sona-ai-voice 0.2.3 → 0.2.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "sona-ai-voice",
-  "version": "0.2.3",
+  "version": "0.2.4",
   "description": "Voice-to-voice AI CLI companion powered by OpenAI Realtime API",
   "author": "zerish",
   "license": "MIT",
@@ -21,8 +21,8 @@
     "sona": "packages/cli/dist/index.js"
   },
   "files": [
+    "README.md",
     "packages/*/dist/**/*",
-    "packages/*/package.json",
     "packages/cli/src/gui/**/*",
     "bin/**/*"
   ],
@@ -38,6 +38,8 @@
     "lint": "eslint packages --ext .ts,.tsx",
     "format": "prettier --write \"packages/**/*.{ts,tsx,json,md}\"",
     "cache:clear": "rm -rf node_modules/.cache .sona-cache",
+    "prepack": "node scripts/prepack.js",
+    "postpack": "node scripts/postpack.js",
     "prepublishOnly": "npm run build"
   },
   "dependencies": {

package/packages/cli/dist/audio-capture.js

@@ -1,8 +1,5 @@
 "use strict";
-/**
- * Audio Capture Module
- * Mic capture with pause/resume for echo cancellation
- */
+
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.AudioCapture = void 0;
 const child_process_1 = require("child_process");
@@ -30,14 +27,14 @@ class AudioCapture extends events_1.EventEmitter {
             this.micProcess = null;
         }
         this.micProcess = (0, child_process_1.spawn)('sox', [
-            '-d', // Default audio device
-            '-t', 'raw', // Raw PCM output
-            '-b', '16', // 16-bit
-            '-e', 'signed-integer', // Signed integer
-            '-r', '24000', // 24kHz (required by OpenAI)
-            '-c', '1', // Mono
-            '-q', // Quiet mode
-            '-', // Output to stdout
+            '-d', 
+            '-t', 'raw', 
+            '-b', '16', 
+            '-e', 'signed-integer', 
+            '-r', '24000', 
+            '-c', '1', 
+            '-q', 
+            '-', 
         ]);
         if (!this.micProcess.stdout) {
             this.emit('error', new Error('Failed to start microphone'));
@@ -48,12 +45,12 @@ class AudioCapture extends events_1.EventEmitter {
                 this.emit('data', chunk);
             }
         });
-        // Capture stderr to detect ALSA/audio device errors
+        
         let stderrOutput = '';
         if (this.micProcess.stderr) {
             this.micProcess.stderr.on('data', (chunk) => {
                 stderrOutput += chunk.toString();
-                // Check for ALSA/audio device errors
+                
                 if (stderrOutput.includes('ALSA') ||
                     stderrOutput.includes('no default audio device') ||
                     stderrOutput.includes('Unknown PCM') ||
@@ -64,9 +61,9 @@ class AudioCapture extends events_1.EventEmitter {
             });
         }
         this.micProcess.on('error', (error) => {
-            // ENOENT means command not found - don't spam errors
+            
             if (error.code === 'ENOENT') {
-                // Error already handled at startup, just emit once
+                
                 if (!this.isRecording) {
                     this.emit('error', new Error('Audio capture tool (sox) not found. Please install sox.'));
                 }
@@ -81,15 +78,12 @@ class AudioCapture extends events_1.EventEmitter {
         this.isPaused = false;
         this.emit('started');
     }
-    /**
-     * Pause audio capture (for echo cancellation during playback)
-     * Kills the mic process to ensure no audio leaks through
-     */
+    
     pause() {
         if (!this.isRecording || this.isPaused)
             return;
         this.isPaused = true;
-        // Kill the mic process to ensure complete silence
+        
         if (this.micProcess) {
             try {
                 this.micProcess.kill('SIGKILL');
@@ -98,14 +92,12 @@ class AudioCapture extends events_1.EventEmitter {
             this.micProcess = null;
         }
     }
-    /**
-     * Resume audio capture after playback ends
-     */
+    
     resume() {
         if (!this.isPaused)
             return;
         this.isPaused = false;
-        // Restart the mic process
+        
         this.spawnMic();
     }
     stop() {
@@ -136,4 +128,3 @@ class AudioCapture extends events_1.EventEmitter {
     }
 }
 exports.AudioCapture = AudioCapture;
-//# sourceMappingURL=audio-capture.js.map

package/packages/cli/dist/audio-playback.js

@@ -1,8 +1,5 @@
 "use strict";
-/**
- * Audio Playback Module
- * Smooth speaker output using sox with buffering to prevent audio cracks
- */
+
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.AudioPlayback = void 0;
 const child_process_1 = require("child_process");
@@ -11,7 +8,7 @@ class AudioPlayback extends events_1.EventEmitter {
     playProcess = null;
     isWritable = false;
     config;
-    // Buffering for smooth playback
+    
     audioBuffer = [];
     isPlaying = false;
     totalBytesQueued = 0;
@@ -26,7 +23,7 @@ class AudioPlayback extends events_1.EventEmitter {
     }
     spawnProcess() {
         if (this.playProcess) {
-            return; // Already have a process
+            return; 
         }
         this.playProcess = (0, child_process_1.spawn)('play', [
             '-t', 'raw',
@@ -35,26 +32,26 @@ class AudioPlayback extends events_1.EventEmitter {
             '-r', String(this.config.sampleRate),
             '-c', '1',
             '-q',
-            '-', // Read from stdin
+            '-', 
         ], {
-            stdio: ['pipe', 'ignore', 'ignore'] // Only need stdin
+            stdio: ['pipe', 'ignore', 'ignore'] 
         });
         if (!this.playProcess.stdin) {
             this.emit('error', new Error('Failed to start speaker'));
             return;
         }
-        // Handle stdin errors gracefully
+        
         this.playProcess.stdin.on('error', (error) => {
             if (error.code !== 'EPIPE' && error.code !== 'ERR_STREAM_DESTROYED') {
                 this.emit('error', error);
             }
         });
-        // Capture stderr to detect ALSA/audio device errors
+        
         let stderrOutput = '';
         if (this.playProcess.stderr) {
             this.playProcess.stderr.on('data', (chunk) => {
                 stderrOutput += chunk.toString();
-                // Check for ALSA/audio device errors
+                
                 if (stderrOutput.includes('ALSA') || stderrOutput.includes('no default audio device') || stderrOutput.includes('Unknown PCM')) {
                     this.emit('error', new Error('Audio device not available. WSL does not support direct audio access.'));
                     this.playProcess?.kill();
@@ -62,9 +59,9 @@ class AudioPlayback extends events_1.EventEmitter {
             });
         }
         this.playProcess.on('error', (error) => {
-            // ENOENT means command not found - don't spam errors
+            
             if (error.code === 'ENOENT') {
-                // Error already handled at startup, just emit once
+                
                 if (!this.isPlaying) {
                     this.emit('error', new Error('Audio playback tool (play) not found. Please install sox.'));
                 }
@@ -74,20 +71,20 @@ class AudioPlayback extends events_1.EventEmitter {
         });
         this.playProcess.on('exit', (code) => {
             if (code !== 0 && code !== null && stderrOutput.includes('ALSA')) {
-                // ALSA error - audio device not available
+                
                 this.emit('error', new Error('Audio device not available. WSL does not support direct audio access.'));
             }
         });
         this.playProcess.on('close', () => {
-            // If we requested finish, don't respawn here - finishPlayback will handle it
-            // Otherwise, respawn for next response
+            
+            
             if (!this.finishRequested) {
                 this.playProcess = null;
                 this.isWritable = false;
                 this.spawnProcess();
             }
             else {
-                // Process closed after finish was requested - this is expected
+                
                 this.playProcess = null;
                 this.isWritable = false;
             }
@@ -95,16 +92,14 @@ class AudioPlayback extends events_1.EventEmitter {
         this.isWritable = true;
         this.emit('started');
     }
-    /**
-     * Queue audio for playback
-     */
+    
     play(audioBuffer) {
         if (this.finishRequested) {
-            return; // Ignoring audio after finish requested
+            return; 
         }
         this.totalBytesQueued += audioBuffer.length;
-        // Write directly to the play process - no drain listener needed
-        // The Node.js stream handles buffering internally
+        
+        
         if (this.playProcess?.stdin && this.isWritable) {
             try {
                 this.playProcess.stdin.write(audioBuffer);
@@ -115,10 +110,7 @@ class AudioPlayback extends events_1.EventEmitter {
         }
         this.isPlaying = true;
     }
-    /**
-     * Signal that we're done sending audio for this response.
-     * Closes stdin to signal end of stream, then waits for all audio to play.
-     */
+    
     finishPlayback() {
         if (this.totalBytesQueued === 0) {
             this.resetForNextResponse();
@@ -126,54 +118,50 @@ class AudioPlayback extends events_1.EventEmitter {
             return;
         }
         this.finishRequested = true;
-        // Close stdin to signal no more data is coming
-        // This ensures the play process knows the stream is complete
+        
+        
         if (this.playProcess?.stdin && this.isWritable) {
             try {
                 this.playProcess.stdin.end();
                 this.isWritable = false;
             }
             catch {
-                // Ignore errors if already closed
+                
             }
         }
-        // Calculate playback duration from bytes queued
-        // For 16-bit PCM: bytes / (sampleRate * 2 bytes per sample) * 1000ms
+        
+        
         const durationMs = (this.totalBytesQueued / (this.config.sampleRate * 2)) * 1000;
-        // Add extra buffer to ensure last word is fully played
-        // Account for system buffering and audio pipeline latency
-        const waitTime = durationMs + 200; // Reduced buffer to 200ms for faster response
-        // Wait for audio to finish playing
+        
+        
+        const waitTime = durationMs + 200; 
+        
         this.drainTimeout = setTimeout(() => {
             this.drainTimeout = null;
-            // Respawn process for next response (since we closed stdin)
+            
             this.spawnProcess();
             this.resetForNextResponse();
             this.emit('finished');
         }, waitTime);
     }
-    /**
-     * Reset state for next response
-     */
+    
     resetForNextResponse() {
         this.audioBuffer = [];
         this.totalBytesQueued = 0;
         this.isPlaying = false;
         this.finishRequested = false;
-        // Don't kill the process - keep it alive for next response
-        // Just let it finish playing current audio
+        
+        
     }
-    /**
-     * Immediately stop playback (for barge-in)
-     */
+    
     stop() {
         if (this.drainTimeout) {
             clearTimeout(this.drainTimeout);
             this.drainTimeout = null;
         }
-        // Cancel finish request
+        
         this.finishRequested = false;
-        // Kill current process to stop audio immediately
+        
         if (this.playProcess) {
             try {
                 this.playProcess.kill('SIGKILL');
@@ -185,7 +173,7 @@ class AudioPlayback extends events_1.EventEmitter {
         this.totalBytesQueued = 0;
         this.isPlaying = false;
         this.isWritable = false;
-        // Spawn fresh process for next response
+        
         this.spawnProcess();
         this.emit('stopped');
     }
@@ -200,4 +188,3 @@ class AudioPlayback extends events_1.EventEmitter {
     }
 }
 exports.AudioPlayback = AudioPlayback;
-//# sourceMappingURL=audio-playback.js.map

package/packages/cli/dist/audio-utils.js

@@ -1,21 +1,16 @@
 "use strict";
-/**
- * Cross-platform audio command detection
- * Detects available audio tools on different platforms
- */
+
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.detectAudioCommands = detectAudioCommands;
 exports.getAudioInstallInstructions = getAudioInstallInstructions;
 exports.getAudioError = getAudioError;
 const os_1 = require("os");
 const child_process_1 = require("child_process");
-/**
- * Detect available audio commands for the current platform
- */
+
 function detectAudioCommands() {
     const os = (0, os_1.platform)();
     if (os === 'win32') {
-        // Windows: Check for sox first, then ffmpeg
+        
         try {
             (0, child_process_1.execSync)('sox --version', { stdio: 'ignore' });
             (0, child_process_1.execSync)('play --version', { stdio: 'ignore' });
@@ -27,7 +22,7 @@ function detectAudioCommands() {
             };
         }
         catch {
-            // Try ffmpeg/ffplay as alternative
+            
             try {
                 (0, child_process_1.execSync)('ffmpeg -version', { stdio: 'ignore' });
                 return {
@@ -48,7 +43,7 @@ function detectAudioCommands() {
         }
     }
     else {
-        // Unix-like (Linux, macOS, WSL): sox/play
+        
         try {
             (0, child_process_1.execSync)('sox --version', { stdio: 'ignore' });
             (0, child_process_1.execSync)('play --version', { stdio: 'ignore' });
@@ -69,9 +64,7 @@ function detectAudioCommands() {
         }
     }
 }
-/**
- * Get platform-specific installation instructions
- */
+
 function getAudioInstallInstructions(commands) {
     if (commands.available) {
         return '';
@@ -86,9 +79,7 @@ function getAudioInstallInstructions(commands) {
         return 'Audio tools not found. Install sox: sudo apt-get install sox libsox-fmt-all';
     }
 }
-/**
- * Get audio command error message
- */
+
 function getAudioError(commands) {
     if (commands.available) {
         return '';
@@ -96,4 +87,3 @@ function getAudioError(commands) {
     const instructions = getAudioInstallInstructions(commands);
     return `Audio tools (${commands.record}/${commands.play}) are not available.\n${instructions}`;
 }
-//# sourceMappingURL=audio-utils.js.map

package/packages/cli/dist/echo-canceller.js

@@ -1,12 +1,5 @@
 "use strict";
-/**
- * Echo Canceller Module
- *
- * Uses audio fingerprinting and correlation to detect when the mic
- * is picking up what we're playing through the speakers.
- *
- * This is the "last line of defense" when hardware AEC isn't available.
- */
+
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.EchoCanceller = void 0;
 class EchoCanceller {
@@ -15,26 +8,24 @@ class EchoCanceller {
     isPlayingAudio = false;
     playbackEndTime = 0;
     config;
-    // For energy-based gating
+    
     recentPlaybackEnergy = 0;
     constructor(config = {}) {
         this.config = {
             sampleRate: 24000,
-            bufferDurationMs: 1500, // Keep 1.5s of playback audio
-            correlationThreshold: 0.35, // Correlation > 0.35 = likely echo
-            energyThreshold: 0.01, // Minimum energy to process
+            bufferDurationMs: 1500, 
+            correlationThreshold: 0.35, 
+            energyThreshold: 0.01, 
             ...config,
         };
         const bufferSize = Math.floor(this.config.sampleRate * this.config.bufferDurationMs / 1000);
         this.playbackBuffer = new Float32Array(bufferSize);
     }
-    /**
-     * Call this whenever we send audio to the speakers
-     */
+    
     feedPlayback(audioBuffer) {
         this.isPlayingAudio = true;
         this.playbackEndTime = Date.now();
-        // Convert PCM16 to float samples and add to ring buffer
+        
         const samples = this.pcm16ToFloat(audioBuffer);
         for (let i = 0; i < samples.length; i++) {
             const pos = this.playbackWritePos;
@@ -43,60 +34,53 @@ class EchoCanceller {
             }
             this.playbackWritePos = (this.playbackWritePos + 1) % this.playbackBuffer.length;
         }
-        // Track recent playback energy
+        
         this.recentPlaybackEnergy = this.calculateEnergy(samples);
     }
-    /**
-     * Mark playback as stopped
-     */
+    
     markPlaybackStopped() {
         this.isPlayingAudio = false;
         this.playbackEndTime = Date.now();
     }
-    /**
-     * Check if mic audio is likely echo from our playback
-     * Returns true if we should DROP this audio (it's echo)
-     */
+    
     isEcho(micBuffer) {
-        // If we haven't played anything recently, it's definitely not echo
+        
         const timeSincePlayback = Date.now() - this.playbackEndTime;
         if (timeSincePlayback > 2000) {
             return false;
         }
         const micSamples = this.pcm16ToFloat(micBuffer);
         const micEnergy = this.calculateEnergy(micSamples);
-        // If mic energy is very low, not worth checking
+        
         if (micEnergy < this.config.energyThreshold) {
             return false;
         }
-        // If we're actively playing audio, apply stricter gating
+        
         if (this.isPlayingAudio) {
-            // During active playback, almost anything is likely echo
-            // unless it's significantly louder than what we're playing
+            
+            
             if (micEnergy < this.recentPlaybackEnergy * 3) {
-                return true; // Probably echo
+                return true; 
             }
         }
-        // Cross-correlation check
+        
         const correlation = this.crossCorrelate(micSamples);
         if (correlation > this.config.correlationThreshold) {
-            return true; // High correlation = echo
+            return true; 
         }
         return false;
     }
-    /**
-     * Calculate cross-correlation between mic audio and playback buffer
-     */
+    
     crossCorrelate(micSamples) {
         if (micSamples.length === 0)
             return 0;
         const playbackLen = this.playbackBuffer.length;
         const micLen = micSamples.length;
-        // We'll check a few different lag positions
+        
         let maxCorrelation = 0;
-        // Check lags from 0 to 500ms (12000 samples at 24kHz)
+        
         const maxLag = Math.min(12000, playbackLen - micLen);
-        const lagStep = 100; // Check every ~4ms
+        const lagStep = 100; 
         for (let lag = 0; lag < maxLag; lag += lagStep) {
             let sum = 0;
             let playbackSum = 0;
@@ -111,7 +95,7 @@ class EchoCanceller {
                     micSum += micSample * micSample;
                 }
             }
-            // Normalized correlation
+            
             const denom = Math.sqrt(playbackSum * micSum);
             if (denom > 0) {
                 const correlation = Math.abs(sum / denom);
@@ -120,9 +104,7 @@ class EchoCanceller {
         }
         return maxCorrelation;
     }
-    /**
-     * Calculate RMS energy of samples
-     */
+    
     calculateEnergy(samples) {
         if (samples.length === 0)
             return 0;
@@ -135,9 +117,7 @@ class EchoCanceller {
         }
         return Math.sqrt(sum / samples.length);
     }
-    /**
-     * Convert PCM16 buffer to float samples
-     */
+    
     pcm16ToFloat(buffer) {
         const samples = new Float32Array(buffer.length / 2);
         for (let i = 0; i < samples.length; i++) {
@@ -145,9 +125,7 @@ class EchoCanceller {
         }
         return samples;
     }
-    /**
-     * Clear the playback buffer (e.g., on new session)
-     */
+    
     reset() {
         this.playbackBuffer.fill(0);
         this.playbackWritePos = 0;
@@ -156,4 +134,3 @@ class EchoCanceller {
     }
 }
 exports.EchoCanceller = EchoCanceller;
-//# sourceMappingURL=echo-canceller.js.map

package/packages/cli/dist/gui-server.js

@@ -1,7 +1,5 @@
 "use strict";
-/**
- * GUI Server - Serves particle visualization and broadcasts state updates
- */
+
 var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
@@ -29,7 +27,7 @@ class GUIServer extends events_1.EventEmitter {
     }
     async start() {
         return new Promise((resolve, reject) => {
-            // Create HTTP server to serve the HTML
+            
             this.httpServer = (0, http_1.createServer)((req, res) => {
                 if (req.url === '/' || req.url?.startsWith('/?')) {
                     try {
@@ -48,11 +46,11 @@ class GUIServer extends events_1.EventEmitter {
                     res.end('Not found');
                 }
             });
-            // Create WebSocket server
+            
             this.wss = new ws_1.WebSocketServer({ server: this.httpServer });
             this.wss.on('connection', (ws) => {
                 this.clients.add(ws);
-                // Send current state immediately
+                
                 ws.send(JSON.stringify({ type: 'state', state: this.currentState }));
                 ws.on('close', () => {
                     this.clients.delete(ws);
@@ -69,7 +67,7 @@ class GUIServer extends events_1.EventEmitter {
             });
             this.httpServer.on('error', (err) => {
                 if (err.code === 'EADDRINUSE') {
-                    // Port in use, try next port
+                    
                     this.port++;
                     this.httpServer?.close();
                     this.start().then(resolve).catch(reject);
@@ -93,45 +91,41 @@ class GUIServer extends events_1.EventEmitter {
                 command = 'cmd';
                 args = ['/c', 'start', url];
                 break;
-            default: // Linux/WSL
+            default: 
                 command = 'xdg-open';
                 args = [url];
                 break;
         }
-        // Try to open browser, but don't crash if command doesn't exist
+        
         try {
             const proc = (0, child_process_1.spawn)(command, args, {
                 detached: true,
                 stdio: 'ignore'
             });
-            // Handle errors gracefully
+            
             proc.on('error', (error) => {
-                // ENOENT means command not found - this is expected in some environments (WSL, headless servers)
+                
                 if (error.code === 'ENOENT') {
-                    // Browser opening command not available - show helpful message
+                    
                     console.log(chalk_1.default.gray(`\nGUI available at: http://localhost:${this.port}`));
                     console.log(chalk_1.default.gray('(Browser not auto-opened - open manually if needed)'));
                 }
-                // Other errors are unexpected but shouldn't crash the app
+                
             });
-            // Don't wait for process to exit
+            
             proc.unref();
         }
         catch (error) {
-            // If spawn itself fails, just continue
-            // Browser opening is optional functionality
+            
+            
         }
     }
-    /**
-     * Broadcast state change to all connected clients
-     */
+    
     setState(state) {
         this.currentState = state;
         this.broadcast({ type: 'state', state });
     }
-    /**
-     * Broadcast audio level to all connected clients
-     */
+    
     setAudioLevel(level) {
         this.broadcast({ type: 'audio', level: Math.min(1, Math.max(0, level)) });
     }
@@ -143,7 +137,7 @@ class GUIServer extends events_1.EventEmitter {
                     client.send(message);
                 }
                 catch {
-                    // Ignore send errors
+                    
                 }
             }
         }
@@ -169,4 +163,3 @@ class GUIServer extends events_1.EventEmitter {
     }
 }
 exports.GUIServer = GUIServer;
-//# sourceMappingURL=gui-server.js.map