Noise suppression tweaks.

Implements 7 voice effects that can be cycled through with F12:
- None (default)
- Echo: Single repeating delay with feedback (250ms)
- Reverb: Multiple short delays without feedback
- High Pitch: Chipmunk voice using cubic interpolation
- Low Pitch: Deep voice effect
- Robot: Ring modulation for robotic sound
- Chorus: Layered voices with pitch variations

The effects are applied after noise suppression and AGC in the audio
pipeline. Selected effect is persisted to config file. Includes
comprehensive documentation in README.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

README.md

@@ -18,13 +18,33 @@ If a user is too soft to hear, you can boost their audio.
 The audio should drastically increase once you have hit the VolumeUp key over 10 times (from the silent/0 position).
 The boost setting is saved per user, just like per user volume.
 
+## Voice Effects
+
+Barnard includes real-time voice effects that can be applied to your outgoing microphone audio. Press F12 to cycle through the available effects.
+
+### Available Effects
+- **None**: No effect applied (default)
+- **Echo**: Single repeating delay effect with feedback (250ms) - creates distinct repetitions that fade away
+- **Reverb**: Multiple short delays (12.5ms, 20ms, 33ms) without feedback - adds thickness and fullness to your voice
+- **High Pitch**: Chipmunk-style voice using pitch shifting
+- **Low Pitch**: Deep voice using pitch shifting
+- **Robot**: Ring modulation effect for robotic sound
+- **Chorus**: Layered voices with slight pitch variations for a rich, ensemble sound
+
+### Controls
+- **F12 key**: Cycle through voice effects (configurable hotkey)
+- **Configuration**: Your selected effect is saved in `~/.barnard.toml`
+
+### How It Works
+Voice effects are applied to your outgoing audio in real-time, after noise suppression and automatic gain control. The effects use various digital signal processing techniques including delay lines, pitch shifting with cubic interpolation, and ring modulation.
+
 ## Noise Suppression
 
 Barnard includes real-time noise suppression for microphone input to filter out background noise such as keyboard typing, computer fans, and other environmental sounds.
 
 ### Features
 - **Real-time processing**: Noise suppression is applied during audio capture with minimal latency
-- **Configurable threshold**: Adjustable noise gate threshold (default: 0.02)
+- **Configurable amount**: Adjustable suppression amount via threshold value (default: `0.08`)
 - **Persistent settings**: Noise suppression preferences are saved in your configuration file
 - **Multiple control methods**: Toggle via hotkey, command line flag, or FIFO commands
 
@@ -32,15 +52,16 @@ Barnard includes real-time noise suppression for microphone input to filter out
 - **F9 key**: Toggle noise suppression on/off (configurable hotkey)
 - **Command line**: Use `--noise-suppression` flag to enable at startup
 - **FIFO command**: Send `noise` command to toggle during runtime
-- **Configuration**: Set `noisesuppressionenabled` and `noisesuppressionthreshold` in `~/.barnard.yaml`
+- **Configuration**: Set `noisesuppressionenabled` and `noisesuppressionthreshold` in `~/.barnard.toml`
 
 ### Configuration Example
-```yaml
-noisesuppressionenabled: true
-noisesuppressionthreshold: 0.02
+```toml
+noisesuppressionenabled = true
+noisesuppressionthreshold = 0.08
 ```
 
-The noise suppression algorithm uses a combination of high-pass filtering and noise gating to reduce unwanted background sounds while preserving voice quality.
+`noisesuppressionthreshold` accepts values from `0.0` to `1.0`, where higher values apply stronger suppression.
+The noise suppression algorithm uses adaptive noise-floor tracking, transient suppression, and smoothed gain reduction to reduce background noise while preserving voice quality.
 
 ## FIFO Control
 
@@ -104,8 +125,8 @@ Our thanks go out to Tim Cooper for the massive amount of work put into this cli
 
 ## Config
 
-By default, the file $HOME/.barnard.yaml will hold the configuration for Barnard.
-You can have barnard read another file by using the -c option, like `./barnard -c ~/.anotherbarnard.yaml`.
+By default, the file $HOME/.barnard.toml will hold the configuration for Barnard.
+You can have barnard read another file by using the -c option, like `./barnard -c ~/.anotherbarnard.toml`.
 It will be created automatically if it doesn't exist.
 If you modify the config file while Barnard is running, your changes may be overwritten.
 
@@ -121,6 +142,18 @@ Pass the -list_devices parameter to barnard to be given a list of audio input an
 Copy lines from the above list into inputdevice and outputdevice as desired.
 To clear your inputdevice or outputdevice options and set them to defaults, set them to "" or delete them entirely.
 
+### Audio Backends (ALSA, PipeWire, PulseAudio)
+
+Barnard uses OpenAL Soft for audio. By default it will pick the first available backend (often ALSA), but you can force a specific driver:
+
+- Command line: `./barnard --audio-driver pipewire` (or `pulse`, `alsa`, `jack`)
+- Config file: add `audiodriver = "pipewire"` to your `~/.barnard.toml`
+- Environment: `ALSOFT_DRIVERS=pipewire ./barnard` (takes precedence over config)
+
+If PipeWire or PulseAudio support is missing, install OpenAL Soft with the corresponding backend enabled (e.g., `libopenal1` or `openal-soft` packages built with PipeWire). After changing drivers, rerun with `--list_devices` to confirm the desired devices appear.
+
+Leaving `audiodriver` empty in the config keeps the OpenAL default ordering (PipeWire/Pulse first if available, then ALSA).
+
 ## Keystrokes
 
 You can see the below keystrokes in your config file.
@@ -149,7 +182,7 @@ If Jim's volume is set to 0.1, and larry's volume is set to 0.9, lowering the ch
 
 You can change the volume for a user once that user has spoken at least once during a session.
 Attempts to change the volume of a user who has not spoken will be ignored.
-If you are unable to hear a user speaking, you can edit the .barnard.yaml file in your home directory, after closing Barnard, and set the volume parameter to 1.0 for a particular user.
+If you are unable to hear a user speaking, you can edit the .barnard.toml file in your home directory, after closing Barnard, and set the volume parameter to 1.0 for a particular user.
 
 ### Technical
 
@@ -191,6 +224,7 @@ After running the command above, `barnard` will be compiled as `$(go env GOPATH)
 
 - <kbd>F1</kbd>: toggle voice transmission
 - <kbd>F9</kbd>: toggle noise suppression
+- <kbd>F12</kbd>: cycle through voice effects
 - <kbd>Ctrl+L</kbd>: clear chat log
 - <kbd>Tab</kbd>: toggle focus between chat and user tree
 - <kbd>Page Up</kbd>: scroll chat up

audio/agc.go

@@ -0,0 +1,159 @@
+package audio
+
+import (
+	"math"
+)
+
+// AGC (Automatic Gain Control) processor for voice normalization
+type AGC struct {
+	targetLevel    float32 // Target RMS level (0.0-1.0)
+	maxGain        float32 // Maximum gain multiplier
+	minGain        float32 // Minimum gain multiplier
+	attackTime     float32 // Attack time coefficient
+	releaseTime    float32 // Release time coefficient
+	currentGain    float32 // Current gain value
+	envelope       float32 // Signal envelope
+	enabled        bool    // Whether AGC is enabled
+	compThreshold  float32 // Compression threshold
+	compRatio      float32 // Compression ratio
+}
+
+// NewAGC creates a new AGC processor with sensible defaults for voice
+func NewAGC() *AGC {
+	return &AGC{
+		targetLevel:    0.18,  // Target 18% of max amplitude (balanced level)
+		maxGain:        8.0,   // Maximum 8x gain (about 18dB)
+		minGain:        0.1,   // Minimum 0.1x gain (-20dB)
+		attackTime:     0.005, // Fast attack (5ms)
+		releaseTime:    0.1,   // Slower release (100ms)
+		currentGain:    1.0,   // Start with unity gain
+		envelope:       0.0,   // Start with zero envelope
+		enabled:        true,  // Enable by default
+		compThreshold:  0.7,   // Compress signals above 70%
+		compRatio:      3.0,   // 3:1 compression ratio
+	}
+}
+
+// ProcessSamples applies AGC processing to audio samples
+func (agc *AGC) ProcessSamples(samples []int16) {
+	if !agc.enabled || len(samples) == 0 {
+		return
+	}
+
+	// Convert samples to float32 for processing
+	floatSamples := make([]float32, len(samples))
+	for i, sample := range samples {
+		floatSamples[i] = float32(sample) / 32768.0
+	}
+
+	// Calculate RMS level for gain control
+	rmsSum := float32(0.0)
+	for _, sample := range floatSamples {
+		rmsSum += sample * sample
+	}
+	rms := float32(math.Sqrt(float64(rmsSum / float32(len(floatSamples)))))
+
+	// Update envelope with peak detection for compression
+	peak := float32(0.0)
+	for _, sample := range floatSamples {
+		absample := float32(math.Abs(float64(sample)))
+		if absample > peak {
+			peak = absample
+		}
+	}
+
+	// Envelope following
+	if peak > agc.envelope {
+		agc.envelope += (peak - agc.envelope) * agc.attackTime
+	} else {
+		agc.envelope += (peak - agc.envelope) * agc.releaseTime
+	}
+
+	// Calculate desired gain based on RMS
+	var desiredGain float32
+	if rms > 0.001 { // Avoid division by zero for very quiet signals
+		desiredGain = agc.targetLevel / rms
+	} else {
+		desiredGain = agc.maxGain // Boost very quiet signals
+	}
+
+	// Apply gain limits
+	if desiredGain > agc.maxGain {
+		desiredGain = agc.maxGain
+	}
+	if desiredGain < agc.minGain {
+		desiredGain = agc.minGain
+	}
+
+	// Smooth gain changes
+	if desiredGain > agc.currentGain {
+		agc.currentGain += (desiredGain - agc.currentGain) * agc.attackTime
+	} else {
+		agc.currentGain += (desiredGain - agc.currentGain) * agc.releaseTime
+	}
+
+	// Apply AGC gain to samples
+	for i, sample := range floatSamples {
+		processed := sample * agc.currentGain
+
+		// Apply compression for loud signals
+		if agc.envelope > agc.compThreshold {
+			// Calculate compression amount
+			overage := agc.envelope - agc.compThreshold
+			compAmount := overage / agc.compRatio
+			compGain := (agc.compThreshold + compAmount) / agc.envelope
+			processed *= compGain
+		}
+
+		// Soft limiting to prevent clipping
+		if processed > 0.90 {
+			processed = 0.90 + (processed-0.90)*0.1
+		} else if processed < -0.90 {
+			processed = -0.90 + (processed+0.90)*0.1
+		}
+
+		// Convert back to int16
+		intSample := int32(processed * 32767.0)
+		if intSample > 32767 {
+			intSample = 32767
+		} else if intSample < -32767 {
+			intSample = -32767
+		}
+		samples[i] = int16(intSample)
+	}
+}
+
+// SetEnabled enables or disables AGC processing
+func (agc *AGC) SetEnabled(enabled bool) {
+	agc.enabled = enabled
+}
+
+// IsEnabled returns whether AGC is enabled
+func (agc *AGC) IsEnabled() bool {
+	return agc.enabled
+}
+
+// SetTargetLevel sets the target RMS level (0.0-1.0)
+func (agc *AGC) SetTargetLevel(level float32) {
+	if level > 0.0 && level < 1.0 {
+		agc.targetLevel = level
+	}
+}
+
+// SetMaxGain sets the maximum gain multiplier
+func (agc *AGC) SetMaxGain(gain float32) {
+	if gain > 1.0 && gain <= 10.0 {
+		agc.maxGain = gain
+	}
+}
+
+// GetCurrentGain returns the current gain being applied
+func (agc *AGC) GetCurrentGain() float32 {
+	return agc.currentGain
+}
+
+// Reset resets the AGC state
+func (agc *AGC) Reset() {
+	agc.currentGain = 1.0
+	agc.envelope = 0.0
+}

audio/effects.go

@@ -0,0 +1,408 @@
+package audio
+
+import (
+	"math"
+)
+
+// VoiceEffect represents different voice effect types
+type VoiceEffect int
+
+const (
+	EffectNone VoiceEffect = iota
+	EffectEcho
+	EffectReverb
+	EffectHighPitch
+	EffectLowPitch
+	EffectRobot
+	EffectChorus
+	EffectCount // Keep this last for cycling
+)
+
+// String returns the name of the effect
+func (e VoiceEffect) String() string {
+	switch e {
+	case EffectNone:
+		return "None"
+	case EffectEcho:
+		return "Echo"
+	case EffectReverb:
+		return "Reverb"
+	case EffectHighPitch:
+		return "High Pitch"
+	case EffectLowPitch:
+		return "Low Pitch"
+	case EffectRobot:
+		return "Robot"
+	case EffectChorus:
+		return "Chorus"
+	default:
+		return "Unknown"
+	}
+}
+
+// EffectsProcessor handles voice effects processing
+type EffectsProcessor struct {
+	currentEffect VoiceEffect
+	enabled       bool
+
+	// Echo parameters
+	echoDelay     int     // Delay in samples
+	echoFeedback  float32 // Echo feedback amount (0-1)
+	echoMix       float32 // Mix of echo with original (0-1)
+	echoBuffer    []int16 // Circular buffer for echo
+	echoPosition  int     // Current position in echo buffer
+
+	// Reverb buffer
+	reverseInputBuffer []int16 // Delay line for reverb
+	reverseInputPos    int     // Write position in buffer
+
+	// Pitch shift parameters
+	pitchRatio    float32 // Pitch shift ratio
+	pitchBuffer   []int16 // Buffer for pitch shifting
+	pitchPhase    float32 // Phase accumulator for resampling
+
+	// Robot voice parameters
+	robotFreq     float32 // Modulation frequency
+	robotPhase    float32 // Phase accumulator
+	sampleRate    float32 // Audio sample rate
+
+	// Chorus parameters
+	chorusDelays  []int   // Multiple delay times
+	chorusBuffers [][]int16 // Multiple delay buffers
+	chorusPositions []int  // Positions in chorus buffers
+	chorusRates   []float32 // LFO rates for each chorus voice
+	chorusPhases  []float32 // LFO phases
+}
+
+// NewEffectsProcessor creates a new voice effects processor
+func NewEffectsProcessor(sampleRate int) *EffectsProcessor {
+	echoDelay := sampleRate / 4 // 250ms delay
+
+	return &EffectsProcessor{
+		currentEffect: EffectNone,
+		enabled:       true,
+		sampleRate:    float32(sampleRate),
+
+		// Echo setup
+		echoDelay:    echoDelay,
+		echoFeedback: 0.4,
+		echoMix:      0.5,
+		echoBuffer:   make([]int16, echoDelay),
+		echoPosition: 0,
+
+		// Reverb setup - 100ms buffer for delay lines
+		reverseInputBuffer: make([]int16, sampleRate/10), // 100ms buffer
+		reverseInputPos:    0,
+
+		// Pitch shift setup
+		pitchRatio:  1.0,
+		pitchBuffer: make([]int16, 4096),
+		pitchPhase:  0.0,
+
+		// Robot voice setup
+		robotFreq:  30.0, // 30 Hz modulation
+		robotPhase: 0.0,
+
+		// Chorus setup (3 voices)
+		chorusDelays:    []int{sampleRate/50, sampleRate/40, sampleRate/35}, // ~20-30ms
+		chorusBuffers:   make([][]int16, 3),
+		chorusPositions: make([]int, 3),
+		chorusRates:     []float32{1.5, 2.0, 2.3}, // LFO rates in Hz
+		chorusPhases:    make([]float32, 3),
+	}
+}
+
+// GetCurrentEffect returns the current effect
+func (ep *EffectsProcessor) GetCurrentEffect() VoiceEffect {
+	return ep.currentEffect
+}
+
+// SetEffect sets the current effect
+func (ep *EffectsProcessor) SetEffect(effect VoiceEffect) {
+	if effect >= 0 && effect < EffectCount {
+		ep.currentEffect = effect
+		ep.resetBuffers()
+	}
+}
+
+// CycleEffect cycles to the next effect
+func (ep *EffectsProcessor) CycleEffect() VoiceEffect {
+	ep.currentEffect = (ep.currentEffect + 1) % EffectCount
+	ep.resetBuffers()
+	return ep.currentEffect
+}
+
+// SetEnabled enables or disables effects processing
+func (ep *EffectsProcessor) SetEnabled(enabled bool) {
+	ep.enabled = enabled
+}
+
+// IsEnabled returns whether effects are enabled
+func (ep *EffectsProcessor) IsEnabled() bool {
+	return ep.enabled
+}
+
+// ProcessSamples applies the current voice effect to audio samples
+func (ep *EffectsProcessor) ProcessSamples(samples []int16) {
+	if !ep.enabled || ep.currentEffect == EffectNone || len(samples) == 0 {
+		return
+	}
+
+	switch ep.currentEffect {
+	case EffectEcho:
+		ep.processEcho(samples)
+	case EffectReverb:
+		ep.processReverb(samples)
+	case EffectHighPitch:
+		ep.processPitchShift(samples, 1.5)
+	case EffectLowPitch:
+		ep.processPitchShift(samples, 0.75)
+	case EffectRobot:
+		ep.processRobot(samples)
+	case EffectChorus:
+		ep.processChorus(samples)
+	}
+}
+
+// processEcho applies echo effect
+func (ep *EffectsProcessor) processEcho(samples []int16) {
+	for i := range samples {
+		// Get delayed sample
+		delayedSample := ep.echoBuffer[ep.echoPosition]
+
+		// Mix original with echo
+		outputSample := float32(samples[i])*(1.0-ep.echoMix) +
+		               float32(delayedSample)*ep.echoMix
+
+		// Create new echo sample (current + feedback)
+		newEchoSample := float32(samples[i]) + float32(delayedSample)*ep.echoFeedback
+
+		// Store in buffer with clipping
+		if newEchoSample > 32767 {
+			newEchoSample = 32767
+		} else if newEchoSample < -32767 {
+			newEchoSample = -32767
+		}
+		ep.echoBuffer[ep.echoPosition] = int16(newEchoSample)
+
+		// Advance buffer position
+		ep.echoPosition = (ep.echoPosition + 1) % len(ep.echoBuffer)
+
+		// Apply to output with clipping
+		if outputSample > 32767 {
+			outputSample = 32767
+		} else if outputSample < -32767 {
+			outputSample = -32767
+		}
+		samples[i] = int16(outputSample)
+	}
+}
+
+// processReverb applies reverb effect - like echo but with multiple short delays
+func (ep *EffectsProcessor) processReverb(samples []int16) {
+	bufLen := len(ep.reverseInputBuffer)
+
+	// Three quick echoes instead of one long repeating echo
+	delays := []int{
+		bufLen / 8,  // ~12.5ms
+		bufLen / 5,  // ~20ms
+		bufLen / 3,  // ~33ms
+	}
+	gains := []float32{0.3, 0.2, 0.15}
+
+	for i := range samples {
+		// Store current sample
+		ep.reverseInputBuffer[ep.reverseInputPos] = samples[i]
+
+		// Add the three quick echoes
+		reverbSample := float32(0)
+		for j := 0; j < len(delays); j++ {
+			readPos := (ep.reverseInputPos - delays[j] + bufLen) % bufLen
+			reverbSample += float32(ep.reverseInputBuffer[readPos]) * gains[j]
+		}
+
+		// Mix dry and wet signal
+		outputSample := float32(samples[i])*0.7 + reverbSample
+
+		// Advance position
+		ep.reverseInputPos = (ep.reverseInputPos + 1) % bufLen
+
+		// Apply with clipping
+		if outputSample > 32767 {
+			outputSample = 32767
+		} else if outputSample < -32767 {
+			outputSample = -32767
+		}
+
+		samples[i] = int16(outputSample)
+	}
+}
+
+// processPitchShift applies pitch shifting using cubic interpolation
+func (ep *EffectsProcessor) processPitchShift(samples []int16, ratio float32) {
+	if ratio == 1.0 {
+		return
+	}
+
+	bufLen := len(ep.pitchBuffer)
+
+	// Copy samples to pitch buffer (maintaining history)
+	copy(ep.pitchBuffer[bufLen-len(samples):], samples)
+
+	// Resample using cubic interpolation for smoother output
+	for i := range samples {
+		// Calculate source position
+		srcPos := float32(bufLen-len(samples)) + float32(i)*ratio
+
+		// Bounds check with extra padding for cubic interpolation
+		if srcPos >= float32(bufLen-2) {
+			srcPos = float32(bufLen - 3)
+		}
+		if srcPos < 1 {
+			srcPos = 1
+		}
+
+		// Cubic interpolation (Hermite interpolation)
+		idx := int(srcPos)
+		frac := srcPos - float32(idx)
+
+		// Get 4 samples around the target position
+		y0 := float32(ep.pitchBuffer[idx-1])
+		y1 := float32(ep.pitchBuffer[idx])
+		y2 := float32(ep.pitchBuffer[idx+1])
+		y3 := float32(ep.pitchBuffer[idx+2])
+
+		// Cubic Hermite interpolation
+		c0 := y1
+		c1 := 0.5 * (y2 - y0)
+		c2 := y0 - 2.5*y1 + 2.0*y2 - 0.5*y3
+		c3 := 0.5*(y3-y0) + 1.5*(y1-y2)
+
+		interpolated := c0 + c1*frac + c2*frac*frac + c3*frac*frac*frac
+
+		// Soft clipping to reduce harshness
+		if interpolated > 32767 {
+			interpolated = 32767
+		} else if interpolated < -32767 {
+			interpolated = -32767
+		}
+
+		samples[i] = int16(interpolated)
+	}
+
+	// Shift buffer for next frame
+	copy(ep.pitchBuffer, ep.pitchBuffer[len(samples):])
+}
+
+// processRobot applies ring modulation for robot voice
+func (ep *EffectsProcessor) processRobot(samples []int16) {
+	phaseIncrement := 2.0 * math.Pi * ep.robotFreq / ep.sampleRate
+
+	for i := range samples {
+		// Generate carrier wave (sine wave)
+		carrier := float32(math.Sin(float64(ep.robotPhase)))
+
+		// Ring modulation: multiply signal by carrier
+		modulated := float32(samples[i]) * (0.5 + carrier*0.5)
+
+		// Advance phase
+		ep.robotPhase += phaseIncrement
+		if ep.robotPhase >= 2.0*math.Pi {
+			ep.robotPhase -= 2.0 * math.Pi
+		}
+
+		// Apply with clipping
+		if modulated > 32767 {
+			modulated = 32767
+		} else if modulated < -32767 {
+			modulated = -32767
+		}
+
+		samples[i] = int16(modulated)
+	}
+}
+
+// processChorus applies chorus effect with multiple delayed voices
+func (ep *EffectsProcessor) processChorus(samples []int16) {
+	// Initialize chorus buffers if needed
+	for j := range ep.chorusBuffers {
+		if len(ep.chorusBuffers[j]) == 0 {
+			ep.chorusBuffers[j] = make([]int16, ep.chorusDelays[j])
+		}
+	}
+
+	for i := range samples {
+		output := float32(samples[i]) * 0.4 // Original signal at 40%
+
+		// Add multiple chorus voices
+		for j := 0; j < len(ep.chorusDelays); j++ {
+			// LFO modulation for slight pitch variation
+			lfoPhaseInc := 2.0 * math.Pi * ep.chorusRates[j] / ep.sampleRate
+			lfo := float32(math.Sin(float64(ep.chorusPhases[j])))
+			ep.chorusPhases[j] += lfoPhaseInc
+			if ep.chorusPhases[j] >= 2.0*math.Pi {
+				ep.chorusPhases[j] -= 2.0 * math.Pi
+			}
+
+			// Get delayed sample with LFO modulation
+			modDelay := int(float32(ep.chorusDelays[j]) * (1.0 + lfo*0.03))
+			if modDelay >= len(ep.chorusBuffers[j]) {
+				modDelay = len(ep.chorusBuffers[j]) - 1
+			}
+
+			readPos := (ep.chorusPositions[j] - modDelay + len(ep.chorusBuffers[j])) % len(ep.chorusBuffers[j])
+			delayedSample := ep.chorusBuffers[j][readPos]
+
+			// Add this voice to output (20% each)
+			output += float32(delayedSample) * 0.2
+
+			// Store current sample in buffer
+			ep.chorusBuffers[j][ep.chorusPositions[j]] = samples[i]
+			ep.chorusPositions[j] = (ep.chorusPositions[j] + 1) % len(ep.chorusBuffers[j])
+		}
+
+		// Apply with clipping
+		if output > 32767 {
+			output = 32767
+		} else if output < -32767 {
+			output = -32767
+		}
+
+		samples[i] = int16(output)
+	}
+}
+
+// resetBuffers clears all effect buffers
+func (ep *EffectsProcessor) resetBuffers() {
+	// Clear echo buffer
+	for i := range ep.echoBuffer {
+		ep.echoBuffer[i] = 0
+	}
+	ep.echoPosition = 0
+
+	// Clear reverb buffer
+	for i := range ep.reverseInputBuffer {
+		ep.reverseInputBuffer[i] = 0
+	}
+	ep.reverseInputPos = 0
+
+	// Clear pitch buffer
+	for i := range ep.pitchBuffer {
+		ep.pitchBuffer[i] = 0
+	}
+	ep.pitchPhase = 0
+
+	// Reset robot phase
+	ep.robotPhase = 0
+
+	// Clear chorus buffers
+	for j := range ep.chorusBuffers {
+		if len(ep.chorusBuffers[j]) > 0 {
+			for i := range ep.chorusBuffers[j] {
+				ep.chorusBuffers[j][i] = 0
+			}
+		}
+		ep.chorusPositions[j] = 0
+		ep.chorusPhases[j] = 0
+	}
+}

barnard-ui

@@ -85,7 +85,7 @@ menulist() {
     # returns: selected tag
     local i
     local menuList
-    for i in $@ ; do
+    for i in "$@" ; do
         menuList+=("$i" "$i")
     done
     dialog --backtitle "$(gettext "Use the up and down arrow keys to find the option you want, then press enter to select it.")" \
@@ -135,10 +135,16 @@ connect() {
     if [[ -z "$serverName" || "$serverName" == "Go Back" ]]; then
         return
     fi
-    local username="$(grep '^username: .*$' ~/.barnard.yaml 2> /dev/null | cut -d ' ' -f2-)"
+    local username
+    username="$(grep -m 1 '^Username = ' ~/.barnard.toml 2> /dev/null | cut -d '=' -f2- | sed "s/^[[:space:]]*//;s/[[:space:]]*$//;s/'//g")"
     username="${username//[[:space:]]/_}"
-    username="${username//\"/}"
-    command barnard -username "${username:-${USER}-${HOSTNAME}}" -server ${mumbleServerList[$serverName]} --fifo ~/.config/barnard/cmd --buffers 16 |& log
+    username="${username:-${USER}-${HOSTNAME}}"
+    local certArgs=()
+    if [[ -f "$certFile" ]]; then
+        certArgs=(-certificate "$certFile")
+    fi
+    # shellcheck disable=SC2086
+    command barnard -username "$username" -server ${mumbleServerList[$serverName]} "${certArgs[@]}" --fifo ~/.config/barnard/cmd --buffers 16 |& log
 }
 
 remove-server() {
@@ -159,9 +165,99 @@ remove-server() {
     msgbox "$(gettext "Removed server") $serverName"
   }  
 
+# Certificate configuration
+certDir="$HOME/.config/barnard"
+certFile="$certDir/barnard.pem"
+
+generate-certificate() {
+    if [[ -f "$certFile" ]]; then
+        if [[ "$(yesno "$(gettext "A certificate already exists. Do you want to replace it? This may affect your registered identity on servers.")")" != "Yes" ]]; then
+            return
+        fi
+    fi
+    local commonName
+    commonName="$(inputbox "$(gettext "Enter a name for your certificate (e.g., your username):")" "barnard")"
+    [[ $? -ne 0 ]] && return
+    [[ -z "$commonName" ]] && commonName="barnard"
+
+    if openssl req -x509 -newkey rsa:2048 -keyout "$certFile" -out "$certFile" -days 3650 -nodes -subj "/CN=$commonName" 2>/dev/null; then
+        chmod 600 "$certFile"
+        msgbox "$(gettext "Certificate generated successfully.")"
+    else
+        msgbox "$(gettext "Failed to generate certificate. Make sure openssl is installed.")"
+    fi
+}
+
+view-certificate() {
+    if [[ ! -f "$certFile" ]]; then
+        msgbox "$(gettext "No certificate found.") $certFile"
+        return
+    fi
+    local certInfo
+    certInfo=$(openssl x509 -in "$certFile" -noout -subject -dates -fingerprint 2>/dev/null)
+    if [[ -n "$certInfo" ]]; then
+        msgbox "$certInfo"
+    else
+        msgbox "$(gettext "Could not read certificate information.")"
+    fi
+}
+
+import-certificate() {
+    local importPath
+    importPath="$(inputbox "$(gettext "Enter the full path to your certificate file (PEM format with certificate and private key):")")"
+    [[ $? -ne 0 ]] && return
+    [[ -z "$importPath" ]] && return
+
+    # Expand ~ if present
+    importPath="${importPath/#\~/$HOME}"
+
+    if [[ ! -f "$importPath" ]]; then
+        msgbox "$(gettext "File not found:") $importPath"
+        return
+    fi
+
+    # Verify it's a valid certificate
+    if ! openssl x509 -in "$importPath" -noout 2>/dev/null; then
+        msgbox "$(gettext "The file does not appear to be a valid PEM certificate.")"
+        return
+    fi
+
+    # Verify it contains a private key
+    if ! openssl rsa -in "$importPath" -check -noout 2>/dev/null && ! openssl ec -in "$importPath" -check -noout 2>/dev/null; then
+        msgbox "$(gettext "The file does not appear to contain a valid private key. The certificate file must contain both the certificate and private key.")"
+        return
+    fi
+
+    if [[ -f "$certFile" ]]; then
+        if [[ "$(yesno "$(gettext "A certificate already exists. Do you want to replace it?")")" != "Yes" ]]; then
+            return
+        fi
+    fi
+
+    if cp "$importPath" "$certFile" && chmod 600 "$certFile"; then
+        msgbox "$(gettext "Certificate imported successfully.")"
+    else
+        msgbox "$(gettext "Failed to import certificate.")"
+    fi
+}
+
+manage-certificate() {
+    while : ; do
+        local certAction
+        certAction="$(menulist "Generate" "View" "Import" "Go_Back")"
+        [[ $? -eq 1 ]] && return
+        case "$certAction" in
+            "Generate") generate-certificate ;;
+            "View") view-certificate ;;
+            "Import") import-certificate ;;
+            "Go_Back"|"") return ;;
+        esac
+    done
+}
+
 # main menu
 while : ; do
-    action="$(menulist "Connect" "Add_server" "Remove_server")"
+    action="$(menulist "Connect" "Add_server" "Remove_server" "Manage_Certificate")"
     [[ $? -eq 1 ]] && exit 0
     action="${action,,}"
     action="${action//_/-}"

barnard.go

@@ -2,8 +2,11 @@ package main
 
 import (
     "crypto/tls"
+    "sync"
 
+    "git.stormux.org/storm/barnard/audio"
     "git.stormux.org/storm/barnard/config"
+    "git.stormux.org/storm/barnard/fileplayback"
     "git.stormux.org/storm/barnard/gumble/gumble"
     "git.stormux.org/storm/barnard/gumble/gumbleopenal"
     "git.stormux.org/storm/barnard/noise"
@@ -44,9 +47,16 @@ type Barnard struct {
 
     // Added for channel muting
     MutedChannels map[uint32]bool
-    
+
     // Added for noise suppression
     NoiseSuppressor *noise.Suppressor
+
+    // Added for voice effects
+    VoiceEffects *audio.EffectsProcessor
+
+    // Added for file playback
+    FileStream      *fileplayback.Player
+    FileStreamMutex sync.Mutex
 }
 
 func (b *Barnard) StopTransmission() {

client.go

@@ -5,9 +5,11 @@ import (
     "net"
     "time"
 
+    "git.stormux.org/storm/barnard/fileplayback"
     "git.stormux.org/storm/barnard/gumble/gumble"
     "git.stormux.org/storm/barnard/gumble/gumbleopenal"
     "git.stormux.org/storm/barnard/gumble/gumbleutil"
+    "git.stormux.org/storm/barnard/gumble/opus"
 )
 
 func (b *Barnard) start() {
@@ -50,6 +52,22 @@ func (b *Barnard) connect(reconnect bool) bool {
     b.Stream = stream
     b.Stream.AttachStream(b.Client)
     b.Stream.SetNoiseProcessor(b.NoiseSuppressor)
+    b.Stream.SetEffectsProcessor(b.VoiceEffects)
+
+    // Initialize stereo encoder for file playback
+    b.Client.AudioEncoderStereo = opus.NewStereoEncoder()
+
+    // Initialize file player
+    b.FileStreamMutex.Lock()
+    b.FileStream = fileplayback.New(b.Client)
+    b.FileStream.SetErrorFunc(func(err error) {
+        // Disable stereo when file finishes or errors
+        b.Client.DisableStereoEncoder()
+        b.AddOutputLine(fmt.Sprintf("File playback: %s", err.Error()))
+    })
+    b.Stream.SetFilePlayer(b.FileStream)
+    b.FileStreamMutex.Unlock()
+
     b.Connected = true
     return true
 }
@@ -161,6 +179,11 @@ func (b *Barnard) OnUserChange(e *gumble.UserChangeEvent) {
     if e.Type.Has(gumble.UserChangeConnected) {
         s = "joined"
         t = "join"
+        // Notify about users joining our channel
+        if e.User.Channel.Name == b.Client.Self.Channel.Name {
+            b.Notify(t, e.User.Name, e.User.Channel.Name)
+            b.AddOutputLine(fmt.Sprintf("%s %s %s", e.User.Name, s, e.User.Channel.Name))
+        }
     }
     if e.Type.Has(gumble.UserChangeDisconnected) {
         s = "left"
@@ -168,10 +191,11 @@ func (b *Barnard) OnUserChange(e *gumble.UserChangeEvent) {
         if e.User == b.selectedUser {
             b.SetSelectedUser(nil)
         }
-    }
-    if e.User.Channel.Name == b.Client.Self.Channel.Name {
-        b.Notify(t, e.User.Name, e.User.Channel.Name)
-        b.AddOutputLine(fmt.Sprintf("%s %s %s", e.User.Name, s, e.User.Channel.Name))
+        // Always notify about disconnects if user has channel info and was in our channel
+        if e.User.Channel != nil && e.User.Channel.Name == b.Client.Self.Channel.Name {
+            b.Notify(t, e.User.Name, e.User.Channel.Name)
+            b.AddOutputLine(fmt.Sprintf("%s %s %s", e.User.Name, s, e.User.Channel.Name))
+        }
     }
     if e.Type.Has(gumble.UserChangeChannel) && e.User == b.Client.Self {
         b.UpdateInputStatus(fmt.Sprintf("[%s]", e.User.Channel.Name))

config/hotkey_config.go

@@ -19,4 +19,5 @@ type Hotkeys struct {
     ScrollToTop      *uiterm.Key
     ScrollToBottom   *uiterm.Key
     NoiseSuppressionToggle *uiterm.Key
+    CycleVoiceEffect *uiterm.Key
 }

config/user_config.go

@@ -1,327 +1,379 @@
 package config
 
 import (
-    "fmt"
-    "git.stormux.org/storm/barnard/uiterm"
-    "gopkg.in/yaml.v2"
-    "git.stormux.org/storm/barnard/gumble/gumble"
-    "io/ioutil"
-    "os"
-    "os/user"
-    "strconv"
-    "strings"
+	"fmt"
+	"git.stormux.org/storm/barnard/gumble/gumble"
+	"git.stormux.org/storm/barnard/uiterm"
+	"github.com/pelletier/go-toml/v2"
+	"io/ioutil"
+	"os"
+	"os/user"
+	"strconv"
+	"strings"
 )
 
 type Config struct {
-    config *exportableConfig
-    fn     string
+	config *exportableConfig
+	fn     string
 }
 
 type exportableConfig struct {
-    Hotkeys       *Hotkeys
-    MicVolume     *float32
-    InputDevice   *string
-    OutputDevice  *string
-    Servers       []*server
-    DefaultServer *string
-    Username      *string
-    NotifyCommand *string
-    NoiseSuppressionEnabled    *bool
-    NoiseSuppressionThreshold *float32
+	Hotkeys                   *Hotkeys
+	AudioDriver               *string
+	MicVolume                 *float32
+	InputDevice               *string
+	OutputDevice              *string
+	Servers                   []*server
+	DefaultServer             *string
+	Username                  *string
+	NotifyCommand             *string
+	NoiseSuppressionEnabled   *bool
+	NoiseSuppressionThreshold *float32
+	VoiceEffect               *int
+	Certificate               *string
 }
 
 type server struct {
-    Host  string
-    Port  int
-    Users []*eUser
+	Host  string
+	Port  int
+	Users []*eUser
 }
 
 type eUser struct {
-    Username    string
-    Boost      uint16
-    Volume     float32
-    LocallyMuted bool    // Changed from Muted to LocallyMuted to match User struct
+	Username     string
+	Boost        uint16
+	Volume       float32
+	LocallyMuted bool // Changed from Muted to LocallyMuted to match User struct
 }
 
 func (c *Config) SaveConfig() {
-    var data []byte
-    data, err := yaml.Marshal(c.config)
-    if err != nil {
-        panic(err)
-    }
-    err = ioutil.WriteFile(c.fn+".tmp", data, 0600)
-    if err != nil {
-        panic(err)
-    }
-    err = os.Rename(c.fn+".tmp", c.fn)
-    if err != nil {
-        panic(err)
-    }
+	var data []byte
+	data, err := toml.Marshal(c.config)
+	if err != nil {
+		panic(err)
+	}
+	err = ioutil.WriteFile(c.fn+".tmp", data, 0600)
+	if err != nil {
+		panic(err)
+	}
+	err = os.Rename(c.fn+".tmp", c.fn)
+	if err != nil {
+		panic(err)
+	}
 }
 
 func key(k uiterm.Key) *uiterm.Key {
-    return &k
+	return &k
 }
 
 func (c *Config) LoadConfig() {
-    var jc exportableConfig
-    jc = exportableConfig{}
-    jc.Hotkeys = &Hotkeys{
-        Talk:             key(uiterm.KeyF1),
-        VolumeDown:       key(uiterm.KeyF5),
-        VolumeUp:         key(uiterm.KeyF6),
-        VolumeReset:      key(uiterm.KeyF8),
-        MuteToggle:       key(uiterm.KeyF7),    // Added mute toggle hotkey
-        Exit:             key(uiterm.KeyF10),
-        ToggleTimestamps: key(uiterm.KeyF3),
-        SwitchViews:      key(uiterm.KeyTab),
-        ScrollUp:         key(uiterm.KeyPgup),
-        ScrollDown:       key(uiterm.KeyPgdn),
-        NoiseSuppressionToggle: key(uiterm.KeyF9),
-    }
-    if fileExists(c.fn) {
-        var data []byte
-        data = readFile(c.fn)
-        if data != nil {
-            err := yaml.UnmarshalStrict(data, &jc)
-            if err != nil {
-                fmt.Fprintf(os.Stderr, "Error parsing \"%s\".\n%s\n", c.fn, err.Error())
-                os.Exit(1)
-            }
-        }
-    }
-    c.config = &jc
-    if c.config.MicVolume == nil {
-        micvol := float32(1.0)
-        jc.MicVolume = &micvol
-    }
-    if c.config.InputDevice == nil {
-        idev := string("")
-        jc.InputDevice = &idev
-    }
-    if c.config.OutputDevice == nil {
-        odev := string("")
-        jc.OutputDevice = &odev
-    }
-    if c.config.DefaultServer == nil {
-        defaultServer := string("localhost:64738")
-        jc.DefaultServer = &defaultServer
-    }
-    if c.config.Username == nil {
-        username := string("")
-        jc.Username = &username
-    }
-    if c.config.NotifyCommand == nil {
-        ncmd := string("")
-        jc.NotifyCommand = &ncmd
-    }
-    if c.config.NoiseSuppressionEnabled == nil {
-        enabled := false
-        jc.NoiseSuppressionEnabled = &enabled
-    }
-    if c.config.NoiseSuppressionThreshold == nil {
-        threshold := float32(0.02)
-        jc.NoiseSuppressionThreshold = &threshold
-    }
+	var jc exportableConfig
+	jc = exportableConfig{}
+	jc.Hotkeys = &Hotkeys{
+		Talk:                   key(uiterm.KeyF1),
+		VolumeDown:             key(uiterm.KeyF5),
+		VolumeUp:               key(uiterm.KeyF6),
+		VolumeReset:            key(uiterm.KeyF8),
+		MuteToggle:             key(uiterm.KeyF7), // Added mute toggle hotkey
+		Exit:                   key(uiterm.KeyF10),
+		ToggleTimestamps:       key(uiterm.KeyF3),
+		SwitchViews:            key(uiterm.KeyTab),
+		ScrollUp:               key(uiterm.KeyPgup),
+		ScrollDown:             key(uiterm.KeyPgdn),
+		NoiseSuppressionToggle: key(uiterm.KeyF9),
+		CycleVoiceEffect:       key(uiterm.KeyF12),
+	}
+	if fileExists(c.fn) {
+		var data []byte
+		data = readFile(c.fn)
+		if data != nil {
+			err := toml.Unmarshal(data, &jc)
+			if err != nil {
+				fmt.Fprintf(os.Stderr, "Error parsing \"%s\".\n%s\n", c.fn, err.Error())
+				os.Exit(1)
+			}
+		}
+	}
+	c.config = &jc
+	if c.config.MicVolume == nil {
+		micvol := float32(1.0)
+		jc.MicVolume = &micvol
+	}
+	if c.config.AudioDriver == nil {
+		driver := string("")
+		jc.AudioDriver = &driver
+	}
+	if c.config.InputDevice == nil {
+		idev := string("")
+		jc.InputDevice = &idev
+	}
+	if c.config.OutputDevice == nil {
+		odev := string("")
+		jc.OutputDevice = &odev
+	}
+	if c.config.DefaultServer == nil {
+		defaultServer := string("localhost:64738")
+		jc.DefaultServer = &defaultServer
+	}
+	if c.config.Username == nil {
+		username := string("")
+		jc.Username = &username
+	}
+	if c.config.NotifyCommand == nil {
+		ncmd := string("/usr/share/barnard/barnard-sound.sh \"%event\" \"%who\" \"%what\"")
+		jc.NotifyCommand = &ncmd
+	}
+	if c.config.NoiseSuppressionEnabled == nil {
+		enabled := false
+		jc.NoiseSuppressionEnabled = &enabled
+	}
+	if c.config.NoiseSuppressionThreshold == nil {
+		threshold := float32(0.08)
+		jc.NoiseSuppressionThreshold = &threshold
+	}
+	if c.config.VoiceEffect == nil {
+		effect := 0 // Default to EffectNone
+		jc.VoiceEffect = &effect
+	}
+	if c.config.Certificate == nil {
+		cert := string("")
+		jc.Certificate = &cert
+	}
 }
 
 func (c *Config) findServer(address string) *server {
-    if c.config.Servers == nil {
-        c.config.Servers = make([]*server, 0)
-    }
-    host, port := makeHostPort(address)
-    var t *server
-    for _, s := range c.config.Servers {
-        if s.Port == port && s.Host == host {
-            t = s
-            break
-        }
-    }
-    if t == nil {
-        t = &server{
-            Host: host,
-            Port: port,
-        }
-        c.config.Servers = append(c.config.Servers, t)
-    }
-    return t
+	if c.config.Servers == nil {
+		c.config.Servers = make([]*server, 0)
+	}
+	host, port := makeHostPort(address)
+	var t *server
+	for _, s := range c.config.Servers {
+		if s.Port == port && s.Host == host {
+			t = s
+			break
+		}
+	}
+	if t == nil {
+		t = &server{
+			Host: host,
+			Port: port,
+		}
+		c.config.Servers = append(c.config.Servers, t)
+	}
+	return t
 }
 
 func (c *Config) findUser(address string, username string) *eUser {
-    var s *server
-    s = c.findServer(address)
-    if s.Users == nil {
-        s.Users = make([]*eUser, 0)
-    }
-    var t *eUser
-    for _, u := range s.Users {
-        if u.Username == username {
-            t = u
-            break
-        }
-    }
-    if t == nil {
-        t = &eUser{
-            Username:    username,
-            Boost:      uint16(1),
-            Volume:     1.0,
-            LocallyMuted: false,  // Initialize local mute state
-        }
-        s.Users = append(s.Users, t)
-    }
-    return t
+	var s *server
+	s = c.findServer(address)
+	if s.Users == nil {
+		s.Users = make([]*eUser, 0)
+	}
+	var t *eUser
+	for _, u := range s.Users {
+		if u.Username == username {
+			t = u
+			break
+		}
+	}
+	if t == nil {
+		t = &eUser{
+			Username:     username,
+			Boost:        uint16(1),
+			Volume:       1.0,
+			LocallyMuted: false, // Initialize local mute state
+		}
+		s.Users = append(s.Users, t)
+	}
+	return t
 }
 
 func (c *Config) ToggleMute(u *gumble.User) {
-    j := c.findUser(u.GetClient().Config.Address, u.Name)
-    j.LocallyMuted = !j.LocallyMuted
-    u.LocallyMuted = j.LocallyMuted
-    c.SaveConfig()
+	j := c.findUser(u.GetClient().Config.Address, u.Name)
+	j.LocallyMuted = !j.LocallyMuted
+	u.LocallyMuted = j.LocallyMuted
+	c.SaveConfig()
 }
 
 func (c *Config) SetMicVolume(v float32) {
-    t := float32(v)
-    c.config.MicVolume = &t
+	t := float32(v)
+	c.config.MicVolume = &t
 }
 
 func (c *Config) GetHotkeys() *Hotkeys {
-    return c.config.Hotkeys
+	return c.config.Hotkeys
 }
 
 func (c *Config) GetNotifyCommand() *string {
-    return c.config.NotifyCommand
+	return c.config.NotifyCommand
+}
+
+func (c *Config) GetAudioDriver() string {
+	if c.config.AudioDriver == nil {
+		return ""
+	}
+	return *c.config.AudioDriver
 }
 
 func (c *Config) GetInputDevice() *string {
-    return c.config.InputDevice
+	return c.config.InputDevice
 }
 
 func (c *Config) GetOutputDevice() *string {
-    return c.config.OutputDevice
+	return c.config.OutputDevice
 }
 
 func (c *Config) GetDefaultServer() *string {
-    return c.config.DefaultServer
+	return c.config.DefaultServer
 }
 
 func (c *Config) GetUsername() *string {
-    return c.config.Username
+	return c.config.Username
+}
+
+func (c *Config) GetCertificate() *string {
+	return c.config.Certificate
 }
 
 func (c *Config) GetNoiseSuppressionEnabled() bool {
-    if c.config.NoiseSuppressionEnabled == nil {
-        return false
-    }
-    return *c.config.NoiseSuppressionEnabled
+	if c.config.NoiseSuppressionEnabled == nil {
+		return false
+	}
+	return *c.config.NoiseSuppressionEnabled
 }
 
 func (c *Config) SetNoiseSuppressionEnabled(enabled bool) {
-    c.config.NoiseSuppressionEnabled = &enabled
-    c.SaveConfig()
+	c.config.NoiseSuppressionEnabled = &enabled
+	c.SaveConfig()
 }
 
 func (c *Config) GetNoiseSuppressionThreshold() float32 {
-    if c.config.NoiseSuppressionThreshold == nil {
-        return 0.02
-    }
-    return *c.config.NoiseSuppressionThreshold
+	if c.config.NoiseSuppressionThreshold == nil {
+		return 0.08
+	}
+	threshold := *c.config.NoiseSuppressionThreshold
+	if threshold < 0.0 {
+		return 0.0
+	}
+	if threshold > 1.0 {
+		return 1.0
+	}
+	return threshold
 }
 
 func (c *Config) SetNoiseSuppressionThreshold(threshold float32) {
-    c.config.NoiseSuppressionThreshold = &threshold
-    c.SaveConfig()
+	if threshold < 0.0 {
+		threshold = 0.0
+	}
+	if threshold > 1.0 {
+		threshold = 1.0
+	}
+	c.config.NoiseSuppressionThreshold = &threshold
+	c.SaveConfig()
+}
+
+func (c *Config) GetVoiceEffect() int {
+	if c.config.VoiceEffect == nil {
+		return 0
+	}
+	return *c.config.VoiceEffect
+}
+
+func (c *Config) SetVoiceEffect(effect int) {
+	c.config.VoiceEffect = &effect
+	c.SaveConfig()
 }
 
 func (c *Config) UpdateUser(u *gumble.User) {
-    var j *eUser
-    var uc *gumble.Client
-    uc = u.GetClient()
-    if uc != nil {
-        j = c.findUser(uc.Config.Address, u.Name)
-        u.Boost = j.Boost
-        u.Volume = j.Volume
-        u.LocallyMuted = j.LocallyMuted  // Update LocallyMuted state from config
-        if u.Boost < 1 {
-            u.Boost = 1
-        }
-    }
+	var j *eUser
+	var uc *gumble.Client
+	uc = u.GetClient()
+	if uc != nil {
+		j = c.findUser(uc.Config.Address, u.Name)
+		u.Boost = j.Boost
+		u.Volume = j.Volume
+		u.LocallyMuted = j.LocallyMuted // Update LocallyMuted state from config
+		if u.Boost < 1 {
+			u.Boost = 1
+		}
+	}
 }
 
 func (c *Config) UpdateConfig(u *gumble.User) {
-    var j *eUser
-    j = c.findUser(u.GetClient().Config.Address, u.Name)
-    j.Boost = u.Boost
-    j.Volume = u.Volume
-    j.LocallyMuted = u.LocallyMuted  // Save LocallyMuted state to config
+	var j *eUser
+	j = c.findUser(u.GetClient().Config.Address, u.Name)
+	j.Boost = u.Boost
+	j.Volume = u.Volume
+	j.LocallyMuted = u.LocallyMuted // Save LocallyMuted state to config
 }
 
 func NewConfig(fn *string) *Config {
-    var c *Config
-    c = &Config{}
-    c.fn = resolvePath(*fn)
-    c.LoadConfig()
-    return c
+	var c *Config
+	c = &Config{}
+	c.fn = resolvePath(*fn)
+	c.LoadConfig()
+	return c
 }
 
 func readFile(path string) []byte {
-    if !fileExists(path) {
-        return nil
-    }
-    dat, err := ioutil.ReadFile(path)
-    if err != nil {
-        return nil
-    }
-    return dat
+	if !fileExists(path) {
+		return nil
+	}
+	dat, err := ioutil.ReadFile(path)
+	if err != nil {
+		return nil
+	}
+	return dat
 }
 
 func fileExists(path string) bool {
-    info, err := os.Stat(path)
-    if os.IsNotExist(err) {
-        return false
-    }
-    return !info.IsDir()
+	info, err := os.Stat(path)
+	if os.IsNotExist(err) {
+		return false
+	}
+	return !info.IsDir()
 }
 
 func resolvePath(path string) string {
-    if strings.HasPrefix(path, "~/") || strings.Contains(path, "$HOME") {
-        usr, err := user.Current()
-        if err != nil {
-            panic(err)
-        }
-        var hd = usr.HomeDir
-        if strings.Contains(path, "$HOME") {
-            path = strings.Replace(path, "$HOME", hd, 1)
-        } else {
-            path = strings.Replace(path, "~", hd, 1)
-        }
-    }
-    return path
+	if strings.HasPrefix(path, "~/") || strings.Contains(path, "$HOME") {
+		usr, err := user.Current()
+		if err != nil {
+			panic(err)
+		}
+		var hd = usr.HomeDir
+		if strings.Contains(path, "$HOME") {
+			path = strings.Replace(path, "$HOME", hd, 1)
+		} else {
+			path = strings.Replace(path, "~", hd, 1)
+		}
+	}
+	return path
 }
 
 func makeHostPort(addr string) (string, int) {
-    parts := strings.Split(addr, ":")
-    host := parts[0]
-    port, err := strconv.Atoi(parts[1])
-    if err != nil {
-        panic(err)
-    }
-    return host, port
+	parts := strings.Split(addr, ":")
+	host := parts[0]
+	port, err := strconv.Atoi(parts[1])
+	if err != nil {
+		panic(err)
+	}
+	return host, port
 }
 
 func Log(s string) {
-    log(s)
+	log(s)
 }
 
 func log(s string) {
-    s += "\n"
-    f, err := os.OpenFile("log.log", os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0644)
-    if err != nil {
-        panic(err)
-    }
-    if _, err := f.Write([]byte(s)); err != nil {
-        panic(err)
-    }
-    if err := f.Close(); err != nil {
-        panic(err)
-    }
+	s += "\n"
+	f, err := os.OpenFile("log.log", os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0644)
+	if err != nil {
+		panic(err)
+	}
+	if _, err := f.Write([]byte(s)); err != nil {
+		panic(err)
+	}
+	if err := f.Close(); err != nil {
+		panic(err)
+	}
 }

fileplayback/player.go

@@ -0,0 +1,241 @@
+package fileplayback
+
+import (
+	"encoding/binary"
+	"errors"
+	"io"
+	"os/exec"
+	"strconv"
+	"sync"
+	"time"
+
+	"git.stormux.org/storm/barnard/gumble/gumble"
+	"git.stormux.org/storm/barnard/gumble/go-openal/openal"
+)
+
+// Player handles file playback and mixing with microphone audio
+type Player struct {
+	client      *gumble.Client
+	filename    string
+	audioChan   chan gumble.AudioBuffer
+	stopChan    chan struct{}
+	mutex       sync.Mutex
+	playing     bool
+	errorFunc   func(error)
+
+	// Local playback
+	localSource *openal.Source
+	localBuffers openal.Buffers
+}
+
+// New creates a new file player
+func New(client *gumble.Client) *Player {
+	return &Player{
+		client:    client,
+		audioChan: make(chan gumble.AudioBuffer, 100),
+		stopChan:  make(chan struct{}),
+	}
+}
+
+// SetErrorFunc sets the error callback function
+func (p *Player) SetErrorFunc(f func(error)) {
+	p.mutex.Lock()
+	defer p.mutex.Unlock()
+	p.errorFunc = f
+}
+
+func (p *Player) reportError(err error) {
+	p.mutex.Lock()
+	errorFunc := p.errorFunc
+	p.mutex.Unlock()
+
+	if errorFunc != nil {
+		errorFunc(err)
+	}
+}
+
+// PlayFile starts playing a file
+func (p *Player) PlayFile(filename string) error {
+	p.mutex.Lock()
+	defer p.mutex.Unlock()
+
+	if p.playing {
+		return errors.New("file already playing")
+	}
+
+	p.filename = filename
+
+	// Initialize local playback
+	source := openal.NewSource()
+	p.localSource = &source
+	p.localSource.SetGain(1.0)
+
+	// Create buffers for local playback
+	p.localBuffers = openal.NewBuffers(64)
+
+	// Start the file reading goroutine
+	p.playing = true
+	p.stopChan = make(chan struct{})
+	go p.readFileAudio()
+
+	return nil
+}
+
+// Stop stops the currently playing file
+func (p *Player) Stop() error {
+	p.mutex.Lock()
+	defer p.mutex.Unlock()
+
+	if !p.playing {
+		return errors.New("no file playing")
+	}
+
+	close(p.stopChan)
+	p.playing = false
+
+	// Clean up local playback
+	if p.localSource != nil {
+		p.localSource.Stop()
+		p.localSource.Delete()
+		p.localSource = nil
+	}
+	if p.localBuffers != nil {
+		p.localBuffers.Delete()
+		p.localBuffers = nil
+	}
+
+	// Drain the audio channel
+	for len(p.audioChan) > 0 {
+		<-p.audioChan
+	}
+
+	return nil
+}
+
+// IsPlaying returns true if a file is currently playing
+func (p *Player) IsPlaying() bool {
+	p.mutex.Lock()
+	defer p.mutex.Unlock()
+	return p.playing
+}
+
+// GetAudioFrame returns the next audio frame from the file, or nil if no file is playing
+func (p *Player) GetAudioFrame() []int16 {
+	select {
+	case frame := <-p.audioChan:
+		return []int16(frame)
+	default:
+		return nil
+	}
+}
+
+// playLocalAudio plays audio through the local OpenAL source
+func (p *Player) playLocalAudio(data []byte) {
+	if p.localSource == nil {
+		return
+	}
+
+	// Reclaim processed buffers
+	if n := p.localSource.BuffersProcessed(); n > 0 {
+		reclaimedBufs := make(openal.Buffers, n)
+		p.localSource.UnqueueBuffers(reclaimedBufs)
+		p.localBuffers = append(p.localBuffers, reclaimedBufs...)
+	}
+
+	// If we have available buffers, queue more audio
+	if len(p.localBuffers) > 0 {
+		buffer := p.localBuffers[len(p.localBuffers)-1]
+		p.localBuffers = p.localBuffers[:len(p.localBuffers)-1]
+
+		// Set buffer data as stereo
+		buffer.SetData(openal.FormatStereo16, data, gumble.AudioSampleRate)
+		p.localSource.QueueBuffer(buffer)
+
+		// Start playing if not already
+		if p.localSource.State() != openal.Playing {
+			p.localSource.Play()
+		}
+	}
+}
+
+// readFileAudio reads audio from the file via ffmpeg
+func (p *Player) readFileAudio() {
+	interval := p.client.Config.AudioInterval
+	frameSize := p.client.Config.AudioFrameSize()
+
+	// Use stereo output from ffmpeg to preserve stereo files
+	// Add -loglevel error to suppress info messages
+	args := []string{"-loglevel", "error", "-i", p.filename}
+	args = append(args, "-ac", "2", "-ar", strconv.Itoa(gumble.AudioSampleRate), "-f", "s16le", "-")
+
+	cmd := exec.Command("ffmpeg", args...)
+	pipe, err := cmd.StdoutPipe()
+	if err != nil {
+		p.mutex.Lock()
+		p.playing = false
+		p.mutex.Unlock()
+		p.reportError(errors.New("failed to create ffmpeg pipe: " + err.Error()))
+		return
+	}
+
+	if err := cmd.Start(); err != nil {
+		p.mutex.Lock()
+		p.playing = false
+		p.mutex.Unlock()
+		p.reportError(errors.New("failed to start ffmpeg: " + err.Error()))
+		return
+	}
+
+	// Stereo has 2 channels, so we need twice the buffer size
+	byteBuffer := make([]byte, frameSize*2*2) // frameSize * 2 channels * 2 bytes per sample
+
+	ticker := time.NewTicker(interval)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-p.stopChan:
+			cmd.Process.Kill()
+			cmd.Wait()
+			return
+		case <-ticker.C:
+			n, err := io.ReadFull(pipe, byteBuffer)
+			if err != nil || n != len(byteBuffer) {
+				// File finished playing
+				p.mutex.Lock()
+				p.playing = false
+				// Clean up local playback
+				if p.localSource != nil {
+					p.localSource.Stop()
+					p.localSource.Delete()
+					p.localSource = nil
+				}
+				if p.localBuffers != nil {
+					p.localBuffers.Delete()
+					p.localBuffers = nil
+				}
+				p.mutex.Unlock()
+				cmd.Wait()
+				// Notify that file finished
+				p.reportError(errors.New("file playback finished"))
+				return
+			}
+
+			// Convert stereo bytes to int16 buffer
+			int16Buffer := make([]int16, frameSize*2) // stereo
+			for i := 0; i < len(int16Buffer); i++ {
+				int16Buffer[i] = int16(binary.LittleEndian.Uint16(byteBuffer[i*2 : (i+1)*2]))
+			}
+
+			// Play locally through OpenAL
+			p.playLocalAudio(byteBuffer[:n])
+
+			// Send to channel (non-blocking)
+			select {
+			case p.audioChan <- gumble.AudioBuffer(int16Buffer):
+			default:
+				// Channel full, skip this frame
+			}
+		}
+	}
+}

go.mod

@@ -7,7 +7,7 @@ require (
 	github.com/golang/protobuf v1.5.3
 	github.com/kennygrant/sanitize v1.2.4
 	github.com/nsf/termbox-go v1.1.1
-	gopkg.in/yaml.v2 v2.4.0
+	github.com/pelletier/go-toml/v2 v2.2.4
 )
 
 require (

go.sum

@@ -11,6 +11,8 @@ github.com/mattn/go-runewidth v0.0.9 h1:Lm995f3rfxdpd6TSmuVCHVb/QhupuXlYr8sCI/Qd
 github.com/mattn/go-runewidth v0.0.9/go.mod h1:H031xJmbD/WCDINGzjvQ9THkh0rPKHF+m2gUSrubnMI=
 github.com/nsf/termbox-go v1.1.1 h1:nksUPLCb73Q++DwbYUBEglYBRPZyoXJdrj5L+TkjyZY=
 github.com/nsf/termbox-go v1.1.1/go.mod h1:T0cTdVuOwf7pHQNtfhnEbzHbcNyCEcVU4YPpouCbVxo=
+github.com/pelletier/go-toml/v2 v2.2.4 h1:mye9XuhQ6gvn5h28+VilKrrPoQVanw5PMw/TB0t5Ec4=
+github.com/pelletier/go-toml/v2 v2.2.4/go.mod h1:2gIqNv+qfxSVS7cM2xJQKtLSTLUE9V8t9Stt+h56mCY=
 golang.org/x/net v0.20.0 h1:aCL9BSgETF1k+blQaYUBx9hJ9LOGP3gAVemcZlf1Kpo=
 golang.org/x/net v0.20.0/go.mod h1:z8BVo6PvndSri0LbOE3hAn0apkU+1YvI6E70E9jsnvY=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543 h1:E7g+9GITq07hpfrRu66IVDexMakfv52eLZ2CXBWiKr4=
@@ -18,7 +20,3 @@ golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8T
 google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw=
 google.golang.org/protobuf v1.26.0 h1:bxAC2xTBsZGibn2RTntX0oH50xLsqy1OxA9tTL3p/lk=
 google.golang.org/protobuf v1.26.0/go.mod h1:9q0QmTI4eRPtz6boOQmLYwt+qCgq0jsYwAQnmE0givc=
-gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
-gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
-gopkg.in/yaml.v2 v2.4.0 h1:D8xgwECY7CYvx+Y2n4sBz93Jn9JRvxdiyyo8CTfuKaY=
-gopkg.in/yaml.v2 v2.4.0/go.mod h1:RDklbk79AGWmwhnvt/jBztapEOGDOx6ZbXqjP6csGnQ=

gumble/gumble/audio.go

@@ -1,35 +1,35 @@
 package gumble
 
 import (
-    "time"
+	"time"
 )
 
 const (
-    // AudioSampleRate is the audio sample rate (in hertz) for incoming and
-    // outgoing audio.
-    AudioSampleRate = 48000
+	// AudioSampleRate is the audio sample rate (in hertz) for incoming and
+	// outgoing audio.
+	AudioSampleRate = 48000
 
-    // AudioDefaultInterval is the default interval that audio packets are sent
-    // at.
-    AudioDefaultInterval = 10 * time.Millisecond
+	// AudioDefaultInterval is the default interval that audio packets are sent
+	// at.
+	AudioDefaultInterval = 10 * time.Millisecond
 
-    // AudioMonoChannels is the number of channels used for voice transmission
-    AudioMonoChannels = 1
+	// AudioMonoChannels is the number of channels used for voice transmission
+	AudioMonoChannels = 1
 
-    // AudioChannels is the number of channels used for playback
-    AudioChannels = 2
+	// AudioChannels is the number of channels used for playback
+	AudioChannels = 2
 
-    // AudioDefaultFrameSize is the number of audio frames that should be sent in
-    // a 10ms window (mono samples)
-    AudioDefaultFrameSize = AudioSampleRate / 100
+	// AudioDefaultFrameSize is the number of audio frames that should be sent in
+	// a 10ms window (mono samples)
+	AudioDefaultFrameSize = AudioSampleRate / 100
 
-    // AudioMaximumFrameSize is the maximum audio frame size from another user
-    // that will be processed (accounting for stereo)
-    AudioMaximumFrameSize = (AudioSampleRate / 1000 * 60) * AudioChannels
+	// AudioMaximumFrameSize is the maximum audio frame size from another user
+	// that will be processed (accounting for stereo)
+	AudioMaximumFrameSize = (AudioSampleRate / 1000 * 60) * AudioChannels
 
-    // AudioDefaultDataBytes is the default number of bytes that an audio frame
-    // can use.
-    AudioDefaultDataBytes = 40
+	// AudioDefaultDataBytes is the default number of bytes that an audio frame
+	// can use.
+	AudioDefaultDataBytes = 40
 )
 
 // AudioListener is the interface that must be implemented by types wishing to
@@ -39,48 +39,55 @@ const (
 // implementer's responsibility to continuously process AudioStreamEvent.C
 // until it is closed.
 type AudioListener interface {
-    OnAudioStream(e *AudioStreamEvent)
+	OnAudioStream(e *AudioStreamEvent)
 }
 
 // AudioStreamEvent is event that is passed to AudioListener.OnAudioStream.
 type AudioStreamEvent struct {
-    Client *Client
-    User   *User
-    C      <-chan *AudioPacket
+	Client *Client
+	User   *User
+	C      <-chan *AudioPacket
 }
 
 // AudioBuffer is a slice of PCM audio samples.
 type AudioBuffer []int16
 
 func (a AudioBuffer) writeAudio(client *Client, seq int64, final bool) error {
-    encoder := client.AudioEncoder
-    if encoder == nil {
-        return nil
-    }
-    dataBytes := client.Config.AudioDataBytes
-    raw, err := encoder.Encode(a, len(a), dataBytes)
-    if final {
-        defer encoder.Reset()
-    }
-    if err != nil {
-        return err
-    }
+	// Choose encoder based on whether buffer size indicates stereo or mono
+	encoder := client.AudioEncoder
+	frameSize := client.Config.AudioFrameSize()
+	if len(a) == frameSize*AudioChannels && client.AudioEncoderStereo != nil {
+		encoder = client.AudioEncoderStereo
+	} else if client.IsStereoEncoderEnabled() && client.AudioEncoderStereo != nil {
+		encoder = client.AudioEncoderStereo
+	}
+	if encoder == nil {
+		return nil
+	}
+	dataBytes := client.Config.AudioDataBytes
+	raw, err := encoder.Encode(a, len(a), dataBytes)
+	if final {
+		defer encoder.Reset()
+	}
+	if err != nil {
+		return err
+	}
 
-    var targetID byte
-    if target := client.VoiceTarget; target != nil {
-        targetID = byte(target.ID)
-    }
-    return client.Conn.WriteAudio(byte(4), targetID, seq, final, raw, nil, nil, nil)
+	var targetID byte
+	if target := client.VoiceTarget; target != nil {
+		targetID = byte(target.ID)
+	}
+	return client.Conn.WriteAudio(byte(4), targetID, seq, final, raw, nil, nil, nil)
 }
 
 // AudioPacket contains incoming audio samples and information.
 type AudioPacket struct {
-    Client *Client
-    Sender *User
-    Target *VoiceTarget
+	Client *Client
+	Sender *User
+	Target *VoiceTarget
 
-    AudioBuffer
+	AudioBuffer
 
-    HasPosition bool
-    X, Y, Z     float32
+	HasPosition bool
+	X, Y, Z     float32
 }

gumble/gumble/client.go

@@ -59,8 +59,10 @@ type Client struct {
 	ContextActions ContextActions
 
 	// The audio encoder used when sending audio to the server.
-	AudioEncoder AudioEncoder
-	audioCodec   AudioCodec
+	AudioEncoder       AudioEncoder
+	AudioEncoderStereo AudioEncoder
+	audioCodec         AudioCodec
+	useStereoEncoder   bool
 	// To whom transmitted audio will be sent. The VoiceTarget must have already
 	// been sent to the server for targeting to work correctly. Setting to nil
 	// will disable voice targeting (i.e. switch back to regular speaking).
@@ -287,3 +289,24 @@ func (c *Client) Do(f func()) {
 func (c *Client) Send(message Message) {
 	message.writeMessage(c)
 }
+
+// EnableStereoEncoder switches to stereo encoding for file playback.
+func (c *Client) EnableStereoEncoder() {
+	c.volatile.Lock()
+	defer c.volatile.Unlock()
+	c.useStereoEncoder = true
+}
+
+// DisableStereoEncoder switches back to mono encoding for voice.
+func (c *Client) DisableStereoEncoder() {
+	c.volatile.Lock()
+	defer c.volatile.Unlock()
+	c.useStereoEncoder = false
+}
+
+// IsStereoEncoderEnabled returns true if stereo encoding is currently active.
+func (c *Client) IsStereoEncoderEnabled() bool {
+	c.volatile.RLock()
+	defer c.volatile.RUnlock()
+	return c.useStereoEncoder
+}

gumble/gumble/handlers.go

@@ -1073,6 +1073,9 @@ func (c *Client) handleCodecVersion(buffer []byte) error {
 			c.volatile.Lock()
 
 			c.AudioEncoder = codec.NewEncoder()
+			// Also create a stereo encoder for file playback
+			// Import the opus package to get NewStereoEncoder
+			c.AudioEncoderStereo = nil // Will be set when needed
 
 			c.volatile.Unlock()
 		}

gumble/gumbleopenal/stream.go

@@ -1,345 +1,586 @@
 package gumbleopenal
 
 import (
-    "encoding/binary"
-    "errors"
-    "os/exec"
-    "time"
+	"encoding/binary"
+	"errors"
+	"os/exec"
+	"time"
 
-    "git.stormux.org/storm/barnard/gumble/gumble"
-    "git.stormux.org/storm/barnard/gumble/go-openal/openal"
+	"git.stormux.org/storm/barnard/audio"
+	"git.stormux.org/storm/barnard/gumble/go-openal/openal"
+	"git.stormux.org/storm/barnard/gumble/gumble"
+	"git.stormux.org/storm/barnard/noise"
 )
 
 // NoiseProcessor interface for noise suppression
 type NoiseProcessor interface {
-    ProcessSamples(samples []int16)
-    IsEnabled() bool
+	ProcessSamples(samples []int16)
+	IsEnabled() bool
+}
+
+// EffectsProcessor interface for voice effects
+type EffectsProcessor interface {
+	ProcessSamples(samples []int16)
+	IsEnabled() bool
+}
+
+// FilePlayer interface for file playback
+type FilePlayer interface {
+	GetAudioFrame() []int16
+	IsPlaying() bool
 }
 
 const (
-    maxBufferSize = 11520  // Max frame size (2880) * bytes per stereo sample (4)
+	maxBufferSize = 11520 // Max frame size (2880) * bytes per stereo sample (4)
 )
 
 var (
-    ErrState        = errors.New("gumbleopenal: invalid state")
-    ErrMic          = errors.New("gumbleopenal: microphone disconnected or misconfigured")
-    ErrInputDevice  = errors.New("gumbleopenal: invalid input device or parameters")
-    ErrOutputDevice = errors.New("gumbleopenal: invalid output device or parameters")
+	ErrState        = errors.New("gumbleopenal: invalid state")
+	ErrMic          = errors.New("gumbleopenal: microphone disconnected or misconfigured")
+	ErrInputDevice  = errors.New("gumbleopenal: invalid input device or parameters")
+	ErrOutputDevice = errors.New("gumbleopenal: invalid output device or parameters")
 )
 
 func beep() {
-    cmd := exec.Command("beep")
-    cmdout, err := cmd.Output()
-    if err != nil {
-        panic(err)
-    }
-    if cmdout != nil {
-    }
+	cmd := exec.Command("beep")
+	cmdout, err := cmd.Output()
+	if err != nil {
+		panic(err)
+	}
+	if cmdout != nil {
+	}
 }
 
 type Stream struct {
-    client *gumble.Client
-    link   gumble.Detacher
+	client *gumble.Client
+	link   gumble.Detacher
 
-    deviceSource    *openal.CaptureDevice
-    sourceFrameSize int
-    micVolume      float32
-    sourceStop     chan bool
+	deviceSource    *openal.CaptureDevice
+	sourceFormat    openal.Format
+	sourceChannels  int
+	sourceFrameSize int
+	micVolume       float32
+	sourceStop      chan bool
 
-    deviceSink  *openal.Device
-    contextSink *openal.Context
-    
-    noiseProcessor NoiseProcessor
+	deviceSink  *openal.Device
+	contextSink *openal.Context
+
+	noiseProcessor        NoiseProcessor
+	noiseProcessorRight   NoiseProcessor
+	micAGC                *audio.AGC
+	micAGCRight           *audio.AGC
+	effectsProcessor      EffectsProcessor
+	effectsProcessorRight EffectsProcessor
+	filePlayer            FilePlayer
 }
 
 func New(client *gumble.Client, inputDevice *string, outputDevice *string, test bool) (*Stream, error) {
-    frmsz := 480
-    if !test {
-        frmsz = client.Config.AudioFrameSize()
-    }
+	frmsz := 480
+	if !test {
+		frmsz = client.Config.AudioFrameSize()
+	}
 
-    // Always use mono for input device
-    idev := openal.CaptureOpenDevice(*inputDevice, gumble.AudioSampleRate, openal.FormatMono16, uint32(frmsz))
-    if idev == nil {
-        return nil, ErrInputDevice
-    }
+	inputFormat := openal.FormatStereo16
+	sourceChannels := 2
+	idev := openal.CaptureOpenDevice(*inputDevice, gumble.AudioSampleRate, inputFormat, uint32(frmsz))
+	if idev == nil {
+		inputFormat = openal.FormatMono16
+		sourceChannels = 1
+		idev = openal.CaptureOpenDevice(*inputDevice, gumble.AudioSampleRate, inputFormat, uint32(frmsz))
+	}
+	if idev == nil {
+		return nil, ErrInputDevice
+	}
 
-    odev := openal.OpenDevice(*outputDevice)
-    if odev == nil {
-        idev.CaptureCloseDevice()
-        return nil, ErrOutputDevice
-    }
+	odev := openal.OpenDevice(*outputDevice)
+	if odev == nil {
+		idev.CaptureCloseDevice()
+		return nil, ErrOutputDevice
+	}
 
-    if test {
-        idev.CaptureCloseDevice()
-        odev.CloseDevice()
-        return nil, nil
-    }
+	if test {
+		idev.CaptureCloseDevice()
+		odev.CloseDevice()
+		return nil, nil
+	}
 
-    s := &Stream{
-        client:          client,
-        sourceFrameSize: frmsz,
-        micVolume:      1.0,
-    }
+	s := &Stream{
+		client:          client,
+		sourceFormat:    inputFormat,
+		sourceChannels:  sourceChannels,
+		sourceFrameSize: frmsz,
+		micVolume:       1.0,
+		micAGC:          audio.NewAGC(), // Always enable AGC for outgoing mic
+	}
+	if sourceChannels == 2 {
+		s.micAGCRight = audio.NewAGC()
+	}
 
-    s.deviceSource = idev
-    if s.deviceSource == nil {
-        return nil, ErrInputDevice
-    }
+	s.deviceSource = idev
+	if s.deviceSource == nil {
+		return nil, ErrInputDevice
+	}
 
-    s.deviceSink = odev
-    if s.deviceSink == nil {
-        return nil, ErrOutputDevice
-    }
-    s.contextSink = s.deviceSink.CreateContext()
-    if s.contextSink == nil {
-        s.Destroy()
-        return nil, ErrOutputDevice
-    }
-    s.contextSink.Activate()
+	s.deviceSink = odev
+	if s.deviceSink == nil {
+		return nil, ErrOutputDevice
+	}
+	s.contextSink = s.deviceSink.CreateContext()
+	if s.contextSink == nil {
+		s.Destroy()
+		return nil, ErrOutputDevice
+	}
+	s.contextSink.Activate()
 
-    return s, nil
+	return s, nil
 }
 
 func (s *Stream) AttachStream(client *gumble.Client) {
-    s.link = client.Config.AttachAudio(s)
+	s.link = client.Config.AttachAudio(s)
 }
 
 func (s *Stream) SetNoiseProcessor(np NoiseProcessor) {
-    s.noiseProcessor = np
+	s.noiseProcessor = np
+	s.noiseProcessorRight = cloneNoiseProcessor(np)
+}
+
+func (s *Stream) SetEffectsProcessor(ep EffectsProcessor) {
+	s.effectsProcessor = ep
+	s.effectsProcessorRight = cloneEffectsProcessor(ep)
+}
+
+func (s *Stream) GetEffectsProcessor() EffectsProcessor {
+	return s.effectsProcessor
+}
+
+func (s *Stream) SetFilePlayer(fp FilePlayer) {
+	s.filePlayer = fp
+}
+
+func (s *Stream) GetFilePlayer() FilePlayer {
+	return s.filePlayer
 }
 
 func (s *Stream) Destroy() {
-    if s.link != nil {
-        s.link.Detach()
-    }
-    if s.deviceSource != nil {
-        s.StopSource()
-        s.deviceSource.CaptureCloseDevice()
-        s.deviceSource = nil
-    }
-    if s.deviceSink != nil {
-        s.contextSink.Destroy()
-        s.deviceSink.CloseDevice()
-        s.contextSink = nil
-        s.deviceSink = nil
-    }
+	if s.link != nil {
+		s.link.Detach()
+	}
+	if s.deviceSource != nil {
+		s.StopSource()
+		s.deviceSource.CaptureCloseDevice()
+		s.deviceSource = nil
+	}
+	if s.deviceSink != nil {
+		s.contextSink.Destroy()
+		s.deviceSink.CloseDevice()
+		s.contextSink = nil
+		s.deviceSink = nil
+	}
 }
 
 func (s *Stream) StartSource(inputDevice *string) error {
-    if s.sourceStop != nil {
-        return ErrState
-    }
-    if s.deviceSource == nil {
-        return ErrMic
-    }
-    s.deviceSource.CaptureStart()
-    s.sourceStop = make(chan bool)
-    go s.sourceRoutine(inputDevice)
-    return nil
+	if s.sourceStop != nil {
+		return ErrState
+	}
+	if s.deviceSource == nil {
+		return ErrMic
+	}
+	s.deviceSource.CaptureStart()
+	s.sourceStop = make(chan bool)
+	go s.sourceRoutine(inputDevice)
+	return nil
 }
 
 func (s *Stream) StopSource() error {
-    if s.deviceSource == nil {
-        return ErrMic
-    }
-    s.deviceSource.CaptureStop()
-    if s.sourceStop == nil {
-        return ErrState
-    }
-    close(s.sourceStop)
-    s.sourceStop = nil
-    return nil
+	if s.deviceSource == nil {
+		return ErrMic
+	}
+	s.deviceSource.CaptureStop()
+	if s.sourceStop == nil {
+		return ErrState
+	}
+	close(s.sourceStop)
+	s.sourceStop = nil
+	return nil
 }
 
 func (s *Stream) GetMicVolume() float32 {
-    return s.micVolume
+	return s.micVolume
 }
 
 func (s *Stream) SetMicVolume(change float32, relative bool) {
-    var val float32
-    if relative {
-        val = s.GetMicVolume() + change
-    } else {
-        val = change
-    }
-    if val >= 1 {
-        val = 1.0
-    }
-    if val <= 0 {
-        val = 0
-    }
-    s.micVolume = val
+	var val float32
+	if relative {
+		val = s.GetMicVolume() + change
+	} else {
+		val = change
+	}
+	if val >= 1 {
+		val = 1.0
+	}
+	if val <= 0 {
+		val = 0
+	}
+	s.micVolume = val
 }
 
 func (s *Stream) OnAudioStream(e *gumble.AudioStreamEvent) {
-    go func(e *gumble.AudioStreamEvent) {
-        var source = openal.NewSource()
-        e.User.AudioSource = &source
-        
-        // Set initial gain based on volume and mute state
-        if e.User.LocallyMuted {
-            e.User.AudioSource.SetGain(0)
-        } else {
-            e.User.AudioSource.SetGain(e.User.Volume)
-        }
+	go func(e *gumble.AudioStreamEvent) {
+		var source = openal.NewSource()
+		e.User.AudioSource = &source
 
-        bufferCount := e.Client.Config.Buffers
-        if bufferCount < 64 {
-            bufferCount = 64
-        }
-        emptyBufs := openal.NewBuffers(bufferCount)
-        
-        reclaim := func() {
-            if n := source.BuffersProcessed(); n > 0 {
-                reclaimedBufs := make(openal.Buffers, n)
-                source.UnqueueBuffers(reclaimedBufs)
-                emptyBufs = append(emptyBufs, reclaimedBufs...)
-            }
-        }
+		// Set initial gain based on volume and mute state
+		if e.User.LocallyMuted {
+			e.User.AudioSource.SetGain(0)
+		} else {
+			e.User.AudioSource.SetGain(e.User.Volume)
+		}
 
-        var raw [maxBufferSize]byte
-        
-        for packet := range e.C {
-            // Skip processing if user is locally muted
-            if e.User.LocallyMuted {
-                continue
-            }
+		bufferCount := e.Client.Config.Buffers
+		if bufferCount < 64 {
+			bufferCount = 64
+		}
+		emptyBufs := openal.NewBuffers(bufferCount)
 
-            var boost uint16 = uint16(1)
-            samples := len(packet.AudioBuffer)
-            if samples > cap(raw)/2 {
-                continue
-            }
-            
-            boost = e.User.Boost
+		reclaim := func() {
+			if n := source.BuffersProcessed(); n > 0 {
+				reclaimedBufs := make(openal.Buffers, n)
+				source.UnqueueBuffers(reclaimedBufs)
+				emptyBufs = append(emptyBufs, reclaimedBufs...)
+			}
+		}
 
-            // Check if sample count suggests stereo data
-            isStereo := samples > gumble.AudioDefaultFrameSize && samples%2 == 0
-            format := openal.FormatMono16
-            if isStereo {
-                format = openal.FormatStereo16
-                samples = samples / 2
-            }
+		var raw [maxBufferSize]byte
 
-            rawPtr := 0
-            if isStereo {
-                // Process stereo samples as pairs
-                for i := 0; i < samples*2; i += 2 {
-                    // Process left channel with saturation protection
-                    sample := packet.AudioBuffer[i]
-                    if boost > 1 {
-                        boosted := int32(sample) * int32(boost)
-                        if boosted > 32767 {
-                            sample = 32767
-                        } else if boosted < -32767 {
-                            sample = -32767
-                        } else {
-                            sample = int16(boosted)
-                        }
-                    }
-                    binary.LittleEndian.PutUint16(raw[rawPtr:], uint16(sample))
-                    rawPtr += 2
+		for packet := range e.C {
+			// Skip processing if user is locally muted
+			if e.User.LocallyMuted {
+				continue
+			}
 
-                    // Process right channel with saturation protection
-                    sample = packet.AudioBuffer[i+1]
-                    if boost > 1 {
-                        boosted := int32(sample) * int32(boost)
-                        if boosted > 32767 {
-                            sample = 32767
-                        } else if boosted < -32767 {
-                            sample = -32767
-                        } else {
-                            sample = int16(boosted)
-                        }
-                    }
-                    binary.LittleEndian.PutUint16(raw[rawPtr:], uint16(sample))
-                    rawPtr += 2
-                }
-            } else {
-                // Process mono samples with saturation protection
-                for i := 0; i < samples; i++ {
-                    sample := packet.AudioBuffer[i]
-                    if boost > 1 {
-                        boosted := int32(sample) * int32(boost)
-                        if boosted > 32767 {
-                            sample = 32767
-                        } else if boosted < -32767 {
-                            sample = -32767
-                        } else {
-                            sample = int16(boosted)
-                        }
-                    }
-                    binary.LittleEndian.PutUint16(raw[rawPtr:], uint16(sample))
-                    rawPtr += 2
-                }
-            }
+			var boost uint16 = uint16(1)
+			samples := len(packet.AudioBuffer)
+			if samples > cap(raw)/2 {
+				continue
+			}
 
-            reclaim()
-            if len(emptyBufs) == 0 {
-                continue
-            }
+			boost = e.User.Boost
 
-            last := len(emptyBufs) - 1
-            buffer := emptyBufs[last]
-            emptyBufs = emptyBufs[:last]
+			// Check if sample count suggests stereo data
+			isStereo := samples > gumble.AudioDefaultFrameSize && samples%2 == 0
+			format := openal.FormatMono16
+			if isStereo {
+				format = openal.FormatStereo16
+				samples = samples / 2
+			}
 
-            buffer.SetData(format, raw[:rawPtr], gumble.AudioSampleRate)
-            source.QueueBuffer(buffer)
+			rawPtr := 0
+			if isStereo {
+				// Process stereo samples as pairs
+				for i := 0; i < samples*2; i += 2 {
+					// Process left channel with saturation protection
+					sample := packet.AudioBuffer[i]
+					if boost > 1 {
+						boosted := int32(sample) * int32(boost)
+						if boosted > 32767 {
+							sample = 32767
+						} else if boosted < -32767 {
+							sample = -32767
+						} else {
+							sample = int16(boosted)
+						}
+					}
+					binary.LittleEndian.PutUint16(raw[rawPtr:], uint16(sample))
+					rawPtr += 2
 
-            if source.State() != openal.Playing {
-                source.Play()
-            }
-        }
-        reclaim()
-        emptyBufs.Delete()
-        source.Delete()
-    }(e)
+					// Process right channel with saturation protection
+					sample = packet.AudioBuffer[i+1]
+					if boost > 1 {
+						boosted := int32(sample) * int32(boost)
+						if boosted > 32767 {
+							sample = 32767
+						} else if boosted < -32767 {
+							sample = -32767
+						} else {
+							sample = int16(boosted)
+						}
+					}
+					binary.LittleEndian.PutUint16(raw[rawPtr:], uint16(sample))
+					rawPtr += 2
+				}
+			} else {
+				// Process mono samples with saturation protection
+				for i := 0; i < samples; i++ {
+					sample := packet.AudioBuffer[i]
+					if boost > 1 {
+						boosted := int32(sample) * int32(boost)
+						if boosted > 32767 {
+							sample = 32767
+						} else if boosted < -32767 {
+							sample = -32767
+						} else {
+							sample = int16(boosted)
+						}
+					}
+					binary.LittleEndian.PutUint16(raw[rawPtr:], uint16(sample))
+					rawPtr += 2
+				}
+			}
+
+			reclaim()
+			if len(emptyBufs) == 0 {
+				continue
+			}
+
+			last := len(emptyBufs) - 1
+			buffer := emptyBufs[last]
+			emptyBufs = emptyBufs[:last]
+
+			buffer.SetData(format, raw[:rawPtr], gumble.AudioSampleRate)
+			source.QueueBuffer(buffer)
+
+			if source.State() != openal.Playing {
+				source.Play()
+			}
+		}
+		reclaim()
+		emptyBufs.Delete()
+		source.Delete()
+	}(e)
 }
 
 func (s *Stream) sourceRoutine(inputDevice *string) {
-    interval := s.client.Config.AudioInterval
-    frameSize := s.client.Config.AudioFrameSize()
+	interval := s.client.Config.AudioInterval
+	frameSize := s.client.Config.AudioFrameSize()
 
-    if frameSize != s.sourceFrameSize {
-        s.deviceSource.CaptureCloseDevice()
-        s.sourceFrameSize = frameSize
-        // Always use mono for input
-        s.deviceSource = openal.CaptureOpenDevice(*inputDevice, gumble.AudioSampleRate, openal.FormatMono16, uint32(s.sourceFrameSize))
-    }
+	if frameSize != s.sourceFrameSize {
+		s.deviceSource.CaptureCloseDevice()
+		s.sourceFrameSize = frameSize
+		s.deviceSource = openal.CaptureOpenDevice(*inputDevice, gumble.AudioSampleRate, s.sourceFormat, uint32(s.sourceFrameSize))
+		if s.deviceSource == nil && s.sourceFormat == openal.FormatStereo16 {
+			s.sourceFormat = openal.FormatMono16
+			s.sourceChannels = 1
+			s.deviceSource = openal.CaptureOpenDevice(*inputDevice, gumble.AudioSampleRate, s.sourceFormat, uint32(s.sourceFrameSize))
+		}
+	}
+	if s.deviceSource == nil {
+		return
+	}
 
-    ticker := time.NewTicker(interval)
-    defer ticker.Stop()
+	ticker := time.NewTicker(interval)
+	defer ticker.Stop()
 
-    stop := s.sourceStop
+	stop := s.sourceStop
 
-    outgoing := s.client.AudioOutgoing()
-    defer close(outgoing)
+	outgoing := s.client.AudioOutgoing()
+	defer close(outgoing)
 
-    for {
-        select {
-        case <-stop:
-            return
-        case <-ticker.C:
-            buff := s.deviceSource.CaptureSamples(uint32(frameSize))
-            if len(buff) != frameSize*2 {
-                continue
-            }
-            int16Buffer := make([]int16, frameSize)
-            for i := range int16Buffer {
-                sample := int16(binary.LittleEndian.Uint16(buff[i*2:]))
-                if s.micVolume != 1.0 {
-                    sample = int16(float32(sample) * s.micVolume)
-                }
-                int16Buffer[i] = sample
-            }
-            
-            // Apply noise suppression if available and enabled
-            if s.noiseProcessor != nil && s.noiseProcessor.IsEnabled() {
-                s.noiseProcessor.ProcessSamples(int16Buffer)
-            }
-            
-            outgoing <- gumble.AudioBuffer(int16Buffer)
-        }
-    }
+	for {
+		select {
+		case <-stop:
+			return
+		case <-ticker.C:
+			sampleCount := frameSize * s.sourceChannels
+			int16Buffer := make([]int16, sampleCount)
+
+			// Capture microphone if available
+			hasMicInput := false
+			buff := s.deviceSource.CaptureSamples(uint32(frameSize))
+			if len(buff) == sampleCount*2 {
+				hasMicInput = true
+				for i := 0; i < sampleCount; i++ {
+					sample := int16(binary.LittleEndian.Uint16(buff[i*2:]))
+					if s.micVolume != 1.0 {
+						sample = int16(float32(sample) * s.micVolume)
+					}
+					int16Buffer[i] = sample
+				}
+
+				if s.sourceChannels == 1 {
+					s.processMonoSamples(int16Buffer)
+				} else {
+					s.processStereoSamples(int16Buffer, frameSize)
+				}
+			}
+
+			// Mix with or use file audio if playing
+			hasFileAudio := false
+			var outputBuffer []int16
+
+			if s.filePlayer != nil && s.filePlayer.IsPlaying() {
+				fileAudio := s.filePlayer.GetAudioFrame()
+				if fileAudio != nil && len(fileAudio) > 0 {
+					hasFileAudio = true
+					// File audio is stereo - send as stereo when file is playing
+					// Create stereo buffer (frameSize * 2 channels)
+					outputBuffer = make([]int16, frameSize*2)
+
+					if hasMicInput {
+						if s.sourceChannels == 2 {
+							// Mix stereo mic with stereo file
+							for i := 0; i < frameSize; i++ {
+								idx := i * 2
+								if idx+1 < len(fileAudio) {
+									left := int32(int16Buffer[idx]) + int32(fileAudio[idx])
+									if left > 32767 {
+										left = 32767
+									} else if left < -32768 {
+										left = -32768
+									}
+									outputBuffer[idx] = int16(left)
+
+									right := int32(int16Buffer[idx+1]) + int32(fileAudio[idx+1])
+									if right > 32767 {
+										right = 32767
+									} else if right < -32768 {
+										right = -32768
+									}
+									outputBuffer[idx+1] = int16(right)
+								}
+							}
+						} else {
+							// Mix mono mic with stereo file
+							for i := 0; i < frameSize; i++ {
+								idx := i * 2
+								if idx+1 < len(fileAudio) {
+									left := int32(int16Buffer[i]) + int32(fileAudio[idx])
+									if left > 32767 {
+										left = 32767
+									} else if left < -32768 {
+										left = -32768
+									}
+									outputBuffer[idx] = int16(left)
+
+									right := int32(int16Buffer[i]) + int32(fileAudio[idx+1])
+									if right > 32767 {
+										right = 32767
+									} else if right < -32768 {
+										right = -32768
+									}
+									outputBuffer[idx+1] = int16(right)
+								}
+							}
+						}
+					} else {
+						// Use file audio only (already stereo)
+						copy(outputBuffer, fileAudio[:frameSize*2])
+					}
+				}
+			}
+
+			// Determine what to send
+			if hasFileAudio {
+				// Send stereo buffer when file is playing
+				outgoing <- gumble.AudioBuffer(outputBuffer)
+			} else if hasMicInput {
+				// Send mic when no file is playing
+				outgoing <- gumble.AudioBuffer(int16Buffer)
+			}
+		}
+	}
+}
+
+func (s *Stream) processMonoSamples(samples []int16) {
+	s.processChannel(samples, s.noiseProcessor, s.micAGC, s.effectsProcessor)
+}
+
+func (s *Stream) processStereoSamples(samples []int16, frameSize int) {
+	if frameSize == 0 || len(samples) < frameSize*2 {
+		return
+	}
+
+	s.ensureStereoProcessors()
+	s.syncStereoProcessors()
+
+	left := make([]int16, frameSize)
+	right := make([]int16, frameSize)
+
+	for i := 0; i < frameSize; i++ {
+		idx := i * 2
+		left[i] = samples[idx]
+		right[i] = samples[idx+1]
+	}
+
+	s.processChannel(left, s.noiseProcessor, s.micAGC, s.effectsProcessor)
+	s.processChannel(right, s.noiseProcessorRight, s.micAGCRight, s.effectsProcessorRight)
+
+	for i := 0; i < frameSize; i++ {
+		idx := i * 2
+		samples[idx] = left[i]
+		samples[idx+1] = right[i]
+	}
+}
+
+func (s *Stream) processChannel(samples []int16, noiseProcessor NoiseProcessor, micAGC *audio.AGC, effectsProcessor EffectsProcessor) {
+	if noiseProcessor != nil && noiseProcessor.IsEnabled() {
+		noiseProcessor.ProcessSamples(samples)
+	}
+	if micAGC != nil {
+		micAGC.ProcessSamples(samples)
+	}
+	if effectsProcessor != nil && effectsProcessor.IsEnabled() {
+		effectsProcessor.ProcessSamples(samples)
+	}
+}
+
+func (s *Stream) ensureStereoProcessors() {
+	if s.micAGCRight == nil {
+		s.micAGCRight = audio.NewAGC()
+	}
+	if s.noiseProcessorRight == nil {
+		s.noiseProcessorRight = cloneNoiseProcessor(s.noiseProcessor)
+	}
+	if s.effectsProcessorRight == nil {
+		s.effectsProcessorRight = cloneEffectsProcessor(s.effectsProcessor)
+	}
+}
+
+func (s *Stream) syncStereoProcessors() {
+	leftSuppressor, leftOk := s.noiseProcessor.(*noise.Suppressor)
+	rightSuppressor, rightOk := s.noiseProcessorRight.(*noise.Suppressor)
+	if leftOk && rightOk {
+		if leftSuppressor.IsEnabled() != rightSuppressor.IsEnabled() {
+			rightSuppressor.SetEnabled(leftSuppressor.IsEnabled())
+		}
+		if leftSuppressor.GetThreshold() != rightSuppressor.GetThreshold() {
+			rightSuppressor.SetThreshold(leftSuppressor.GetThreshold())
+		}
+	}
+
+	leftEffects, leftOk := s.effectsProcessor.(*audio.EffectsProcessor)
+	rightEffects, rightOk := s.effectsProcessorRight.(*audio.EffectsProcessor)
+	if leftOk && rightOk {
+		if leftEffects.IsEnabled() != rightEffects.IsEnabled() {
+			rightEffects.SetEnabled(leftEffects.IsEnabled())
+		}
+		if leftEffects.GetCurrentEffect() != rightEffects.GetCurrentEffect() {
+			rightEffects.SetEffect(leftEffects.GetCurrentEffect())
+		}
+	}
+}
+
+func cloneNoiseProcessor(np NoiseProcessor) NoiseProcessor {
+	if np == nil {
+		return nil
+	}
+	if suppressor, ok := np.(*noise.Suppressor); ok {
+		clone := noise.NewSuppressor()
+		clone.SetEnabled(suppressor.IsEnabled())
+		clone.SetThreshold(suppressor.GetThreshold())
+		return clone
+	}
+	return nil
+}
+
+func cloneEffectsProcessor(ep EffectsProcessor) EffectsProcessor {
+	if ep == nil {
+		return nil
+	}
+	if processor, ok := ep.(*audio.EffectsProcessor); ok {
+		clone := audio.NewEffectsProcessor(gumble.AudioSampleRate)
+		clone.SetEnabled(processor.IsEnabled())
+		clone.SetEffect(processor.GetCurrentEffect())
+		return clone
+	}
+	return nil
 }

gumble/opus/opus.go

@@ -35,6 +35,16 @@ func (*generator) NewEncoder() gumble.AudioEncoder {
     }
 }
 
+// NewStereoEncoder creates a stereo encoder for file playback
+func NewStereoEncoder() gumble.AudioEncoder {
+    // Create stereo encoder for file playback
+    e, _ := opus.NewEncoder(gumble.AudioSampleRate, gumble.AudioChannels, opus.AppAudio)
+    e.SetBitrateToMax()
+    return &Encoder{
+        e,
+    }
+}
+
 func (*generator) NewDecoder() gumble.AudioDecoder {
     // Create decoder with stereo support
     d, _ := opus.NewDecoder(gumble.AudioSampleRate, gumble.AudioChannels)

main.go

@@ -15,14 +15,15 @@ import (
 	//"github.com/google/shlex"
 	"crypto/tls"
 	"flag"
-	"github.com/alessio/shellescape"
+	"git.stormux.org/storm/barnard/audio"
 	"git.stormux.org/storm/barnard/config"
 	"git.stormux.org/storm/barnard/noise"
+	"github.com/alessio/shellescape"
 
+	"git.stormux.org/storm/barnard/gumble/go-openal/openal"
 	"git.stormux.org/storm/barnard/gumble/gumble"
 	_ "git.stormux.org/storm/barnard/gumble/opus"
 	"git.stormux.org/storm/barnard/uiterm"
-	"git.stormux.org/storm/barnard/gumble/go-openal/openal"
 )
 
 func show_devs(name string, args []string) {
@@ -107,7 +108,8 @@ func main() {
 	password := flag.String("password", "", "the password of the server")
 	insecure := flag.Bool("insecure", false, "skip server certificate verification")
 	certificate := flag.String("certificate", "", "PEM encoded certificate and private key")
-	cfgfn := flag.String("config", "~/.barnard.yaml", "Path to YAML formatted configuration file")
+	cfgfn := flag.String("config", "~/.barnard.toml", "Path to TOML formatted configuration file")
+	audioDriver := flag.String("audio-driver", "", "preferred OpenAL backend (pipewire, pulse, alsa, jack)")
 	list_devices := flag.Bool("list_devices", false, "do not connect; instead, list available audio devices and exit")
 	fifo := flag.String("fifo", "", "path of a FIFO from which to read commands")
 	serverSet := false
@@ -126,12 +128,15 @@ func main() {
 
 	userConfig := config.NewConfig(cfgfn)
 
+	certificateSet := false
 	flag.CommandLine.Visit(func(theFlag *flag.Flag) {
 		switch theFlag.Name {
 		case "server":
 			serverSet = true
 		case "username":
 			usernameSet = true
+		case "certificate":
+			certificateSet = true
 		}
 	})
 
@@ -141,6 +146,22 @@ func main() {
 	if !usernameSet {
 		username = userConfig.GetUsername()
 	}
+	if !certificateSet {
+		certificate = userConfig.GetCertificate()
+	}
+
+	driver := strings.TrimSpace(*audioDriver)
+	if driver == "" {
+		// Environment variable takes precedence over config
+		if envDriver := os.Getenv("ALSOFT_DRIVERS"); envDriver != "" {
+			driver = envDriver
+		} else {
+			driver = strings.TrimSpace(userConfig.GetAudioDriver())
+		}
+	}
+	if driver != "" {
+		os.Setenv("ALSOFT_DRIVERS", driver)
+	}
 
 	if os.Getenv("ALSOFT_LOGLEVEL") == "" {
 		os.Setenv("ALSOFT_LOGLEVEL", "0")
@@ -157,17 +178,18 @@ func main() {
 
 	// Initialize
 	b := Barnard{
-		Config:     gumble.NewConfig(),
-		UserConfig: userConfig,
-		Address:    *server,
-		MutedChannels: make(map[uint32]bool),
+		Config:          gumble.NewConfig(),
+		UserConfig:      userConfig,
+		Address:         *server,
+		MutedChannels:   make(map[uint32]bool),
 		NoiseSuppressor: noise.NewSuppressor(),
+		VoiceEffects:    audio.NewEffectsProcessor(gumble.AudioSampleRate),
 	}
 	b.Config.Buffers = *buffers
 
 	b.Hotkeys = b.UserConfig.GetHotkeys()
 	b.UserConfig.SaveConfig()
-	
+
 	// Configure noise suppression
 	enabled := b.UserConfig.GetNoiseSuppressionEnabled()
 	if *noiseSuppressionEnabled {
@@ -176,7 +198,10 @@ func main() {
 	}
 	b.NoiseSuppressor.SetEnabled(enabled)
 	b.NoiseSuppressor.SetThreshold(b.UserConfig.GetNoiseSuppressionThreshold())
-	
+
+	// Configure voice effects
+	b.VoiceEffects.SetEffect(audio.VoiceEffect(b.UserConfig.GetVoiceEffect()))
+
 	b.Config.Username = *username
 	b.Config.Password = *password
 
@@ -196,12 +221,12 @@ func main() {
 	if err != nil {
 		b.exitMessage = err.Error()
 		b.exitStatus = 1
-		handle_error(b)
+		handle_error(&b)
 	}
 	b.notifyChannel = setup_notify_runner(*b.UserConfig.GetNotifyCommand())
 	b.Ui = uiterm.New(&b)
 	b.Ui.Run(reader)
-	handle_error(b)
+	handle_error(&b)
 }
 
 func handle_raw_error(e error) {
@@ -209,7 +234,7 @@ func handle_raw_error(e error) {
 	os.Exit(1)
 }
 
-func handle_error(b Barnard) {
+func handle_error(b *Barnard) {
 	if b.exitMessage != "" {
 		fmt.Fprintf(os.Stderr, "%s\n", b.exitMessage)
 	}

noise/suppression.go

@@ -2,6 +2,7 @@ package noise
 
 import (
 	"math"
+	"sync"
 )
 
 // Ensure Suppressor implements the NoiseProcessor interface
@@ -12,129 +13,200 @@ var _ interface {
 
 // Suppressor handles noise suppression for audio samples
 type Suppressor struct {
-	enabled    bool
-	threshold  float32
-	gainFactor float32
-	
-	// Simple high-pass filter state for DC removal
+	mu sync.Mutex
+
+	enabled   bool
+	threshold float32
+
+	// High-pass filter state for low-frequency rumble/DC removal.
 	prevInput  float32
 	prevOutput float32
-	alpha      float32
-	
-	// Click detection state
-	clickThreshold    float32
-	clickDecay        float32
-	recentClickEnergy float32
+	hpAlpha    float32
+
+	// Adaptive suppression state.
+	envelope        float32
+	noiseFloor      float32
+	suppressionGain float32
+	clickEnergy     float32
+
+	// Tunables.
+	envelopeAttack  float32
+	envelopeRelease float32
+	noiseAttack     float32
+	noiseRelease    float32
+	gainAttack      float32
+	gainRelease     float32
+	speechRatio     float32
+	clickDecay      float32
+	minNoiseFloor   float32
 }
 
 // NewSuppressor creates a new noise suppressor
 func NewSuppressor() *Suppressor {
-	return &Suppressor{
-		enabled:           false,
-		threshold:         0.01,     // Reduced noise threshold level for less aggressive filtering
-		gainFactor:        0.9,      // Less aggressive gain reduction for noise
-		alpha:             0.98,     // More stable high-pass filter coefficient
-		clickThreshold:    0.15,     // Threshold for detecting keyboard clicks
-		clickDecay:        0.95,     // How quickly click energy decays
-		recentClickEnergy: 0.0,      // Tracks recent click activity
+	s := &Suppressor{
+		enabled:         false,
+		threshold:       0.08,
+		hpAlpha:         0.995,
+		envelopeAttack:  0.18,
+		envelopeRelease: 0.02,
+		noiseAttack:     0.08,
+		noiseRelease:    0.002,
+		gainAttack:      0.35,
+		gainRelease:     0.02,
+		speechRatio:     4.0,
+		clickDecay:      0.93,
+		minNoiseFloor:   0.0008,
+		suppressionGain: 1.0,
 	}
+	s.resetStateLocked()
+	return s
 }
 
 // SetEnabled enables or disables noise suppression
 func (s *Suppressor) SetEnabled(enabled bool) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if s.enabled == enabled {
+		return
+	}
 	s.enabled = enabled
+	s.resetStateLocked()
 }
 
 // IsEnabled returns whether noise suppression is enabled
 func (s *Suppressor) IsEnabled() bool {
+	s.mu.Lock()
+	defer s.mu.Unlock()
 	return s.enabled
 }
 
 // SetThreshold sets the noise threshold (0.0 to 1.0)
 func (s *Suppressor) SetThreshold(threshold float32) {
-	if threshold >= 0.0 && threshold <= 1.0 {
-		s.threshold = threshold
-	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	s.threshold = clampFloat32(threshold, 0.0, 1.0)
 }
 
 // GetThreshold returns the current noise threshold
 func (s *Suppressor) GetThreshold() float32 {
+	s.mu.Lock()
+	defer s.mu.Unlock()
 	return s.threshold
 }
 
 // ProcessSamples applies noise suppression to audio samples
 func (s *Suppressor) ProcessSamples(samples []int16) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+
 	if !s.enabled || len(samples) == 0 {
 		return
 	}
 
-	// Calculate frame energy for click detection
-	var frameEnergy float32 = 0.0
-	for _, sample := range samples {
-		floatSample := float32(sample) / 32767.0
-		frameEnergy += floatSample * floatSample
-	}
-	frameEnergy = float32(math.Sqrt(float64(frameEnergy / float32(len(samples)))))
-	
-	// Detect sudden energy spikes (likely keyboard clicks)
-	energySpike := frameEnergy - s.recentClickEnergy
-	isClick := energySpike > s.clickThreshold && frameEnergy > 0.05
-	
-	// Update recent click energy with decay
-	s.recentClickEnergy = s.recentClickEnergy*s.clickDecay + frameEnergy*(1.0-s.clickDecay)
-	
-	// Improved noise suppression algorithm
+	intensity := s.thresholdToIntensity()
+	minGain := 1.0 - (0.92 * intensity)
+	eps := float32(1e-6)
+
 	for i, sample := range samples {
-		// Convert to float for processing
-		floatSample := float32(sample) / 32767.0
-		
-		// Apply high-pass filter for DC removal
-		filtered := s.highPassFilter(floatSample)
-		
-		// Calculate signal strength (RMS-like)
-		strength := float32(math.Abs(float64(filtered)))
-		
-		// Apply noise gate with smooth transition
-		var gainReduction float32 = 1.0
-		
-		// If we detected a click, apply stronger suppression
-		if isClick {
-			gainReduction = s.gainFactor * 0.3 // Much stronger reduction for clicks
-		} else if strength < s.threshold {
-			// Normal noise gate for low-level sounds
-			gainReduction = strength / s.threshold
-			if gainReduction < s.gainFactor {
-				gainReduction = s.gainFactor
-			}
-		}
-		
-		// Apply gain reduction
-		processed := filtered * gainReduction
-		
-		// Convert back to int16 with proper clipping
-		processedInt := processed * 32767.0
-		if processedInt > 32767 {
-			processedInt = 32767
-		} else if processedInt < -32767 {
-			processedInt = -32767
-		}
-		
-		samples[i] = int16(processedInt)
+		floatSample := float32(sample) / 32768.0
+		filtered := s.highPassFilterLocked(floatSample)
+		absSample := float32(math.Abs(float64(filtered)))
+
+		s.updateEnvelopeLocked(absSample)
+		s.updateNoiseFloorLocked()
+
+		snr := s.envelope / (s.noiseFloor + eps)
+		voicePresence := clampFloat32((snr-1.0)/(s.speechRatio-1.0), 0.0, 1.0)
+
+		targetGain := minGain + ((1.0 - minGain) * voicePresence)
+		targetGain = s.applyTransientSuppressionLocked(absSample, voicePresence, minGain, targetGain)
+
+		s.applyGainSmoothingLocked(targetGain)
+
+		processed := filtered * s.suppressionGain
+		processed = clampFloat32(processed, -1.0, 1.0)
+		samples[i] = int16(processed * 32767.0)
 	}
 }
 
-// highPassFilter applies a simple high-pass filter to remove DC component
-func (s *Suppressor) highPassFilter(input float32) float32 {
+func (s *Suppressor) highPassFilterLocked(input float32) float32 {
 	// Simple high-pass filter: y[n] = alpha * (y[n-1] + x[n] - x[n-1])
-	output := s.alpha * (s.prevOutput + input - s.prevInput)
+	output := s.hpAlpha * (s.prevOutput + input - s.prevInput)
 	s.prevInput = input
 	s.prevOutput = output
 	return output
 }
 
+func (s *Suppressor) thresholdToIntensity() float32 {
+	// Keep lower legacy threshold values meaningful while allowing up to very aggressive suppression.
+	return 1.0 - float32(math.Exp(float64(-28.0*clampFloat32(s.threshold, 0.0, 1.0))))
+}
+
+func (s *Suppressor) updateEnvelopeLocked(absSample float32) {
+	if absSample > s.envelope {
+		s.envelope += s.envelopeAttack * (absSample - s.envelope)
+	} else {
+		s.envelope += s.envelopeRelease * (absSample - s.envelope)
+	}
+	if s.envelope < s.minNoiseFloor {
+		s.envelope = s.minNoiseFloor
+	}
+}
+
+func (s *Suppressor) updateNoiseFloorLocked() {
+	coef := s.noiseRelease
+	if s.envelope < s.noiseFloor*2.2 {
+		coef = s.noiseAttack
+	}
+	s.noiseFloor += coef * (s.envelope - s.noiseFloor)
+	if s.noiseFloor < s.minNoiseFloor {
+		s.noiseFloor = s.minNoiseFloor
+	}
+}
+
+func (s *Suppressor) applyTransientSuppressionLocked(absSample float32, voicePresence float32, minGain float32, targetGain float32) float32 {
+	s.clickEnergy = (s.clickEnergy * s.clickDecay) + (absSample * (1.0 - s.clickDecay))
+	transient := absSample - s.clickEnergy
+	transientThreshold := 0.04 + (0.08 * (1.0 - voicePresence))
+	if transient > transientThreshold && voicePresence < 0.65 {
+		clickGain := minGain * 0.55
+		if clickGain < targetGain {
+			targetGain = clickGain
+		}
+	}
+	return clampFloat32(targetGain, 0.02, 1.0)
+}
+
+func (s *Suppressor) applyGainSmoothingLocked(targetGain float32) {
+	if targetGain < s.suppressionGain {
+		s.suppressionGain += s.gainAttack * (targetGain - s.suppressionGain)
+	} else {
+		s.suppressionGain += s.gainRelease * (targetGain - s.suppressionGain)
+	}
+	s.suppressionGain = clampFloat32(s.suppressionGain, 0.02, 1.0)
+}
+
+func (s *Suppressor) resetStateLocked() {
+	s.prevInput = 0.0
+	s.prevOutput = 0.0
+	s.envelope = s.minNoiseFloor
+	s.noiseFloor = s.minNoiseFloor
+	s.suppressionGain = 1.0
+	s.clickEnergy = 0.0
+}
+
+func clampFloat32(value float32, min float32, max float32) float32 {
+	if value < min {
+		return min
+	}
+	if value > max {
+		return max
+	}
+	return value
+}
+
 // ProcessSamplesAdvanced applies more sophisticated noise suppression
-// This is a placeholder for future RNNoise integration
+// Placeholder for future RNNoise integration.
 func (s *Suppressor) ProcessSamplesAdvanced(samples []int16) {
-	// TODO: Integrate RNNoise or other advanced algorithms
 	s.ProcessSamples(samples)
-}
+}

noise/suppression_test.go

@@ -0,0 +1,103 @@
+package noise
+
+import (
+	"math"
+	"testing"
+)
+
+func TestSuppressorDisabledBypassesSamples(t *testing.T) {
+	suppressor := NewSuppressor()
+	samples := []int16{100, -200, 300, -400, 500}
+	original := append([]int16(nil), samples...)
+
+	suppressor.ProcessSamples(samples)
+
+	for i := range samples {
+		if samples[i] != original[i] {
+			t.Fatalf("expected sample %d to remain unchanged, got %d want %d", i, samples[i], original[i])
+		}
+	}
+}
+
+func TestSuppressorAttenuatesLowLevelNoise(t *testing.T) {
+	suppressor := NewSuppressor()
+	suppressor.SetEnabled(true)
+	suppressor.SetThreshold(0.08)
+
+	input := makeSineFrame(600, 700)
+	originalRMS := frameRMS(input)
+	processed := append([]int16(nil), input...)
+	suppressor.ProcessSamples(processed)
+	processedRMS := frameRMS(processed)
+
+	if processedRMS >= originalRMS*0.8 {
+		t.Fatalf("expected low-level noise attenuation, got RMS %.2f from %.2f", processedRMS, originalRMS)
+	}
+}
+
+func TestSuppressorPreservesSpeechLikeSignal(t *testing.T) {
+	suppressor := NewSuppressor()
+	suppressor.SetEnabled(true)
+	suppressor.SetThreshold(0.08)
+
+	voice := makeSineFrame(1000, 9000)
+	originalRMS := frameRMS(voice)
+	processed := append([]int16(nil), voice...)
+	suppressor.ProcessSamples(processed)
+	processedRMS := frameRMS(processed)
+
+	if processedRMS <= originalRMS*0.6 {
+		t.Fatalf("expected speech-like signal to be mostly preserved, got RMS %.2f from %.2f", processedRMS, originalRMS)
+	}
+}
+
+func TestHigherThresholdAppliesStrongerSuppression(t *testing.T) {
+	lowSuppressor := NewSuppressor()
+	lowSuppressor.SetEnabled(true)
+	lowSuppressor.SetThreshold(0.02)
+
+	highSuppressor := NewSuppressor()
+	highSuppressor.SetEnabled(true)
+	highSuppressor.SetThreshold(0.20)
+
+	noiseFrame := makeSineFrame(500, 700)
+	lowRMS := runFrameWarmup(lowSuppressor, noiseFrame, 8)
+	highRMS := runFrameWarmup(highSuppressor, noiseFrame, 8)
+
+	if highRMS >= lowRMS*0.8 {
+		t.Fatalf("expected stronger suppression at higher threshold, got low %.2f high %.2f", lowRMS, highRMS)
+	}
+}
+
+func runFrameWarmup(suppressor *Suppressor, frame []int16, repeats int) float64 {
+	var processed []int16
+	for i := 0; i < repeats; i++ {
+		processed = append([]int16(nil), frame...)
+		suppressor.ProcessSamples(processed)
+	}
+	return frameRMS(processed)
+}
+
+func makeSineFrame(frequency float64, amplitude float64) []int16 {
+	const sampleRate = 48000.0
+	const frameSize = 480
+
+	frame := make([]int16, frameSize)
+	for i := 0; i < frameSize; i++ {
+		value := math.Sin((2.0 * math.Pi * frequency * float64(i)) / sampleRate)
+		frame[i] = int16(value * amplitude)
+	}
+	return frame
+}
+
+func frameRMS(samples []int16) float64 {
+	if len(samples) == 0 {
+		return 0.0
+	}
+	var sumSquares float64
+	for _, sample := range samples {
+		normalized := float64(sample) / 32768.0
+		sumSquares += normalized * normalized
+	}
+	return math.Sqrt(sumSquares / float64(len(samples)))
+}

ui.go

@@ -2,7 +2,7 @@ package main
 
 import (
     "fmt"
-    "github.com/nsf/termbox-go"
+    "os"
     "os/exec"
     "strings"
     "time"
@@ -10,6 +10,7 @@ import (
     "git.stormux.org/storm/barnard/gumble/gumble"
     "git.stormux.org/storm/barnard/uiterm"
     "github.com/kennygrant/sanitize"
+    "github.com/nsf/termbox-go"
 )
 
 const (
@@ -70,7 +71,7 @@ func (b *Barnard) UpdateInputStatus(status string) {
 
 func (b *Barnard) AddOutputLine(line string) {
     now := time.Now()
-    b.UiOutput.AddLine(fmt.Sprintf("[%02d:%02d:%02d] %s", now.Hour(), now.Minute(), now.Second(), line))
+    b.UiOutput.AddLine(fmt.Sprintf("%s [%02d:%02d:%02d]", line, now.Hour(), now.Minute(), now.Second()))
 }
 
 func (b *Barnard) AddOutputMessage(sender *gumble.User, message string) {
@@ -99,7 +100,7 @@ func (b *Barnard) OnNoiseSuppressionToggle(ui *uiterm.Ui, key uiterm.Key) {
     enabled := !b.UserConfig.GetNoiseSuppressionEnabled()
     b.UserConfig.SetNoiseSuppressionEnabled(enabled)
     b.NoiseSuppressor.SetEnabled(enabled)
-    
+
     if enabled {
         b.UpdateGeneralStatus("Noise suppression: ON", false)
     } else {
@@ -107,6 +108,13 @@ func (b *Barnard) OnNoiseSuppressionToggle(ui *uiterm.Ui, key uiterm.Key) {
     }
 }
 
+func (b *Barnard) OnVoiceEffectCycle(ui *uiterm.Ui, key uiterm.Key) {
+    effect := b.VoiceEffects.CycleEffect()
+    b.UserConfig.SetVoiceEffect(int(effect))
+    b.UpdateGeneralStatus(fmt.Sprintf("Voice effect: %s", effect.String()), false)
+}
+
+
 func (b *Barnard) UpdateGeneralStatus(text string, notice bool) {
     if notice {
         b.UiStatus.Fg = uiterm.ColorWhite | uiterm.AttrBold
@@ -143,7 +151,7 @@ func (b *Barnard) CommandNoiseSuppressionToggle(ui *uiterm.Ui, cmd string) {
     enabled := !b.UserConfig.GetNoiseSuppressionEnabled()
     b.UserConfig.SetNoiseSuppressionEnabled(enabled)
     b.NoiseSuppressor.SetEnabled(enabled)
-    
+
     if enabled {
         b.AddOutputLine("Noise suppression enabled")
     } else {
@@ -151,6 +159,102 @@ func (b *Barnard) CommandNoiseSuppressionToggle(ui *uiterm.Ui, cmd string) {
     }
 }
 
+func (b *Barnard) CommandPlayFile(ui *uiterm.Ui, cmd string) {
+    // cmd contains just the filename part (everything after "/file ")
+    filename := strings.TrimSpace(cmd)
+    if filename == "" {
+        b.AddOutputLine("Usage: /file <filename or URL>")
+        return
+    }
+
+    // Check if it's a URL
+    isURL := strings.HasPrefix(filename, "http://") ||
+             strings.HasPrefix(filename, "https://") ||
+             strings.HasPrefix(filename, "ftp://") ||
+             strings.HasPrefix(filename, "rtmp://")
+
+    if !isURL {
+        // Expand ~ to home directory for local files
+        if strings.HasPrefix(filename, "~") {
+            homeDir := os.Getenv("HOME")
+            filename = strings.Replace(filename, "~", homeDir, 1)
+        }
+
+        // Check if local file exists
+        if _, err := os.Stat(filename); os.IsNotExist(err) {
+            b.AddOutputLine(fmt.Sprintf("File not found: %s", filename))
+            return
+        }
+    }
+
+    if !b.Connected {
+        b.AddOutputLine("Not connected to server")
+        return
+    }
+
+    b.FileStreamMutex.Lock()
+    defer b.FileStreamMutex.Unlock()
+
+    if b.FileStream != nil && b.FileStream.IsPlaying() {
+        b.AddOutputLine("Already playing a file. Use /stop first.")
+        return
+    }
+
+    err := b.FileStream.PlayFile(filename)
+    if err != nil {
+        b.AddOutputLine(fmt.Sprintf("Error playing file: %s", err.Error()))
+        return
+    }
+
+    // Enable stereo encoder for file playback
+    b.Client.EnableStereoEncoder()
+
+    // Auto-start transmission if not already transmitting
+    if !b.Tx {
+        err := b.Stream.StartSource(b.UserConfig.GetInputDevice())
+        if err != nil {
+            b.AddOutputLine(fmt.Sprintf("Error starting transmission: %s", err.Error()))
+            b.FileStream.Stop()
+            b.Client.DisableStereoEncoder()
+            return
+        }
+        b.Tx = true
+        b.UpdateGeneralStatus(" File ", true)
+    }
+
+    if isURL {
+        b.AddOutputLine(fmt.Sprintf("Streaming: %s (stereo)", filename))
+    } else {
+        b.AddOutputLine(fmt.Sprintf("Playing: %s (stereo)", filename))
+    }
+}
+
+func (b *Barnard) CommandStopFile(ui *uiterm.Ui, cmd string) {
+    b.FileStreamMutex.Lock()
+    defer b.FileStreamMutex.Unlock()
+
+    if b.FileStream == nil || !b.FileStream.IsPlaying() {
+        b.AddOutputLine("No file playing")
+        return
+    }
+
+    err := b.FileStream.Stop()
+    if err != nil {
+        b.AddOutputLine(fmt.Sprintf("Error stopping file: %s", err.Error()))
+        return
+    }
+
+    // Disable stereo encoder when file stops
+    b.Client.DisableStereoEncoder()
+
+    b.AddOutputLine("File playback stopped")
+
+    // Note: We keep transmission active even after file stops
+    // User can manually stop with talk key or it will stop when they're done talking
+    b.UpdateGeneralStatus(" Idle ", false)
+}
+
+
 func (b *Barnard) setTransmit(ui *uiterm.Ui, val int) {
     if b.Tx && val == 1 {
         return
@@ -249,6 +353,43 @@ func (b *Barnard) OnTextInput(ui *uiterm.Ui, textbox *uiterm.Textbox, text strin
     if text == "" {
         return
     }
+
+    // Check if this is a command (starts with /)
+    if strings.HasPrefix(text, "/") {
+        // Remove the leading slash and process as command
+        cmdText := strings.TrimPrefix(text, "/")
+        parts := strings.SplitN(cmdText, " ", 2)
+        cmdName := parts[0]
+        cmdArgs := ""
+        if len(parts) > 1 {
+            cmdArgs = parts[1]
+        }
+
+        // Handle built-in commands
+        switch cmdName {
+        case "file":
+            b.CommandPlayFile(ui, cmdArgs)
+        case "stop":
+            b.CommandStopFile(ui, cmdArgs)
+        case "exit":
+            b.CommandExit(ui, cmdArgs)
+        case "status":
+            b.CommandStatus(ui, cmdArgs)
+        case "noise":
+            b.CommandNoiseSuppressionToggle(ui, cmdArgs)
+        case "micup":
+            b.CommandMicUp(ui, cmdArgs)
+        case "micdown":
+            b.CommandMicDown(ui, cmdArgs)
+        case "toggle", "talk":
+            b.CommandTalk(ui, cmdArgs)
+        default:
+            b.AddOutputLine(fmt.Sprintf("Unknown command: /%s", cmdName))
+        }
+        return
+    }
+
+    // Not a command, send as chat message
     if b.Client != nil && b.Client.Self != nil {
         if b.selectedUser != nil {
             b.selectedUser.Send(text)
@@ -317,10 +458,13 @@ func (b *Barnard) OnUiInitialize(ui *uiterm.Ui) {
     b.Ui.AddCommandListener(b.CommandExit, "exit")
     b.Ui.AddCommandListener(b.CommandStatus, "status")
     b.Ui.AddCommandListener(b.CommandNoiseSuppressionToggle, "noise")
+    b.Ui.AddCommandListener(b.CommandPlayFile, "file")
+    b.Ui.AddCommandListener(b.CommandStopFile, "stop")
     b.Ui.AddKeyListener(b.OnFocusPress, b.Hotkeys.SwitchViews)
     b.Ui.AddKeyListener(b.OnVoiceToggle, b.Hotkeys.Talk)
     b.Ui.AddKeyListener(b.OnTimestampToggle, b.Hotkeys.ToggleTimestamps)
     b.Ui.AddKeyListener(b.OnNoiseSuppressionToggle, b.Hotkeys.NoiseSuppressionToggle)
+    b.Ui.AddKeyListener(b.OnVoiceEffectCycle, b.Hotkeys.CycleVoiceEffect)
     b.Ui.AddKeyListener(b.OnQuitPress, b.Hotkeys.Exit)
     b.Ui.AddKeyListener(b.OnScrollOutputUp, b.Hotkeys.ScrollUp)
     b.Ui.AddKeyListener(b.OnScrollOutputDown, b.Hotkeys.ScrollDown)

ui_tree.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+    "fmt"
     "git.stormux.org/storm/barnard/gumble/gumble"
     "git.stormux.org/storm/barnard/uiterm"
     "sort"
@@ -11,7 +12,10 @@ func (ti TreeItem) String() string {
         if ti.User.LocallyMuted {
             return "[MUTED] " + ti.User.Name
         }
-        return ti.User.Name
+        // Calculate total volume as percentage
+        boostPercent := float32(ti.User.Boost-1) * 10
+        totalVolume := ti.User.Volume*100 + boostPercent
+        return fmt.Sprintf("%s [%.0f%%]", ti.User.Name, totalVolume)
     }
     if ti.Channel != nil {
         return "#" + ti.Channel.Name

uiterm/key_toml.go

@@ -0,0 +1,15 @@
+package uiterm
+
+// MarshalText implements the encoding.TextMarshaler interface for Key
+// This allows TOML to serialize Key values as strings
+func (i Key) MarshalText() ([]byte, error) {
+	return []byte(i.String()), nil
+}
+
+// UnmarshalText implements the encoding.TextUnmarshaler interface for Key
+// This allows TOML to deserialize strings back into Key values
+func (i *Key) UnmarshalText(data []byte) error {
+	var err error
+	*i, err = KeyString(string(data))
+	return err
+}