Voice-to-text dictation app with local (Nvidia Parakeet/Whisper) and cloud models (BYOK). Privacy-first and available cross-platform.
JavaScript aianthropiccross-platformgeminigroqlinuxmacosnvidiaopen-sourceopenaiparakeetspeech-to-texttranscribewhisperwindows
chore: python formatting
Gabriel Stein0c690e4a87a2
diff --git a/package.json b/package.jsonindex af67ee389..5dfeed268 100644--- a/package.json+++ b/package.json@@ -29,9 +29,9 @@"predist": "npm run compile:globe","dist": "npm run build:renderer && electron-builder","format:js": "eslint . --fix && cd src && eslint . --fix && prettier --write \"**/*.{js,jsx,ts,tsx,json,css,md}\"",- "format:py": "python -m ruff check --fix *.py && python -m ruff format *.py",+ "format:py": "python3 -m ruff check --fix *.py && python3 -m ruff format *.py","format": "npm run format:js && npm run format:py",- "format:check": "eslint . && cd src && eslint . && prettier --check \"**/*.{js,jsx,ts,tsx,json,css,md}\" && python -m ruff check *.py",+ "format:check": "eslint . && cd src && eslint . && prettier --check \"**/*.{js,jsx,ts,tsx,json,css,md}\" && python3 -m ruff check *.py","preview": "cd src && vite preview","clean": "node cleanup.js"},diff --git a/pyproject.toml b/pyproject.tomlindex a0a03b555..ff08f5e50 100644--- a/pyproject.toml+++ b/pyproject.toml@@ -13,7 +13,6 @@ select = [ignore = ["E501", # line too long (let formatter handle it)"E402", # module level import not at top- "W503", # line break before binary operator][tool.ruff.format]diff --git a/whisper_bridge.py b/whisper_bridge.pyindex f846f4047..8dc5c0417 100755--- a/whisper_bridge.py+++ b/whisper_bridge.py@@ -6,16 +6,15 @@ Handles local speech-to-text processing using OpenAI's Whisper modelimport sysimport json-import tempfileimport osimport argparse-from pathlib import Pathimport whisperimport threadingimport timeimport requestsimport gc+def get_whisper_cache_dir():"""Get platform-appropriate cache directory for Whisper models"""if sys.platform == "win32":@@ -30,30 +29,31 @@ def get_whisper_cache_dir():# macOS/Linux: use ~/.cache/whisperreturn os.path.expanduser("~/.cache/whisper")+def get_ffmpeg_path():"""Get path to bundled FFmpeg executable with proper production support"""# Check environment variables firstenv_paths = [("FFMPEG_PATH", os.environ.get("FFMPEG_PATH")),- ("FFMPEG_EXECUTABLE", os.environ.get("FFMPEG_EXECUTABLE")),- ("FFMPEG_BINARY", os.environ.get("FFMPEG_BINARY"))+ ("FFMPEG_EXECUTABLE", os.environ.get("FFMPEG_EXECUTABLE")),+ ("FFMPEG_BINARY", os.environ.get("FFMPEG_BINARY")),]-+for env_name, env_path in env_paths:# On Windows, skip X_OK check since it doesn't use Unix-style execute permissionsif env_path and os.path.exists(env_path):if sys.platform == "win32" or os.access(env_path, os.X_OK):return env_path-+# Determine base path- if getattr(sys, 'frozen', False):+ if getattr(sys, "frozen", False):base_path = sys._MEIPASSelse:base_path = os.path.dirname(os.path.abspath(__file__))-+# Try multiple possible paths for production Electron apppossible_paths = []-+if sys.platform == "darwin": # macOSpossible_paths = [# Unpacked ASAR locations@@ -66,7 +66,9 @@ def get_ffmpeg_path():]elif sys.platform == "win32": # Windowspossible_paths = [- os.path.join(base_path, "..", "..", "..", "app.asar.unpacked", "node_modules", "ffmpeg-static", "ffmpeg.exe"),+ os.path.join(+ base_path, "..", "..", "..", "app.asar.unpacked", "node_modules", "ffmpeg-static", "ffmpeg.exe"+ ),os.path.join(base_path, "..", "app.asar.unpacked", "node_modules", "ffmpeg-static", "ffmpeg.exe"),os.path.join(base_path, "node_modules", "ffmpeg-static", "ffmpeg.exe"),os.path.join(base_path, "..", "node_modules", "ffmpeg-static", "ffmpeg.exe"),@@ -78,7 +80,7 @@ def get_ffmpeg_path():os.path.join(base_path, "node_modules", "ffmpeg-static", "ffmpeg"),os.path.join(base_path, "..", "node_modules", "ffmpeg-static", "ffmpeg"),]-+# Try each possible pathfor ffmpeg_path in possible_paths:abs_path = os.path.abspath(ffmpeg_path)@@ -86,40 +88,42 @@ def get_ffmpeg_path():if os.path.exists(abs_path):if sys.platform == "win32" or os.access(abs_path, os.X_OK):return abs_path-+# Try system FFmpeg as last resortif sys.platform == "darwin":common_ffmpeg_paths = ["/opt/homebrew/bin/ffmpeg", # Homebrew on Apple Silicon- "/usr/local/bin/ffmpeg", # Homebrew on Intel or manual installs- "/usr/bin/ffmpeg", # System location- "ffmpeg" # In PATH+ "/usr/local/bin/ffmpeg", # Homebrew on Intel or manual installs+ "/usr/bin/ffmpeg", # System location+ "ffmpeg", # In PATH]else:common_ffmpeg_paths = ["ffmpeg"]-+for ffmpeg_cmd in common_ffmpeg_paths:try:import subprocess+result = subprocess.run([ffmpeg_cmd, "-version"], capture_output=True, text=True, timeout=5)if result.returncode == 0:return ffmpeg_cmdexcept Exception:continue-+return None+# Set FFmpeg path for Whisperffmpeg_path = get_ffmpeg_path()if ffmpeg_path:os.environ["FFMPEG_BINARY"] = ffmpeg_path-+# CRITICAL: Add ffmpeg directory to PATH so Whisper can find itffmpeg_dir = os.path.dirname(os.path.abspath(ffmpeg_path))current_path = os.environ.get("PATH", "")if ffmpeg_dir not in current_path:os.environ["PATH"] = f"{ffmpeg_dir}{os.pathsep}{current_path}"-+# For Whisper library, we need to ensure 'ffmpeg' command works# Create a symlink if needed (for macOS/Linux)if sys.platform != "win32" and os.path.isfile(ffmpeg_path) and os.path.basename(ffmpeg_path) != "ffmpeg":@@ -133,124 +137,126 @@ if ffmpeg_path:# Global model cache to avoid reloading_model_cache = {}+def load_model(model_name="base"):"""Load Whisper model with caching for performance"""global _model_cache-+# Return cached model if availableif model_name in _model_cache:return _model_cache[model_name]-+try:model = whisper.load_model(model_name)-+if len(_model_cache) >= 2:oldest_key = next(iter(_model_cache))del _model_cache[oldest_key]gc.collect()-+_model_cache[model_name] = modelreturn model- except Exception as e:+ except Exception:return None+def get_expected_model_size(model_name):"""Get expected file size for a model by checking the remote URL"""try:model_url = whisper._MODELS[model_name]response = requests.head(model_url, timeout=10)if response.status_code == 200:- content_length = response.headers.get('content-length')+ content_length = response.headers.get("content-length")if content_length:return int(content_length)except Exception:pass-+approximate_sizes = {"tiny": 39 * 1024 * 1024,"base": 74 * 1024 * 1024,"small": 244 * 1024 * 1024,"medium": 769 * 1024 * 1024,"large": 1550 * 1024 * 1024,- "turbo": 809 * 1024 * 1024+ "turbo": 809 * 1024 * 1024,}return approximate_sizes.get(model_name, 100 * 1024 * 1024)+def monitor_download_progress(model_name, expected_size, stop_event):"""Monitor download progress by watching file size growth"""cache_dir = get_whisper_cache_dir()model_url = whisper._MODELS[model_name]model_file = os.path.join(cache_dir, os.path.basename(model_url))-+os.makedirs(cache_dir, exist_ok=True)-+last_size = 0last_update_time = time.time()speed_samples = []last_progress_update = 0-+while not stop_event.is_set():try:current_size = 0if os.path.exists(model_file):current_size = os.path.getsize(model_file)-+current_time = time.time()time_diff = current_time - last_update_time-+speed_mbps = 0if last_size > 0 and time_diff > 0 and current_size > last_size:bytes_per_second = (current_size - last_size) / time_diffspeed_mbps = (bytes_per_second * 8) / (1024 * 1024)-+speed_samples.append(speed_mbps)if len(speed_samples) > 10:speed_samples.pop(0)speed_mbps = sum(speed_samples) / len(speed_samples)-+percentage = min((current_size / expected_size * 100) if expected_size > 0 else 0, 100)-- if (current_time - last_progress_update > 0.5 or- abs(percentage - last_progress_update) > 1.0):-++ if current_time - last_progress_update > 0.5 or abs(percentage - last_progress_update) > 1.0:progress_data = {"type": "progress","model": model_name,"downloaded_bytes": current_size,"total_bytes": expected_size,"percentage": round(percentage, 1),- "speed_mbps": round(speed_mbps, 2) if speed_mbps > 0 else 0+ "speed_mbps": round(speed_mbps, 2) if speed_mbps > 0 else 0,}-+print(f"PROGRESS:{json.dumps(progress_data)}", file=sys.stderr)last_progress_update = percentage-+if current_size >= expected_size * 0.95:break-+if current_size == last_size and current_time - last_update_time > 10:if current_size > expected_size * 0.9:break-+last_size = current_sizelast_update_time = current_time-+except Exception:pass-+time.sleep(0.5)+def download_model(model_name="base"):"""Download Whisper model with real-time progress monitoring"""stop_event = threading.Event()progress_thread = None-+try:# Check if model is already downloadedcache_dir = get_whisper_cache_dir()model_url = whisper._MODELS[model_name]model_file = os.path.join(cache_dir, os.path.basename(model_url))-+if os.path.exists(model_file):file_size = os.path.getsize(model_file)return {@@ -259,70 +265,59 @@ def download_model(model_name="base"):"path": model_file,"size_bytes": file_size,"size_mb": round(file_size / (1024 * 1024), 1),- "success": True+ "success": True,}-+# Get expected file sizeexpected_size = get_expected_model_size(model_name)-+# Start progress monitoring in background threadprogress_thread = threading.Thread(- target=monitor_download_progress,- args=(model_name, expected_size, stop_event),- daemon=True+ target=monitor_download_progress, args=(model_name, expected_size, stop_event), daemon=True)progress_thread.start()-+# Start the actual download (this will block until complete)- model = whisper.load_model(model_name)-+ whisper.load_model(model_name)+# Stop progress monitoringstop_event.set()-+# Wait for progress thread to finishif progress_thread and progress_thread.is_alive():progress_thread.join(timeout=1)-+# Get final file infofinal_size = 0if os.path.exists(model_file):final_size = os.path.getsize(model_file)-+# Send completion signalcompletion_data = {"type": "complete","model": model_name,"downloaded_bytes": final_size,"total_bytes": expected_size,- "percentage": 100+ "percentage": 100,}print(f"PROGRESS:{json.dumps(completion_data)}", file=sys.stderr)-+return {"model": model_name,"downloaded": True,"path": model_file,"size_bytes": final_size,"size_mb": round(final_size / (1024 * 1024), 1),- "success": True+ "success": True,}-+except KeyboardInterrupt:stop_event.set()- return {- "model": model_name,- "downloaded": False,- "error": "Download interrupted by user",- "success": False- }+ return {"model": model_name, "downloaded": False, "error": "Download interrupted by user", "success": False}except Exception as e:stop_event.set()- return {- "model": model_name,- "downloaded": False,- "error": str(e),- "success": False- }+ return {"model": model_name, "downloaded": False, "error": str(e), "success": False}+def check_model_status(model_name="base"):"""Check if a model is already downloaded"""@@ -330,7 +325,7 @@ def check_model_status(model_name="base"):cache_dir = get_whisper_cache_dir()model_url = whisper._MODELS[model_name]model_file = os.path.join(cache_dir, os.path.basename(model_url))-+if os.path.exists(model_file):file_size = os.path.getsize(model_file)return {@@ -339,35 +334,25 @@ def check_model_status(model_name="base"):"path": model_file,"size_bytes": file_size,"size_mb": round(file_size / (1024 * 1024), 1),- "success": True+ "success": True,}else:- return {- "model": model_name,- "downloaded": False,- "success": True- }+ return {"model": model_name, "downloaded": False, "success": True}except Exception as e:- return {- "model": model_name,- "error": str(e),- "success": False- }+ return {"model": model_name, "error": str(e), "success": False}+def list_models():"""List all available models and their download status"""models = ["tiny", "base", "small", "medium", "large", "turbo"]model_info = []-+for model in models:status = check_model_status(model)model_info.append(status)-- return {- "models": model_info,- "cache_dir": get_whisper_cache_dir(),- "success": True- }++ return {"models": model_info, "cache_dir": get_whisper_cache_dir(), "success": True}+def delete_model(model_name="base"):"""Delete a downloaded Whisper model"""@@ -375,7 +360,7 @@ def delete_model(model_name="base"):cache_dir = get_whisper_cache_dir()model_url = whisper._MODELS[model_name]model_file = os.path.join(cache_dir, os.path.basename(model_url))-+if os.path.exists(model_file):file_size = os.path.getsize(model_file)os.remove(model_file)@@ -384,117 +369,92 @@ def delete_model(model_name="base"):"deleted": True,"freed_bytes": file_size,"freed_mb": round(file_size / (1024 * 1024), 1),- "success": True+ "success": True,}else:- return {- "model": model_name,- "deleted": False,- "error": "Model not found",- "success": False- }+ return {"model": model_name, "deleted": False, "error": "Model not found", "success": False}except Exception as e:- return {- "model": model_name,- "deleted": False,- "error": str(e),- "success": False- }+ return {"model": model_name, "deleted": False, "error": str(e), "success": False}+def transcribe_audio(audio_path, model_name="base", language=None):"""Transcribe audio file using Whisper with optimizations"""-+if not os.path.exists(audio_path):return {"error": f"Audio file not found: {audio_path}", "success": False}-+try:# Load model (uses cache for performance)model = load_model(model_name)if model is None:return {"error": "Failed to load Whisper model", "success": False}-+options = {"fp16": False,"verbose": False,}if language:options["language"] = language-+result = model.transcribe(audio_path, **options)-+text = result.get("text", "").strip()language = result.get("language", "unknown")-- return {- "text": text,- "language": language,- "success": True- }-++ return {"text": text, "language": language, "success": True}+except Exception as e:- return {- "error": str(e),- "success": False- }+ return {"error": str(e), "success": False}+def check_ffmpeg():"""Check if FFmpeg is available and working"""try:import subprocess+test_path = ffmpeg_path or "ffmpeg"-- result = subprocess.run([test_path, "-version"],- capture_output=True, text=True, timeout=10)-++ result = subprocess.run([test_path, "-version"], capture_output=True, text=True, timeout=10)+if result.returncode == 0:- version_line = result.stdout.split('\n')[0] if result.stdout else "Unknown"- return {- "available": True,- "path": test_path,- "version": version_line,- "success": True- }+ version_line = result.stdout.split("\n")[0] if result.stdout else "Unknown"+ return {"available": True, "path": test_path, "version": version_line, "success": True}else:return {"available": False,"error": f"FFmpeg returned code {result.returncode}: {result.stderr}",- "success": False+ "success": False,}except subprocess.TimeoutExpired:- return {- "available": False,- "error": "FFmpeg check timed out",- "success": False- }+ return {"available": False, "error": "FFmpeg check timed out", "success": False}except FileNotFoundError:- return {- "available": False,- "error": "FFmpeg not found in PATH",- "success": False- }+ return {"available": False, "error": "FFmpeg not found in PATH", "success": False}except Exception as e:- return {- "available": False,- "error": str(e),- "success": False- }+ return {"available": False, "error": str(e), "success": False}+def main():parser = argparse.ArgumentParser(description="Whisper Bridge for OpenWhispr")- parser.add_argument("--mode", default="transcribe",- choices=["transcribe", "download", "check", "list", "delete", "check-ffmpeg"],- help="Operation mode (default: transcribe)")+ parser.add_argument(+ "--mode",+ default="transcribe",+ choices=["transcribe", "download", "check", "list", "delete", "check-ffmpeg"],+ help="Operation mode (default: transcribe)",+ )parser.add_argument("audio_file", nargs="?", help="Path to audio file to transcribe")- parser.add_argument("--model", default="base",- choices=["tiny", "base", "small", "medium", "large", "turbo"],- help="Whisper model to use (default: base)")+ parser.add_argument(+ "--model",+ default="base",+ choices=["tiny", "base", "small", "medium", "large", "turbo"],+ help="Whisper model to use (default: base)",+ )parser.add_argument("--language", help="Language code (optional)")- parser.add_argument("--output-format", default="json",- choices=["json", "text"],- help="Output format (default: json)")-+ parser.add_argument(+ "--output-format", default="json", choices=["json", "text"], help="Output format (default: json)"+ )+args = parser.parse_args()-+# Handle different modesif args.mode == "download":result = download_model(args.model)@@ -522,15 +482,15 @@ def main():error_result = {"error": "Audio file required for transcription mode", "success": False}print(json.dumps(error_result))sys.exit(1)-+if not os.path.exists(args.audio_file):error_result = {"error": f"Audio file not found: {args.audio_file}", "success": False}print(json.dumps(error_result))sys.exit(1)-+# Transcriberesult = transcribe_audio(args.audio_file, args.model, args.language)-+# Output resultsif args.output_format == "json":print(json.dumps(result))@@ -541,5 +501,6 @@ def main():print(f"Error: {result.get('error', 'Unknown error')}", file=sys.stderr)sys.exit(1)+if __name__ == "__main__":- main()\ No newline at end of file+ main()