lattifai 1.0.1__py3-none-any.whl → 1.0.2__py3-none-any.whl

lattifai/alignment/lattice1_worker.py

@@ -30,28 +30,42 @@ class Lattice1Worker:
         sess_options.execution_mode = ort.ExecutionMode.ORT_PARALLEL
         sess_options.add_session_config_entry("session.intra_op.allow_spinning", "0")
 
+        acoustic_model_path = f"{model_path}/acoustic_opt.onnx"
+
         providers = []
-        if device.startswith("cuda") and ort.get_all_providers().count("CUDAExecutionProvider") > 0:
+        all_providers = ort.get_all_providers()
+        if device.startswith("cuda") and all_providers.count("CUDAExecutionProvider") > 0:
             providers.append("CUDAExecutionProvider")
-        elif device.startswith("mps") and ort.get_all_providers().count("MPSExecutionProvider") > 0:
+        if "MPSExecutionProvider" in all_providers:
             providers.append("MPSExecutionProvider")
+        if "CoreMLExecutionProvider" in all_providers:
+            if "quant" in acoustic_model_path:
+                # NOTE: CPUExecutionProvider is faster for quantized models
+                pass
+            else:
+                providers.append("CoreMLExecutionProvider")
 
         try:
             self.acoustic_ort = ort.InferenceSession(
-                f"{model_path}/acoustic_opt.onnx",
+                acoustic_model_path,
                 sess_options,
-                providers=providers + ["CPUExecutionProvider", "CoreMLExecutionProvider"],
+                providers=providers + ["CPUExecutionProvider"],
             )
         except Exception as e:
             raise ModelLoadError(f"acoustic model from {model_path}", original_error=e)
 
-        try:
-            config = FbankConfig(num_mel_bins=80, device=device, snip_edges=False)
-            config_dict = config.to_dict()
-            config_dict.pop("device")
-            self.extractor = Wav2LogFilterBank(**config_dict).to(device).eval()
-        except Exception as e:
-            raise ModelLoadError(f"feature extractor for device {device}", original_error=e)
+        # get input_names
+        input_names = [inp.name for inp in self.acoustic_ort.get_inputs()]
+        if "audios" not in input_names:
+            try:
+                config = FbankConfig(num_mel_bins=80, device=device, snip_edges=False)
+                config_dict = config.to_dict()
+                config_dict.pop("device")
+                self.extractor = Wav2LogFilterBank(**config_dict).to(device).eval()
+            except Exception as e:
+                raise ModelLoadError(f"feature extractor for device {device}", original_error=e)
+        else:
+            self.extractor = None  # ONNX model includes feature extractor
 
         self.device = torch.device(device)
         self.timings = defaultdict(lambda: 0.0)
@@ -63,28 +77,57 @@ class Lattice1Worker:
     @torch.inference_mode()
     def emission(self, audio: torch.Tensor) -> torch.Tensor:
         _start = time.time()
-        # audio -> features -> emission
-        features = self.extractor(audio)  # (1, T, D)
-        if features.shape[1] > 6000:
-            features_list = torch.split(features, 6000, dim=1)
-            emissions = []
-            for features in features_list:
+        if self.extractor is not None:
+            # audio -> features -> emission
+            features = self.extractor(audio)  # (1, T, D)
+            if features.shape[1] > 6000:
+                features_list = torch.split(features, 6000, dim=1)
+                emissions = []
+                for features in features_list:
+                    ort_inputs = {
+                        "features": features.cpu().numpy(),
+                        "feature_lengths": np.array([features.size(1)], dtype=np.int64),
+                    }
+                    emission = self.acoustic_ort.run(None, ort_inputs)[0]  # (1, T, vocab_size) numpy
+                    emissions.append(emission)
+                emission = torch.cat(
+                    [torch.from_numpy(emission).to(self.device) for emission in emissions], dim=1
+                )  # (1, T, vocab_size)
+            else:
                 ort_inputs = {
                     "features": features.cpu().numpy(),
                     "feature_lengths": np.array([features.size(1)], dtype=np.int64),
                 }
                 emission = self.acoustic_ort.run(None, ort_inputs)[0]  # (1, T, vocab_size) numpy
-                emissions.append(emission)
-            emission = torch.cat(
-                [torch.from_numpy(emission).to(self.device) for emission in emissions], dim=1
-            )  # (1, T, vocab_size)
+                emission = torch.from_numpy(emission).to(self.device)
         else:
-            ort_inputs = {
-                "features": features.cpu().numpy(),
-                "feature_lengths": np.array([features.size(1)], dtype=np.int64),
-            }
-            emission = self.acoustic_ort.run(None, ort_inputs)[0]  # (1, T, vocab_size) numpy
-            emission = torch.from_numpy(emission).to(self.device)
+            CHUNK_SIZE = 60 * 16000  # 60 seconds
+            if audio.shape[1] > CHUNK_SIZE:
+                audio_list = torch.split(audio, CHUNK_SIZE, dim=1)
+                emissions = []
+                for audios in audio_list:
+                    emission = self.acoustic_ort.run(
+                        None,
+                        {
+                            "audios": audios.cpu().numpy(),
+                        },
+                    )[
+                        0
+                    ]  # (1, T, vocab_size) numpy
+                    emissions.append(emission)
+                emission = torch.cat(
+                    [torch.from_numpy(emission).to(self.device) for emission in emissions], dim=1
+                )  # (1, T, vocab_size)
+            else:
+                emission = self.acoustic_ort.run(
+                    None,
+                    {
+                        "audios": audio.cpu().numpy(),
+                    },
+                )[
+                    0
+                ]  # (1, T, vocab_size) numpy
+                emission = torch.from_numpy(emission).to(self.device)
 
         self.timings["emission"] += time.time() - _start
         return emission  # (1, T, vocab_size) torch

lattifai/cli/app_installer.py

@@ -0,0 +1,140 @@
+"""CLI tool to install lai-app (frontend web application)."""
+
+import platform
+import subprocess
+import sys
+from pathlib import Path
+
+
+def check_command_exists(cmd: str) -> bool:
+    """Check if a command exists in PATH."""
+    try:
+        subprocess.run([cmd, "--version"], check=True, capture_output=True, text=True)
+        return True
+    except (subprocess.CalledProcessError, FileNotFoundError):
+        return False
+
+
+def install_nodejs():
+    """Install Node.js based on the operating system."""
+    system = platform.system().lower()
+
+    print("📦 Node.js not found. Installing Node.js...\n")
+
+    try:
+        if system == "darwin":  # macOS
+            # Check if Homebrew is installed
+            if check_command_exists("brew"):
+                print("🍺 Using Homebrew to install Node.js...")
+                subprocess.run(["brew", "install", "node"], check=True)
+                print("✓ Node.js installed via Homebrew\n")
+            else:
+                print("❌ Homebrew not found.")
+                print("   Please install Homebrew first:")
+                print(
+                    '   /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"'
+                )
+                print("\n   Or install Node.js manually from: https://nodejs.org/")
+                sys.exit(1)
+
+        elif system == "linux":
+            # Try common package managers
+            if check_command_exists("apt"):
+                print("🐧 Using apt to install Node.js...")
+                subprocess.run(["sudo", "apt", "update"], check=True)
+                subprocess.run(["sudo", "apt", "install", "-y", "nodejs", "npm"], check=True)
+                print("✓ Node.js installed via apt\n")
+            elif check_command_exists("yum"):
+                print("🐧 Using yum to install Node.js...")
+                subprocess.run(["sudo", "yum", "install", "-y", "nodejs", "npm"], check=True)
+                print("✓ Node.js installed via yum\n")
+            elif check_command_exists("dnf"):
+                print("🐧 Using dnf to install Node.js...")
+                subprocess.run(["sudo", "dnf", "install", "-y", "nodejs", "npm"], check=True)
+                print("✓ Node.js installed via dnf\n")
+            elif check_command_exists("pacman"):
+                print("🐧 Using pacman to install Node.js...")
+                subprocess.run(["sudo", "pacman", "-S", "--noconfirm", "nodejs", "npm"], check=True)
+                print("✓ Node.js installed via pacman\n")
+            else:
+                print("❌ No supported package manager found (apt/yum/dnf/pacman).")
+                print("   Please install Node.js manually from: https://nodejs.org/")
+                sys.exit(1)
+
+        elif system == "windows":
+            print("❌ Automatic installation on Windows is not supported.")
+            print("   Please download and install Node.js from: https://nodejs.org/")
+            print("   Then run this command again.")
+            sys.exit(1)
+
+        else:
+            print(f"❌ Unsupported operating system: {system}")
+            print("   Please install Node.js manually from: https://nodejs.org/")
+            sys.exit(1)
+
+        # Verify installation
+        if not check_command_exists("npm"):
+            print("❌ Node.js installation verification failed.")
+            print("   Please restart your terminal and try again.")
+            sys.exit(1)
+
+    except subprocess.CalledProcessError as e:
+        print(f"\n❌ Error during Node.js installation: {e}")
+        print("   Please install Node.js manually from: https://nodejs.org/")
+        sys.exit(1)
+
+
+def main():
+    """Install lai-app Node.js application."""
+    # Get the app directory relative to this package
+    app_dir = Path(__file__).parent.parent.parent.parent / "app"
+
+    if not app_dir.exists():
+        print(f"❌ Error: app directory not found at {app_dir}")
+        print("   Make sure you're in the lattifai-python repository.")
+        sys.exit(1)
+
+    print("🚀 Installing lai-app (LattifAI Web Application)...\n")
+
+    # Check if npm is installed, if not, install Node.js
+    if not check_command_exists("npm"):
+        install_nodejs()
+    else:
+        npm_version = subprocess.run(["npm", "--version"], capture_output=True, text=True, check=True).stdout.strip()
+        print(f"✓ npm is already installed (v{npm_version})\n")
+
+    # Change to app directory and run installation
+    try:
+        print(f"📁 Working directory: {app_dir}\n")
+
+        # Install dependencies
+        print("📦 Installing dependencies...")
+        subprocess.run(["npm", "install"], cwd=app_dir, check=True)
+        print("✓ Dependencies installed\n")
+
+        # Build the application
+        print("🔨 Building application...")
+        subprocess.run(["npm", "run", "build"], cwd=app_dir, check=True)
+        print("✓ Application built\n")
+
+        # Link globally
+        print("🔗 Linking lai-app command globally...")
+        subprocess.run(["npm", "link"], cwd=app_dir, check=True)
+        print("✓ lai-app command linked globally\n")
+
+        print("=" * 60)
+        print("✅ lai-app installed successfully!")
+        print("=" * 60)
+        print("\n🎉 You can now run:")
+        print("   lai-app              # Start the web application")
+        print("   lai-app --help       # Show help")
+        print("   lai-app --port 8080  # Use custom port")
+        print("\n📖 For more information, see app/CLI_USAGE.md\n")
+
+    except subprocess.CalledProcessError as e:
+        print(f"\n❌ Error during installation: {e}")
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

lattifai/cli/server.py

@@ -1,3 +1,4 @@
+import argparse
 import os
 
 import colorful
@@ -6,13 +7,35 @@ import uvicorn
 
 def main():
     """Launch the LattifAI Web Interface."""
-    print(colorful.bold_green("🚀 Launching LattifAI Web Interface..."))
-    print(colorful.cyan("See http://localhost:8001"))
+    parser = argparse.ArgumentParser(description="LattifAI Backend Server")
+    parser.add_argument(
+        "-p",
+        "--port",
+        type=int,
+        default=8001,
+        help="Port to run the server on (default: 8001)",
+    )
+    parser.add_argument(
+        "--host",
+        type=str,
+        default="0.0.0.0",
+        help="Host to bind the server to (default: 0.0.0.0)",
+    )
+    parser.add_argument(
+        "--no-reload",
+        action="store_true",
+        help="Disable auto-reload on code changes",
+    )
 
-    # Ensure the directory contains the app
-    # We might need to adjust python path or just rely on installed package
+    args = parser.parse_args()
 
-    uvicorn.run("lattifai.server.app:app", host="0.0.0.0", port=8001, reload=True, log_level="info")
+    print(colorful.bold_green("🚀 Launching LattifAI Backend Server..."))
+    print(colorful.cyan(f"Server running at http://localhost:{args.port}"))
+    print(colorful.yellow(f"Host: {args.host}"))
+    print(colorful.yellow(f"Auto-reload: {'disabled' if args.no_reload else 'enabled'}"))
+    print()
+
+    uvicorn.run("lattifai.server.app:app", host=args.host, port=args.port, reload=not args.no_reload, log_level="info")
 
 
 if __name__ == "__main__":

lattifai/server/app.py

@@ -58,6 +58,12 @@ app.add_middleware(
 )
 
 
+@app.get("/health")
+async def health_check():
+    """Health check endpoint for server status monitoring."""
+    return {"status": "ok", "message": "LattifAI backend server is running"}
+
+
 def mask_api_key(key: str) -> str:
     """Mask API key for display, showing only first 6 and last 4 characters."""
     if len(key) <= 10:

{lattifai-1.0.1.dist-info → lattifai-1.0.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lattifai
-Version: 1.0.1
+Version: 1.0.2
 Summary: Lattifai Python SDK: Seamless Integration with Lattifai's Speech and Video AI Services
 Author-email: Lattifai Technologies <tech@lattifai.com>
 Maintainer-email: Lattice <tech@lattifai.com>
@@ -155,7 +155,7 @@ Advanced forced alignment and subtitle generation powered by [ 🤗 Lattice-1](h
 pip install install-k2
 install-k2 --torch-version 2.9.1  # if not set will auto-detect PyTorch version and install compatible k2
 
-pip install lattifai==1.0.0
+pip install lattifai==1.0.2
 ```
 
 **Using uv (Recommended - 10-100x faster):**
@@ -173,8 +173,8 @@ uv pip install install-k2
 uv pip install pip
 uv run install-k2 --torch-version 2.9.1
 
-# Install LattifAI (v1.0.0)
-uv pip install lattifai==1.0.0
+# Install LattifAI (v1.0.2)
+uv pip install lattifai==1.0.2
 ```
 
 > **Note**: `install-k2` automatically detects your PyTorch version (up to 2.9) and installs the compatible k2 wheel.
@@ -257,19 +257,69 @@ That's it! Your aligned subtitles are saved to `aligned.srt`.
 
 ### Web Interface
 
-1. **Start the backend server:**
+![web Demo](assets/web.png)
+
+1. **Install the web application (one-time setup):**
+   ```bash
+   lai-app-install
+   ```
+
+   This command will:
+   - Check if Node.js/npm is installed (and install if needed)
+   - Install frontend dependencies
+   - Build the application
+   - Setup the `lai-app` command globally
+
+2. **Start the backend server:**
    ```bash
    lai-server
+
+   # Custom port (default: 8001)
+   lai-server --port 9000
+
+   # Custom host
+   lai-server --host 127.0.0.1 --port 9000
+
+   # Production mode (disable auto-reload)
+   lai-server --no-reload
    ```
 
-2. **Start the frontend (in a new terminal):**
+   **Backend Server Options:**
+   - `-p, --port` - Server port (default: 8001)
+   - `--host` - Host address (default: 0.0.0.0)
+   - `--no-reload` - Disable auto-reload for production
+   - `-h, --help` - Show help message
+
+3. **Start the frontend application:**
    ```bash
-   cd app
-   npm install
-   npm run dev
+   lai-app
+
+   # Custom port (default: 5173)
+   lai-app --port 8080
+
+   # Custom backend URL
+   lai-app --backend http://localhost:9000
+
+   # Don't auto-open browser
+   lai-app --no-open
    ```
 
-   Visit `http://localhost:5173` to open the web interface.
+   **Frontend Application Options:**
+   - `-p, --port` - Frontend server port (default: 5173)
+   - `--backend` - Backend API URL (default: http://localhost:8001)
+   - `--no-open` - Don't automatically open browser
+   - `-h, --help` - Show help message
+
+   The web interface will automatically open in your browser at `http://localhost:5173`.
+
+**Features:**
+- ✅ Automatic backend server status detection
+- ✅ Visual file upload with drag-and-drop
+- ✅ Real-time alignment progress
+- ✅ Multiple subtitle format support
+- ✅ Built-in transcription with multiple models
+- ✅ API key management interface
+- ✅ Download aligned subtitles in various formats
 
 ---
 

{lattifai-1.0.1.dist-info → lattifai-1.0.2.dist-info}/RECORD

@@ -8,7 +8,7 @@ lattifai/types.py,sha256=SjYBfwrCBOXlICvH04niFQJ7OzTx7oTaa_npfRkB67U,659
 lattifai/utils.py,sha256=TqOPrd_Et7KxrbfI_JbBNIGZ5-oGJY8ZUyJMPDTih1I,3848
 lattifai/alignment/__init__.py,sha256=ehpkKfjNIYUx7_M-RWD_8Efcrzd9bE-NSm0QgMMVLW0,178
 lattifai/alignment/lattice1_aligner.py,sha256=soBRZ98jRIju-wN5eqYUmQfF56KiEUxVGw0UvtRcx4A,4464
-lattifai/alignment/lattice1_worker.py,sha256=U3izvaBCnrRmr2jBnaEcHLJLGHbZlHNb1qroFxTH8oQ,7440
+lattifai/alignment/lattice1_worker.py,sha256=XGICEzLygspqC7SMj5s45M7VNSj-l2zemkawdKKjssw,9233
 lattifai/alignment/phonemizer.py,sha256=fbhN2DOl39lW4nQWKzyUUTMUabg7v61lB1kj8SKK-Sw,1761
 lattifai/alignment/segmenter.py,sha256=-FKtIwv9Z4fU9Fs08jhL9VyREVSYcfcwuTqb8jxCiuo,6228
 lattifai/alignment/tokenizer.py,sha256=WilqU9Ecdkl_cW86IkB1mh_PFlHN-35Jsreiyse2r-8,22355
@@ -20,8 +20,9 @@ lattifai/caption/supervision.py,sha256=DRrM8lfKU_x9aVBcLG6xnT0xIJrnc8jzHpzcSwQOg
 lattifai/caption/text_parser.py,sha256=XDb8KTt031uJ1hg6dpbINglGOTX-6pBcghbg3DULM1I,4633
 lattifai/cli/__init__.py,sha256=dIUmrpN-OwR4h6BqMhXp87_5ZwgO41ShPru_iZGnpQs,463
 lattifai/cli/alignment.py,sha256=uKMTE95_JMikfbyCcwLbQxms-EQmZXEj7oYugiupk9I,5890
+lattifai/cli/app_installer.py,sha256=0xBQnJZKhyx4JT_PkHXkZ0XlAWxCGz9o0Jjq99poKew,5680
 lattifai/cli/caption.py,sha256=ucgYxJ43ab71nGpZBAiVn8QA0DAVht2QMZFE5IdgxP0,6853
-lattifai/cli/server.py,sha256=k3kuQsZ2JNHZD_eMDxGSn5OFHiUrC2ybHN_xQQlLEgU,493
+lattifai/cli/server.py,sha256=Vo6_ANgwu7WtC5h4BebQLmhqLNpqzPoYrPQPANpP7rw,1142
 lattifai/cli/transcribe.py,sha256=6uJfvtB1o_u1uQwxt4fje_koyfN93mGaFLlskmjqx2c,7406
 lattifai/cli/youtube.py,sha256=9_erdIkhX8pCiy7BRzNstEiO9saM-VKZ1WVqvbXbmrc,5267
 lattifai/config/__init__.py,sha256=Z8OudvS6fgfLNLu_2fvoXartQiYCECOnNfzDt-PfCN4,543
@@ -33,7 +34,7 @@ lattifai/config/media.py,sha256=5JOPjifXDM2WWQERySDZen4-7YfgQNcYM2NkkKp0LjQ,1361
 lattifai/config/transcription.py,sha256=bzghOGgcNWzTnDYd_cqCOB7GT8OnzHDiyam7LSixqxM,2901
 lattifai/diarization/__init__.py,sha256=MgBDQ1ehL2qDnZprEp8KqON7CmbG-qaP37gzBsV0jzk,119
 lattifai/diarization/lattifai.py,sha256=SE2BpIZ3_deKyhXdBqe77bsDLXIUV9AQV34gfINv7_s,2657
-lattifai/server/app.py,sha256=ndMBC_FuAFZGs9bvrgHUMtgDDDeYGhCkW0Lx49SEh7k,15193
+lattifai/server/app.py,sha256=UpHsKJHtK1-sdp5mtDPBSxEl8xYTbe7cVO8dLp9Xiuo,15380
 lattifai/transcription/__init__.py,sha256=mEoMTbs5jAgtXQn1jTjlFY_GUr-S0WmPn8uZ6WZCkU0,2643
 lattifai/transcription/base.py,sha256=59b4nQHFMyTRyyzBJTM8ZpEuUy1KjwA2o6rNfrNluKY,3911
 lattifai/transcription/gemini.py,sha256=1VNi9gl-Kpkw3ljZcOZG5oq_OY8fMC9Xv4kOwyQpI0Q,7992
@@ -47,9 +48,9 @@ lattifai/workflow/agents.py,sha256=yEOnxnhcTvr1iOhCorNvp8B76P6nQsLRXJCu_rCYFfM,3
 lattifai/workflow/base.py,sha256=8QoVIBZwJfr5mppJbtUFafHv5QR9lL-XrULjTWD0oBg,6257
 lattifai/workflow/file_manager.py,sha256=d106KHLY8A9amLy5h1vR32e4od8mmJGqMD-iDyiRPLI,32917
 lattifai/workflow/youtube.py,sha256=n8L1c6tl8FuYzAzKZ-B76zf5yZsvVggZEJ9mPdbEWGQ,22989
-lattifai-1.0.1.dist-info/licenses/LICENSE,sha256=_IkHdwOWLAWcE1M_tIpDoRWdNSJwFdtIqI-XSkK3yPU,1066
-lattifai-1.0.1.dist-info/METADATA,sha256=3w5o4QmSENLp9PDaqwmnXJnv2l57q1oKEae_3QeUooI,23109
-lattifai-1.0.1.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
-lattifai-1.0.1.dist-info/entry_points.txt,sha256=Ay-Rxb214lYSuJIXO_wuNlUu1LbvOXtMJ-G-PcvGZFw,424
-lattifai-1.0.1.dist-info/top_level.txt,sha256=tHSoXF26r-IGfbIP_JoYATqbmf14h5NrnNJGH4j5reI,9
-lattifai-1.0.1.dist-info/RECORD,,
+lattifai-1.0.2.dist-info/licenses/LICENSE,sha256=_IkHdwOWLAWcE1M_tIpDoRWdNSJwFdtIqI-XSkK3yPU,1066
+lattifai-1.0.2.dist-info/METADATA,sha256=Ew2BnhBAeZyUBCKJtvl4qJz1IomUsPL2-3COXZc0fqw,24571
+lattifai-1.0.2.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
+lattifai-1.0.2.dist-info/entry_points.txt,sha256=F8Akof3VtKtrbnYSav1umgoo9Xbv34rUcKn-ioRfeGQ,474
+lattifai-1.0.2.dist-info/top_level.txt,sha256=tHSoXF26r-IGfbIP_JoYATqbmf14h5NrnNJGH4j5reI,9
+lattifai-1.0.2.dist-info/RECORD,,

{lattifai-1.0.1.dist-info → lattifai-1.0.2.dist-info}/entry_points.txt

@@ -1,5 +1,6 @@
 [console_scripts]
 lai-align = lattifai.cli.alignment:main
+lai-app-install = lattifai.cli.app_installer:main
 lai-server = lattifai.cli.server:main
 lai-transcribe = lattifai.cli.transcribe:main
 lai-youtube = lattifai.cli.youtube:main