voice-mode 3.34.3__py3-none-any.whl → 4.1.0__py3-none-any.whl

voice_mode/__version__.py

@@ -1,3 +1,3 @@
 # This file is automatically updated by 'make release'
 # Do not edit manually
-__version__ = "3.34.3"
+__version__ = "4.1.0"

voice_mode/cli.py

@@ -1359,12 +1359,20 @@ def cli():
 
 # Import subcommand groups
 from voice_mode.cli_commands import exchanges as exchanges_cmd
+from voice_mode.cli_commands import transcribe as transcribe_cmd
+from voice_mode.cli_commands import pronounce_commands
 
 # Add subcommands to legacy CLI
 cli.add_command(exchanges_cmd.exchanges)
+cli.add_command(transcribe_cmd.transcribe)
+cli.add_command(pronounce_commands.pronounce_group)
 
 # Add exchanges to main CLI
 voice_mode_main_cli.add_command(exchanges_cmd.exchanges)
+voice_mode_main_cli.add_command(pronounce_commands.pronounce_group)
+
+# Add transcribe to main CLI
+voice_mode_main_cli.add_command(transcribe_cmd.transcribe)
 
 
 # Converse command - direct voice conversation from CLI

voice_mode/cli_commands/pronounce_commands.py

@@ -0,0 +1,223 @@
+"""CLI commands for managing pronunciation rules."""
+
+import click
+import yaml
+import json
+from pathlib import Path
+from typing import Optional
+
+from voice_mode.pronounce import get_manager
+
+
+@click.group(name='pronounce')
+def pronounce_group():
+    """Manage pronunciation rules for TTS and STT."""
+    pass
+
+
+@pronounce_group.command(name='list')
+@click.option('--direction', '-d', type=click.Choice(['tts', 'stt', 'all']), default='all',
+              help='Filter by direction (tts/stt/all)')
+@click.option('--enabled-only', '-e', is_flag=True, help='Show only enabled rules')
+@click.option('--show-private', '-p', is_flag=True, help='Include private rules')
+@click.option('--format', '-f', type=click.Choice(['table', 'yaml', 'json']), default='table',
+              help='Output format')
+def list_rules(direction: str, enabled_only: bool, show_private: bool, format: str):
+    """List pronunciation rules."""
+    manager = get_manager()
+    
+    # Get rules
+    if direction == 'all':
+        rules = manager.list_rules(include_private=show_private)
+    else:
+        rules = manager.list_rules(direction=direction, include_private=show_private)
+    
+    # Filter if needed
+    if enabled_only:
+        rules = [r for r in rules if r['enabled']]
+    
+    # Format output
+    if format == 'table':
+        if not rules:
+            click.echo("No rules found.")
+            return
+        
+        # Count private rules that were hidden
+        all_rules = manager.list_rules(include_private=True)
+        private_count = len(all_rules) - len(rules)
+        
+        # Simple table format without tabulate
+        click.echo("\nPronunciation Rules:")
+        click.echo("=" * 80)
+        
+        for rule in rules:
+            status = '✓' if rule['enabled'] else '✗'
+            click.echo(f"\n{status} [{rule['direction'].upper()}] {rule['name']} (order: {rule['order']})")
+            click.echo(f"  Pattern:  {rule['pattern']}")
+            click.echo(f"  Replace:  {rule['replacement']}")
+            if rule['description']:
+                click.echo(f"  Desc:     {rule['description']}")
+        
+        if private_count > 0 and not show_private:
+            click.echo(f"\n({private_count} private rules hidden. Use --show-private to display)")
+    
+    elif format == 'yaml':
+        import yaml
+        click.echo(yaml.dump(rules, default_flow_style=False))
+    
+    elif format == 'json':
+        import json
+        click.echo(json.dumps(rules, indent=2))
+
+
+@pronounce_group.command(name='test')
+@click.argument('text')
+@click.option('--direction', '-d', type=click.Choice(['tts', 'stt']), default='tts',
+              help='Test direction (tts/stt)')
+def test_rule(text: str, direction: str):
+    """Test pronunciation rules on text."""
+    manager = get_manager()
+    result = manager.test_rule(text, direction)
+    
+    if text != result:
+        click.echo(f"Original: {text}")
+        click.echo(f"Modified: {result}")
+        
+        # Show which rules were applied if logging is enabled
+        import os
+        if os.environ.get('VOICEMODE_PRONUNCIATION_LOG_SUBSTITUTIONS', '').lower() == 'true':
+            click.echo("\n(Check logs for applied rules)")
+    else:
+        click.echo(f"No changes: {text}")
+
+
+@pronounce_group.command(name='add')
+@click.option('--direction', '-d', type=click.Choice(['tts', 'stt']), required=True,
+              help='Rule direction (tts/stt)')
+@click.option('--pattern', '-p', required=True, help='Regex pattern to match')
+@click.option('--replacement', '-r', required=True, help='Replacement text')
+@click.option('--name', '-n', help='Rule name (auto-generated if not provided)')
+@click.option('--description', help='Rule description')
+@click.option('--order', type=int, default=100, help='Processing order (lower = earlier)')
+@click.option('--disabled', is_flag=True, help='Create rule as disabled')
+def add_rule(direction: str, pattern: str, replacement: str, name: Optional[str],
+             description: str, order: int, disabled: bool):
+    """Add a new pronunciation rule."""
+    manager = get_manager()
+    
+    success = manager.add_rule(
+        direction=direction,
+        pattern=pattern,
+        replacement=replacement,
+        name=name,
+        description=description or "",
+        enabled=not disabled,
+        order=order,
+        private=False  # CLI-created rules are not private
+    )
+    
+    if success:
+        click.echo(f"✓ Rule added successfully")
+    else:
+        click.echo("✗ Failed to add rule (check pattern validity)", err=True)
+
+
+@pronounce_group.command(name='remove')
+@click.option('--direction', '-d', type=click.Choice(['tts', 'stt']), required=True,
+              help='Rule direction (tts/stt)')
+@click.argument('name')
+def remove_rule(direction: str, name: str):
+    """Remove a pronunciation rule by name."""
+    manager = get_manager()
+    
+    success = manager.remove_rule(direction, name)
+    
+    if success:
+        click.echo(f"✓ Rule '{name}' removed")
+    else:
+        click.echo(f"✗ Rule '{name}' not found", err=True)
+
+
+@pronounce_group.command(name='enable')
+@click.option('--direction', '-d', type=click.Choice(['tts', 'stt']), required=True,
+              help='Rule direction (tts/stt)')
+@click.argument('name')
+def enable_rule(direction: str, name: str):
+    """Enable a pronunciation rule."""
+    manager = get_manager()
+    
+    success = manager.enable_rule(direction, name)
+    
+    if success:
+        click.echo(f"✓ Rule '{name}' enabled")
+    else:
+        click.echo(f"✗ Failed to enable rule '{name}' (not found or private)", err=True)
+
+
+@pronounce_group.command(name='disable')
+@click.option('--direction', '-d', type=click.Choice(['tts', 'stt']), required=True,
+              help='Rule direction (tts/stt)')
+@click.argument('name')
+def disable_rule(direction: str, name: str):
+    """Disable a pronunciation rule."""
+    manager = get_manager()
+    
+    success = manager.disable_rule(direction, name)
+    
+    if success:
+        click.echo(f"✓ Rule '{name}' disabled")
+    else:
+        click.echo(f"✗ Failed to disable rule '{name}' (not found or private)", err=True)
+
+
+@pronounce_group.command(name='reload')
+def reload_rules():
+    """Reload pronunciation rules from configuration files."""
+    manager = get_manager()
+    manager.reload_rules()
+    click.echo("✓ Pronunciation rules reloaded")
+
+
+@pronounce_group.command(name='edit')
+@click.option('--system', is_flag=True, help='Edit system default rules (requires sudo)')
+def edit_config(system: bool):
+    """Open pronunciation config in editor."""
+    import os
+    import subprocess
+    
+    if system:
+        # Edit system defaults
+        config_path = Path(__file__).parent.parent / 'data' / 'default_pronunciation.yaml'
+        if not config_path.exists():
+            click.echo(f"System config not found: {config_path}", err=True)
+            return
+        # Might need sudo
+        editor = os.environ.get('EDITOR', 'nano')
+        subprocess.run(['sudo', editor, str(config_path)])
+    else:
+        # Edit user config
+        config_path = Path.home() / '.voicemode' / 'config' / 'pronunciation.yaml'
+        if not config_path.exists():
+            # Create default config
+            config_path.parent.mkdir(parents=True, exist_ok=True)
+            default_config = {
+                'version': 1,
+                'tts_rules': [],
+                'stt_rules': []
+            }
+            with open(config_path, 'w') as f:
+                yaml.dump(default_config, f, default_flow_style=False)
+        
+        editor = os.environ.get('EDITOR', 'nano')
+        subprocess.run([editor, str(config_path)])
+    
+    # Reload after editing
+    manager = get_manager()
+    manager.reload_rules()
+    click.echo("✓ Configuration edited and reloaded")
+
+
+# Register the command group
+def register_commands(cli):
+    """Register pronunciation commands with the main CLI."""
+    cli.add_command(pronounce_group)

voice_mode/cli_commands/transcribe.py

@@ -0,0 +1,141 @@
+"""CLI command for audio transcription."""
+
+import click
+import json
+import asyncio
+from pathlib import Path
+from typing import Optional
+
+from voice_mode.tools.transcription import (
+    transcribe_audio,
+    TranscriptionBackend,
+    OutputFormat
+)
+
+
+@click.group()
+def transcribe():
+    """Audio transcription with word-level timestamps."""
+    pass
+
+
+@transcribe.command("audio")
+@click.argument('audio_file', type=click.Path(exists=True))
+@click.option('--words', is_flag=True, help='Include word-level timestamps')
+@click.option(
+    '--backend',
+    type=click.Choice(['openai', 'whisperx', 'whisper-cpp']),
+    default='openai',
+    help='Transcription backend to use'
+)
+@click.option(
+    '--format',
+    'output_format',
+    type=click.Choice(['json', 'srt', 'vtt', 'csv']),
+    default='json',
+    help='Output format for transcription'
+)
+@click.option('--output', '-o', type=click.Path(), help='Save transcription to file')
+@click.option('--language', help='Language code (e.g., en, es, fr)')
+@click.option('--model', default='whisper-1', help='Model to use (for OpenAI backend)')
+def audio_command(
+    audio_file: str,
+    words: bool,
+    backend: str,
+    output_format: str,
+    output: Optional[str],
+    language: Optional[str],
+    model: str
+):
+    """
+    Transcribe audio with optional word-level timestamps.
+    
+    Examples:
+    
+        voice-mode transcribe audio recording.mp3
+        
+        voice-mode transcribe audio interview.wav --words
+        
+        voice-mode transcribe audio podcast.mp3 --words --format srt -o subtitles.srt
+        
+        voice-mode transcribe audio spanish.mp3 --language es --backend whisperx
+    """
+    async def run():
+        # Perform transcription
+        result = await transcribe_audio(
+            audio_file=audio_file,
+            word_timestamps=words,
+            backend=TranscriptionBackend(backend),
+            output_format=OutputFormat(output_format),
+            language=language,
+            model=model
+        )
+        
+        # Check for errors
+        if not result.get("success", False):
+            error_msg = result.get("error", "Unknown error occurred")
+            click.echo(f"Error: {error_msg}", err=True)
+            return
+        
+        # Format output
+        if output_format == 'json':
+            # Remove internal fields for cleaner output
+            output_result = {k: v for k, v in result.items() 
+                           if k not in ['formatted_content']}
+            content = json.dumps(output_result, indent=2)
+        elif "formatted_content" in result:
+            content = result["formatted_content"]
+        else:
+            # Fallback to JSON if format conversion failed
+            content = json.dumps(result, indent=2)
+        
+        # Write output
+        if output:
+            Path(output).write_text(content)
+            click.echo(f"Transcription saved to {output}")
+        else:
+            click.echo(content)
+    
+    # Run async function
+    asyncio.run(run())
+
+
+# For backward compatibility, also provide a direct command
+@click.command('transcribe-audio')
+@click.argument('audio_file', type=click.Path(exists=True))
+@click.option('--words', is_flag=True, help='Include word-level timestamps')
+@click.option(
+    '--backend',
+    type=click.Choice(['openai', 'whisperx', 'whisper-cpp']),
+    default='openai',
+    help='Transcription backend'
+)
+@click.option(
+    '--format',
+    'output_format',
+    type=click.Choice(['json', 'srt', 'vtt', 'csv']),
+    default='json',
+    help='Output format'
+)
+@click.option('--output', '-o', type=click.Path(), help='Save to file')
+@click.option('--language', help='Language code')
+@click.option('--model', default='whisper-1', help='Model to use')
+def transcribe_audio_command(
+    audio_file: str,
+    words: bool,
+    backend: str,
+    output_format: str,
+    output: Optional[str],
+    language: Optional[str],
+    model: str
+):
+    """Direct transcription command for backward compatibility."""
+    audio_command.callback(
+        audio_file=audio_file,
+        words=words,
+        backend=backend,
+        output_format=output_format,
+        output=output,
+        language=language,
+        model=model
+    )

voice_mode/config.py

@@ -15,21 +15,66 @@ from datetime import datetime
 
 # ==================== ENVIRONMENT CONFIGURATION ====================
 
+def find_voicemode_env_files() -> list[Path]:
+    """
+    Find .voicemode.env files by walking up the directory tree.
+    
+    Looks for (in order of priority - closest to current directory wins):
+    1. .voicemode.env in current or parent directories  
+    2. .voicemode/voicemode.env in current or parent directories
+    3. ~/.voicemode/voicemode.env in user home (global config)
+    
+    Returns:
+        List of Path objects in loading order (global first, then project-specific)
+    """
+    config_files = []
+    
+    # First add global config (lowest priority - loaded first)
+    global_config = Path.home() / ".voicemode" / "voicemode.env"
+    
+    # Backwards compatibility: check for old filename
+    if not global_config.exists():
+        old_global = Path.home() / ".voicemode" / ".voicemode.env"
+        if old_global.exists():
+            global_config = old_global
+    
+    if global_config.exists():
+        config_files.append(global_config)
+    
+    # Then walk up directory tree for project-specific configs (higher priority)
+    current_dir = Path.cwd()
+    project_configs = []
+    
+    while current_dir != current_dir.parent:
+        # Check for standalone .voicemode.env first
+        standalone_file = current_dir / ".voicemode.env"
+        if standalone_file.exists():
+            project_configs.append(standalone_file)
+            break  # Stop at first found (closest wins)
+            
+        # Then check .voicemode/voicemode.env  
+        dir_file = current_dir / ".voicemode" / "voicemode.env"
+        # Skip if this is the global config file (already added)
+        if dir_file.exists() and dir_file != global_config:
+            project_configs.append(dir_file)
+            break  # Stop at first found (closest wins)
+        
+        current_dir = current_dir.parent
+    
+    # Add project configs (they were collected closest-first, so add as-is)
+    config_files.extend(project_configs)
+    
+    return config_files
+
+
 def load_voicemode_env():
-    """Load configuration from voicemode.env file if it exists, creating a default if not."""
-    # Try new filename first
-    config_path = Path.home() / ".voicemode" / "voicemode.env"
-    
-    # Backwards compatibility: check for old filename if new doesn't exist
-    if not config_path.exists():
-        old_path = Path.home() / ".voicemode" / ".voicemode.env"
-        if old_path.exists():
-            config_path = old_path
-            print(f"Warning: Using deprecated .voicemode.env - please rename to voicemode.env")
-    
-    if not config_path.exists():
-        # Create default template
-        config_path.parent.mkdir(parents=True, exist_ok=True)
+    """Load configuration from voicemode.env files, with cascading from global to project-specific."""
+    config_files = find_voicemode_env_files()
+    
+    # If no config files found, create default global config
+    if not config_files:
+        default_path = Path.home() / ".voicemode" / "voicemode.env"
+        default_path.parent.mkdir(parents=True, exist_ok=True)
         default_config = '''# Voice Mode Configuration File
 # This file is automatically generated and can be customized
 # Environment variables always take precedence over this file
@@ -66,8 +111,8 @@ def load_voicemode_env():
 # Comma-separated list of STT endpoints
 # VOICEMODE_STT_BASE_URLS=http://127.0.0.1:2022/v1,https://api.openai.com/v1
 
-# Comma-separated list of preferred voices
-# VOICEMODE_TTS_VOICES=af_sky,alloy
+# Comma-separated list of preferred voices  
+# VOICEMODE_VOICES=af_sky,alloy
 
 # Comma-separated list of preferred models
 # VOICEMODE_TTS_MODELS=tts-1,tts-1-hd,gpt-4o-mini-tts
@@ -127,26 +172,28 @@ def load_voicemode_env():
 # LIVEKIT_API_KEY=devkey
 # LIVEKIT_API_SECRET=secret
 '''
-        with open(config_path, 'w') as f:
+        with open(default_path, 'w') as f:
             f.write(default_config)
-        os.chmod(config_path, 0o600)  # Secure permissions
-    
-    # Load configuration from file
-    if config_path.exists():
-        with open(config_path, 'r') as f:
-            for line in f:
-                line = line.strip()
-                # Skip comments and empty lines
-                if not line or line.startswith('#'):
-                    continue
-                # Parse KEY=VALUE format
-                if '=' in line:
-                    key, value = line.split('=', 1)
-                    key = key.strip()
-                    value = value.strip()
-                    # Only set if not already in environment
-                    if key and key not in os.environ:
-                        os.environ[key] = value
+        os.chmod(default_path, 0o600)  # Secure permissions
+        config_files = [default_path]
+    
+    # Load configuration from all files in order (global first, project-specific last)
+    for config_path in config_files:
+        if config_path.exists():
+            with open(config_path, 'r') as f:
+                for line in f:
+                    line = line.strip()
+                    # Skip comments and empty lines
+                    if not line or line.startswith('#'):
+                        continue
+                    # Parse KEY=VALUE format
+                    if '=' in line:
+                        key, value = line.split('=', 1)
+                        key = key.strip()
+                        value = value.strip()
+                        # Only set if not already in environment (env vars take precedence)
+                        if key and key not in os.environ:
+                            os.environ[key] = value
 
 # Load configuration file before other configuration
 load_voicemode_env()
@@ -222,13 +269,68 @@ def parse_comma_list(env_var: str, fallback: str) -> list:
 # New provider endpoint lists configuration
 TTS_BASE_URLS = parse_comma_list("VOICEMODE_TTS_BASE_URLS", "http://127.0.0.1:8880/v1,https://api.openai.com/v1")
 STT_BASE_URLS = parse_comma_list("VOICEMODE_STT_BASE_URLS", "http://127.0.0.1:2022/v1,https://api.openai.com/v1")
-TTS_VOICES = parse_comma_list("VOICEMODE_TTS_VOICES", "af_sky,alloy")
+TTS_VOICES = parse_comma_list("VOICEMODE_VOICES", "af_sky,alloy")
 TTS_MODELS = parse_comma_list("VOICEMODE_TTS_MODELS", "tts-1,tts-1-hd,gpt-4o-mini-tts")
 
+# Voice preferences cache
+_cached_voice_preferences: Optional[list] = None
+_voice_preferences_loaded = False
+
+def get_voice_preferences() -> list[str]:
+    """
+    Get voice preferences from configuration.
+    
+    Uses the VOICEMODE_VOICES configuration which is loaded from:
+    1. Environment variables (highest priority)
+    2. Project-specific .voicemode.env files 
+    3. Global ~/.voicemode/voicemode.env file
+    4. Built-in defaults
+    
+    Returns:
+        List of voice names in preference order
+    """
+    global _cached_voice_preferences, _voice_preferences_loaded
+    
+    # Return cached preferences if already loaded
+    if _voice_preferences_loaded:
+        return _cached_voice_preferences or []
+    
+    _voice_preferences_loaded = True
+    
+    # Get voices from TTS_VOICES configuration
+    _cached_voice_preferences = TTS_VOICES.copy()
+    
+    logger.info(f"Voice preferences loaded: {_cached_voice_preferences}")
+    return _cached_voice_preferences
+
+def clear_voice_preferences_cache():
+    """Clear the voice preferences cache, forcing a reload on next access."""
+    global _cached_voice_preferences, _voice_preferences_loaded
+    _cached_voice_preferences = None
+    _voice_preferences_loaded = False
+    logger.debug("Voice preferences cache cleared")
+
+def reload_configuration():
+    """Reload configuration from files and clear all caches."""
+    # Clear voice preferences cache
+    clear_voice_preferences_cache()
+    
+    # Reload environment configuration
+    load_voicemode_env()
+    
+    # Update global configuration variables
+    global TTS_VOICES, TTS_MODELS, TTS_BASE_URLS, STT_BASE_URLS
+    TTS_BASE_URLS = parse_comma_list("VOICEMODE_TTS_BASE_URLS", "http://127.0.0.1:8880/v1,https://api.openai.com/v1")
+    STT_BASE_URLS = parse_comma_list("VOICEMODE_STT_BASE_URLS", "http://127.0.0.1:2022/v1,https://api.openai.com/v1")
+    TTS_VOICES = parse_comma_list("VOICEMODE_VOICES", "af_sky,alloy")
+    TTS_MODELS = parse_comma_list("VOICEMODE_TTS_MODELS", "tts-1,tts-1-hd,gpt-4o-mini-tts")
+    
+    logger.info("Configuration reloaded successfully")
+
 # Legacy variables have been removed - use the new list-based configuration:
 # - VOICEMODE_TTS_BASE_URLS (comma-separated list)
 # - VOICEMODE_STT_BASE_URLS (comma-separated list)
-# - VOICEMODE_TTS_VOICES (comma-separated list)
+# - VOICEMODE_VOICES (comma-separated list)
 # - VOICEMODE_TTS_MODELS (comma-separated list)
 
 # LiveKit configuration