claude-voice 1.3.6 → 1.3.7

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-voice",
-  "version": "1.3.6",
+  "version": "1.3.7",
   "description": "Voice interface extension for Claude Code - TTS, STT, and wake word detection",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

package/scripts/postinstall.js

@@ -23,6 +23,8 @@ const CONFIG_DIR = path.join(os.homedir(), '.claude-voice');
 const CONFIG_FILE = path.join(CONFIG_DIR, 'config.json');
 const DEFAULT_CONFIG = path.join(__dirname, '..', 'config', 'default.json');
 const HOOKS_DIR = path.join(__dirname, '..', 'hooks');
+const MODELS_DIR = path.join(CONFIG_DIR, 'models');
+const VOICES_DIR = path.join(CONFIG_DIR, 'voices');
 
 console.log('\n╔════════════════════════════════════════════════════════════╗');
 console.log('║          Claude Voice Extension - Auto Setup               ║');
@@ -89,16 +91,94 @@ try {
 }
 
 // 6. Install Piper TTS and download default voice
-const binPath = path.join(__dirname, '..', 'bin', 'claude-voice');
 console.log('\nStep 4/7: Installing Piper TTS engine...');
 console.log('  (First-time install may take 1-2 minutes)\n');
 try {
-  execSync(`"${binPath}" voice download en_US-joe-medium`, {
-    stdio: 'inherit',
-    timeout: 300000  // 5 min timeout
-  });
+  // Direct download without going through CLI
+  const voiceId = 'en_US-joe-medium';
+  const onnxPath = path.join(VOICES_DIR, `${voiceId}.onnx`);
+  const jsonPath = path.join(VOICES_DIR, `${voiceId}.onnx.json`);
+
+  if (fs.existsSync(onnxPath) && fs.existsSync(jsonPath)) {
+    console.log(`  [✓] Voice already installed: ${voiceId}`);
+  } else {
+    // Create voices directory
+    if (!fs.existsSync(VOICES_DIR)) {
+      fs.mkdirSync(VOICES_DIR, { recursive: true });
+    }
+
+    // Get voice URLs from HuggingFace
+    const parts = voiceId.split('-');
+    const langCode = parts[0];
+    const lang = langCode.split('_')[0];
+    const voiceName = parts[1];
+    const quality = parts[2];
+    const baseUrl = 'https://huggingface.co/rhasspy/piper-voices/resolve/main';
+    const voicePath = `${lang}/${langCode}/${voiceName}/${quality}`;
+    const onnxUrl = `${baseUrl}/${voicePath}/${voiceId}.onnx`;
+    const jsonUrl = `${baseUrl}/${voicePath}/${voiceId}.onnx.json`;
+
+    console.log(`  Voice: Joe (US English)`);
+    console.log(`  [1/2] Downloading voice model (~50MB)...`);
+    execSync(`curl -L --progress-bar -o "${onnxPath}" "${onnxUrl}"`, { stdio: 'inherit' });
+
+    console.log(`  [2/2] Downloading voice config...`);
+    execSync(`curl -sL -o "${jsonPath}" "${jsonUrl}"`, { stdio: 'inherit' });
+
+    console.log(`  [✓] Voice installed: ${voiceId}`);
+  }
+
+  // Install Piper TTS via pip if needed
+  const PIPER_DIR = path.join(CONFIG_DIR, 'piper');
+  const PIPER_VENV = path.join(PIPER_DIR, 'venv');
+  const PIPER_BIN = path.join(PIPER_VENV, 'bin', 'piper');
+
+  if (!fs.existsSync(PIPER_BIN)) {
+    // Find Python 3.9-3.13
+    const pythonCandidates = [
+      '/opt/homebrew/bin/python3.12', '/opt/homebrew/bin/python3.11', '/opt/homebrew/bin/python3',
+      '/usr/local/bin/python3.12', '/usr/local/bin/python3.11', '/usr/local/bin/python3',
+      '/usr/bin/python3.12', '/usr/bin/python3.11', '/usr/bin/python3.10', '/usr/bin/python3.9', '/usr/bin/python3',
+      'python3.12', 'python3.11', 'python3.10', 'python3'
+    ];
+
+    let python = null;
+    for (const py of pythonCandidates) {
+      try {
+        const version = execSync(`${py} --version 2>&1`, { encoding: 'utf-8' });
+        const match = version.match(/Python 3\.(\d+)/);
+        if (match) {
+          const minor = parseInt(match[1], 10);
+          if (minor >= 9 && minor <= 13) {
+            python = py;
+            break;
+          }
+        }
+      } catch {}
+    }
+
+    if (python) {
+      console.log(`  [1/3] Found Python: ${python}`);
+      if (!fs.existsSync(PIPER_DIR)) {
+        fs.mkdirSync(PIPER_DIR, { recursive: true });
+      }
+      console.log('  [2/3] Creating Python virtual environment...');
+      execSync(`${python} -m venv "${PIPER_VENV}"`, { stdio: 'pipe' });
+      console.log('  [3/3] Installing piper-tts package...');
+      const pip = path.join(PIPER_VENV, 'bin', 'pip');
+      execSync(`"${pip}" install --quiet piper-tts`, { stdio: 'pipe' });
+      console.log('  [✓] Piper TTS installed successfully');
+    } else {
+      const installCmd = os.platform() === 'darwin'
+        ? 'brew install python@3.12'
+        : 'sudo apt install python3.12 python3.12-venv';
+      console.log(`  [!] Python 3.9-3.13 not found. Install with: ${installCmd}`);
+    }
+  } else {
+    console.log('  [✓] Piper TTS already installed');
+  }
 } catch (err) {
-  console.log('  [!] Could not install Piper voice.');
+  console.log('  [!] Could not install Piper voice:', err.message);
   console.log('      Run manually: claude-voice voice download en_US-joe-medium');
 }
 
@@ -106,12 +186,41 @@ try {
 console.log('\nStep 5/7: Downloading Whisper STT model...');
 console.log('  (This may take 2-3 minutes depending on connection)\n');
 try {
-  execSync(`"${binPath}" model download whisper-tiny`, {
-    stdio: 'inherit',
-    timeout: 300000  // 5 min timeout
-  });
+  // Direct download without going through CLI
+  const modelId = 'whisper-tiny';
+  const modelFolder = 'sherpa-onnx-whisper-tiny';
+  const modelUrl = 'https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-tiny.tar.bz2';
+  const modelPath = path.join(MODELS_DIR, modelFolder);
+
+  if (fs.existsSync(modelPath)) {
+    console.log(`  [✓] Model already installed: ${modelId}`);
+  } else {
+    // Create models directory
+    if (!fs.existsSync(MODELS_DIR)) {
+      fs.mkdirSync(MODELS_DIR, { recursive: true });
+    }
+
+    const archivePath = path.join(MODELS_DIR, `${modelId}.tar.bz2`);
+
+    console.log(`  Model: Whisper Tiny (75MB)`);
+    console.log(`  [1/2] Downloading model...`);
+    execSync(`curl -L --progress-bar -o "${archivePath}" "${modelUrl}"`, {
+      stdio: 'inherit',
+      cwd: MODELS_DIR
+    });
+
+    console.log('  [2/2] Extracting model files...');
+    execSync(`tar -xjf "${archivePath}"`, {
+      stdio: 'pipe',
+      cwd: MODELS_DIR
+    });
+
+    // Cleanup archive
+    fs.unlinkSync(archivePath);
+    console.log(`  [✓] Model installed: ${modelId}`);
+  }
 } catch (err) {
-  console.log('  [!] Could not download STT model.');
+  console.log('  [!] Could not download STT model:', err.message);
   console.log('      Run manually: claude-voice model download whisper-tiny');
 }
 
@@ -119,12 +228,41 @@ try {
 console.log('\nStep 6/7: Downloading Sherpa-ONNX Wake Word model...');
 console.log('  (Sherpa-ONNX keyword spotting - ~19MB)\n');
 try {
-  execSync(`"${binPath}" model download kws-zipformer-gigaspeech`, {
-    stdio: 'inherit',
-    timeout: 300000  // 5 min timeout
-  });
+  // Direct download without going through CLI
+  const kwsModelId = 'kws-zipformer-gigaspeech';
+  const kwsModelFolder = 'sherpa-onnx-kws-zipformer-gigaspeech-3.3M-2024-01-01';
+  const kwsModelUrl = 'https://github.com/k2-fsa/sherpa-onnx/releases/download/kws-models/sherpa-onnx-kws-zipformer-gigaspeech-3.3M-2024-01-01.tar.bz2';
+  const kwsModelPath = path.join(MODELS_DIR, kwsModelFolder);
+
+  if (fs.existsSync(kwsModelPath)) {
+    console.log(`  [✓] Model already installed: ${kwsModelId}`);
+  } else {
+    // Create models directory
+    if (!fs.existsSync(MODELS_DIR)) {
+      fs.mkdirSync(MODELS_DIR, { recursive: true });
+    }
+
+    const archivePath = path.join(MODELS_DIR, `${kwsModelId}.tar.bz2`);
+
+    console.log(`  Model: Keyword Spotter English (19MB)`);
+    console.log(`  [1/2] Downloading model...`);
+    execSync(`curl -L --progress-bar -o "${archivePath}" "${kwsModelUrl}"`, {
+      stdio: 'inherit',
+      cwd: MODELS_DIR
+    });
+
+    console.log('  [2/2] Extracting model files...');
+    execSync(`tar -xjf "${archivePath}"`, {
+      stdio: 'pipe',
+      cwd: MODELS_DIR
+    });
+
+    // Cleanup archive
+    fs.unlinkSync(archivePath);
+    console.log(`  [✓] Model installed: ${kwsModelId}`);
+  }
 } catch (err) {
-  console.log('  [!] Could not download wake word model.');
+  console.log('  [!] Could not download wake word model:', err.message);
   console.log('      Run manually: claude-voice model download kws-zipformer-gigaspeech');
 }