lattifai 1.0.5__py3-none-any.whl → 1.1.0__py3-none-any.whl

lattifai/diarization/lattifai.py

@@ -1,11 +1,12 @@
 """LattifAI speaker diarization implementation."""
 
 import logging
-from collections import defaultdict
-from typing import List, Optional, Tuple
+from dataclasses import dataclass
+from typing import Callable, List, Optional, Tuple
 
-import torch
-from tgt import Interval, IntervalTier, TextGrid
+import numpy as np
+from lattifai_core.diarization import DiarizationOutput
+from tgt import TextGrid
 
 from lattifai.audio2 import AudioData
 from lattifai.caption import Supervision
@@ -60,7 +61,7 @@ class LattifAIDiarizer:
         num_speakers: Optional[int] = None,
         min_speakers: Optional[int] = None,
         max_speakers: Optional[int] = None,
-    ) -> TextGrid:
+    ) -> DiarizationOutput:
         """Perform speaker diarization on the input audio."""
         return self.diarizer.diarize(
             input_media,
@@ -73,11 +74,16 @@ class LattifAIDiarizer:
         self,
         input_media: AudioData,
         alignments: List[Supervision],
-        diarization: Optional[TextGrid] = None,
+        diarization: Optional[DiarizationOutput] = None,
         num_speakers: Optional[int] = None,
         min_speakers: Optional[int] = None,
         max_speakers: Optional[int] = None,
-    ) -> Tuple[TextGrid, List[Supervision]]:
+        alignment_fn: Optional[Callable] = None,
+        transcribe_fn: Optional[Callable] = None,
+        separate_fn: Optional[Callable] = None,
+        debug: bool = False,
+        output_path: Optional[str] = None,
+    ) -> Tuple[DiarizationOutput, List[Supervision]]:
         """Diarize the given media input and return alignments with refined speaker labels."""
         return self.diarizer.diarize_with_alignments(
             input_media,
@@ -86,4 +92,9 @@ class LattifAIDiarizer:
             num_speakers=num_speakers,
             min_speakers=min_speakers,
             max_speakers=max_speakers,
+            alignment_fn=alignment_fn,
+            transcribe_fn=transcribe_fn,
+            separate_fn=separate_fn,
+            debug=debug,
+            output_path=output_path,
         )

lattifai/mixin.py

@@ -184,7 +184,9 @@ class LattifAIClientMixin:
         from lattifai.utils import _resolve_model_path
 
         if transcription_config is not None:
-            transcription_config.lattice_model_path = _resolve_model_path(alignment_config.model_name)
+            transcription_config.lattice_model_path = _resolve_model_path(
+                alignment_config.model_name, getattr(alignment_config, "model_hub", "huggingface")
+            )
 
         # Set client_wrapper for all configs
         alignment_config.client_wrapper = self
@@ -380,6 +382,7 @@ class LattifAIClientMixin:
         media_file: Union[str, Path, AudioData],
         source_lang: Optional[str],
         is_async: bool = False,
+        output_dir: Optional[Path] = None,
     ) -> Caption:
         """
         Get captions by downloading or transcribing.
@@ -406,6 +409,9 @@ class LattifAIClientMixin:
             safe_print(colorful.green("         ✓ Transcription completed."))
 
             if "gemini" in self.transcriber.name.lower():
+                safe_print(colorful.yellow("🔍 Gemini raw output:"))
+                safe_print(colorful.yellow(f"{transcription[:1000]}..."))  # Print first 1000 chars
+
                 # write to temp file and use Caption read
                 # On Windows, we need to close the file before writing to it
                 tmp_file = tempfile.NamedTemporaryFile(
@@ -428,6 +434,18 @@ class LattifAIClientMixin:
                     # Clean up temp file
                     if tmp_path.exists():
                         tmp_path.unlink()
+            else:
+                safe_print(colorful.yellow(f"🔍 {self.transcriber.name} raw output:"))
+                if isinstance(transcription, Caption):
+                    safe_print(colorful.yellow(f"Caption with {len(transcription.transcription)} segments"))
+                    if transcription.transcription:
+                        safe_print(colorful.yellow(f"First segment: {transcription.transcription[0].text}"))
+
+            if output_dir:
+                # Generate transcript file path
+                transcript_file = output_dir / f"{Path(str(media_file)).stem}_{self.transcriber.file_name}"
+                await asyncio.to_thread(self.transcriber.write, transcription, transcript_file, encoding="utf-8")
+                safe_print(colorful.green(f"         ✓ Transcription saved to: {transcript_file}"))
 
             return transcription
 

lattifai/transcription/__init__.py

@@ -70,7 +70,7 @@ def create_transcriber(
         raise ValueError(
             f"Cannot determine transcriber for model_name='{transcription_config.model_name}'. "
             f"Supported patterns: \n"
-            f"  - Gemini API models: 'gemini-2.5-pro', 'gemini-3-pro-preview'\n"
+            f"  - Gemini API models: 'gemini-2.5-pro', 'gemini-3-pro-preview', 'gemini-3-flash-preview'\n"
             f"  - Local HF models: 'nvidia/parakeet-*', 'iic/SenseVoiceSmall', etc.\n"
             f"Please specify a valid model_name."
         )

lattifai/transcription/base.py

@@ -2,10 +2,12 @@
 
 from abc import ABC, abstractmethod
 from pathlib import Path
-from typing import Optional, Union
+from typing import List, Optional, Union
+
+import numpy as np
 
 from lattifai.audio2 import AudioData
-from lattifai.caption import Caption
+from lattifai.caption import Caption, Supervision
 from lattifai.config import TranscriptionConfig
 from lattifai.logging import get_logger
 
@@ -96,6 +98,23 @@ class BaseTranscriber(ABC):
             language: Optional language code for transcription.
         """
 
+    @abstractmethod
+    def transcribe_numpy(
+        self,
+        audio: Union[np.ndarray, List[np.ndarray]],
+        language: Optional[str] = None,
+    ) -> Union[Supervision, List[Supervision]]:
+        """
+        Transcribe audio from a numpy array and return Supervision.
+
+        Args:
+            audio_array: Audio data as numpy array (shape: [samples]).
+            language: Optional language code for transcription.
+
+        Returns:
+            Supervision object with transcription info.
+        """
+
     @abstractmethod
     def write(self, transcript: Union[str, Caption], output_file: Path, encoding: str = "utf-8") -> Path:
         """

lattifai/transcription/gemini.py

@@ -2,12 +2,14 @@
 
 import asyncio
 from pathlib import Path
-from typing import Optional, Union
+from typing import List, Optional, Union
 
+import numpy as np
 from google import genai
 from google.genai.types import GenerateContentConfig, Part, ThinkingConfig
 
 from lattifai.audio2 import AudioData
+from lattifai.caption import Supervision
 from lattifai.config import TranscriptionConfig
 from lattifai.transcription.base import BaseTranscriber
 from lattifai.transcription.prompts import get_prompt_loader
@@ -118,6 +120,130 @@ class GeminiTranscriber(BaseTranscriber):
             self.logger.error(f"Gemini transcription failed: {str(e)}")
             raise RuntimeError(f"Gemini transcription failed: {str(e)}")
 
+    def transcribe_numpy(
+        self,
+        audio: Union[np.ndarray, List[np.ndarray]],
+        language: Optional[str] = None,
+    ) -> Union[Supervision, List[Supervision]]:
+        """
+        Transcribe audio from a numpy array (or list of arrays) and return Supervision.
+
+        Note: Gemini API does not support word-level alignment. The returned
+        Supervision will contain only the full transcription text without alignment.
+
+        Args:
+            audio: Audio data as numpy array (shape: [samples]),
+                   or a list of such arrays for batch processing.
+            language: Optional language code for transcription.
+
+        Returns:
+            Supervision object (or list of Supervision objects) with transcription text (no alignment).
+
+        Raises:
+            ValueError: If API key not provided
+            RuntimeError: If transcription fails
+        """
+        # Handle batch processing
+        if isinstance(audio, list):
+            return [self.transcribe_numpy(arr, language=language) for arr in audio]
+
+        audio_array = audio
+        # Use default sample rate of 16000 Hz
+        sample_rate = 16000
+
+        if self.config.verbose:
+            self.logger.info(f"🎤 Starting Gemini transcription for numpy array (sample_rate={sample_rate})")
+
+        # Ensure audio is in the correct shape
+        if audio_array.ndim == 1:
+            audio_array = audio_array.reshape(1, -1)
+        elif audio_array.ndim > 2:
+            raise ValueError(f"Audio array must be 1D or 2D, got shape {audio_array.shape}")
+
+        # Save numpy array to temporary file
+        import tempfile
+
+        import soundfile as sf
+
+        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp_file:
+            # Transpose to (samples, channels) for soundfile
+            sf.write(tmp_file.name, audio_array.T, sample_rate)
+            tmp_path = Path(tmp_file.name)
+
+        try:
+            # Transcribe using simple ASR prompt
+            import asyncio
+
+            transcript = asyncio.run(self._transcribe_with_simple_prompt(tmp_path, language=language))
+
+            # Create Supervision object from transcript
+            duration = audio_array.shape[-1] / sample_rate
+            supervision = Supervision(
+                id="gemini-transcription",
+                recording_id="numpy-array",
+                start=0.0,
+                duration=duration,
+                text=transcript,
+                speaker=None,
+                alignment=None,  # Gemini does not provide word-level alignment
+            )
+
+            return supervision
+
+        finally:
+            # Clean up temporary file
+            if tmp_path.exists():
+                tmp_path.unlink()
+
+    async def _transcribe_with_simple_prompt(self, media_file: Path, language: Optional[str] = None) -> str:
+        """
+        Transcribe audio using a simple ASR prompt instead of complex instructions.
+
+        Args:
+            media_file: Path to audio file
+            language: Optional language code
+
+        Returns:
+            Transcribed text
+        """
+        client = self._get_client()
+
+        # Upload audio file
+        if self.config.verbose:
+            self.logger.info("📤 Uploading audio file to Gemini...")
+        uploaded_file = client.files.upload(file=str(media_file))
+
+        # Simple ASR prompt
+        system_prompt = "Transcribe the audio."
+        if language:
+            system_prompt = f"Transcribe the audio in {language}."
+
+        # Create simple generation config
+        simple_config = GenerateContentConfig(
+            system_instruction=system_prompt,
+            response_modalities=["TEXT"],
+        )
+
+        contents = Part.from_uri(file_uri=uploaded_file.uri, mime_type=uploaded_file.mime_type)
+        response = await asyncio.get_event_loop().run_in_executor(
+            None,
+            lambda: client.models.generate_content(
+                model=self.config.model_name,
+                contents=contents,
+                config=simple_config,
+            ),
+        )
+
+        if not response.text:
+            raise RuntimeError("Empty response from Gemini API")
+
+        transcript = response.text.strip()
+
+        if self.config.verbose:
+            self.logger.info(f"✅ Transcription completed: {len(transcript)} characters")
+
+        return transcript
+
     def _get_transcription_prompt(self) -> str:
         """Get (and cache) transcription system prompt from prompts module."""
         if self._system_prompt is not None:

lattifai/transcription/lattifai.py

@@ -1,10 +1,12 @@
 """Transcription module with config-driven architecture."""
 
 from pathlib import Path
-from typing import Optional, Union
+from typing import List, Optional, Union
+
+import numpy as np
 
 from lattifai.audio2 import AudioData
-from lattifai.caption import Caption
+from lattifai.caption import Caption, Supervision
 from lattifai.config import TranscriptionConfig
 from lattifai.transcription.base import BaseTranscriber
 from lattifai.transcription.prompts import get_prompt_loader  # noqa: F401
@@ -74,6 +76,32 @@ class LattifAITranscriber(BaseTranscriber):
 
         return caption
 
+    def transcribe_numpy(
+        self,
+        audio: Union[np.ndarray, List[np.ndarray]],
+        language: Optional[str] = None,
+    ) -> Union[Supervision, List[Supervision]]:
+        """
+        Transcribe audio from a numpy array (or list of arrays) and return Supervision.
+
+        Args:
+            audio: Audio data as numpy array (shape: [samples]),
+                   or a list of such arrays for batch processing.
+            language: Optional language code for transcription.
+
+        Returns:
+            Supervision object (or list of Supervision objects) with transcription and alignment info.
+        """
+        if self._transcriber is None:
+            from lattifai_core.transcription import LattifAITranscriber as CoreLattifAITranscriber
+
+            self._transcriber = CoreLattifAITranscriber.from_pretrained(model_config=self.config)
+
+        # Delegate to core transcriber which handles both single arrays and lists
+        return self._transcriber.transcribe(
+            audio, language=language, return_hypotheses=True, progress_bar=False, timestamps=True
+        )[0]
+
     def write(
         self, transcript: Caption, output_file: Path, encoding: str = "utf-8", cache_audio_events: bool = True
     ) -> Path:

lattifai/utils.py

@@ -1,10 +1,9 @@
 """Shared utility helpers for the LattifAI SDK."""
 
-import os
 import sys
 from datetime import datetime, timedelta
 from pathlib import Path
-from typing import Any, Optional, Type
+from typing import Optional
 
 from lattifai.errors import ModelLoadError
 
@@ -88,42 +87,74 @@ def _create_cache_marker(cache_dir: Path) -> None:
     marker_path.touch()
 
 
-def _resolve_model_path(model_name_or_path: str) -> str:
-    """Resolve model path, downloading from Hugging Face when necessary."""
+def _resolve_model_path(model_name_or_path: str, model_hub: str = "huggingface") -> str:
+    """Resolve model path, downloading from the specified model hub when necessary.
+
+    Args:
+        model_name_or_path: Local path or remote model identifier.
+        model_hub: Which hub to use for downloads. Supported: "huggingface", "modelscope".
+    """
     if Path(model_name_or_path).expanduser().exists():
         return str(Path(model_name_or_path).expanduser())
 
-    from huggingface_hub import snapshot_download
-    from huggingface_hub.constants import HF_HUB_CACHE
-    from huggingface_hub.errors import LocalEntryNotFoundError
+    # Normalize hub name
+    hub = (model_hub or "huggingface").lower()
 
-    # Determine cache directory for this model
-    cache_dir = Path(HF_HUB_CACHE) / f'models--{model_name_or_path.replace("/", "--")}'
+    if hub not in ("huggingface", "modelscope"):
+        raise ValueError(f"Unsupported model_hub: {model_hub}. Supported: 'huggingface', 'modelscope'.")
 
-    # Check if we have a valid cached version
-    if _is_cache_valid(cache_dir):
-        # Return the snapshot path (latest version)
-        snapshots_dir = cache_dir / "snapshots"
-        if snapshots_dir.exists():
-            snapshot_dirs = [d for d in snapshots_dir.iterdir() if d.is_dir()]
-            if snapshot_dirs:
-                # Return the most recent snapshot
-                latest_snapshot = max(snapshot_dirs, key=lambda p: p.stat().st_mtime)
-                return str(latest_snapshot)
+    # If local path exists, return it regardless of hub
+    if Path(model_name_or_path).expanduser().exists():
+        return str(Path(model_name_or_path).expanduser())
+
+    if hub == "huggingface":
+        from huggingface_hub import snapshot_download
+        from huggingface_hub.constants import HF_HUB_CACHE
+        from huggingface_hub.errors import LocalEntryNotFoundError
+
+        # Determine cache directory for this model
+        cache_dir = Path(HF_HUB_CACHE) / f'models--{model_name_or_path.replace("/", "--")}'
+
+        # Check if we have a valid cached version
+        if _is_cache_valid(cache_dir):
+            # Return the snapshot path (latest version)
+            snapshots_dir = cache_dir / "snapshots"
+            if snapshots_dir.exists():
+                snapshot_dirs = [d for d in snapshots_dir.iterdir() if d.is_dir()]
+                if snapshot_dirs:
+                    # Return the most recent snapshot
+                    latest_snapshot = max(snapshot_dirs, key=lambda p: p.stat().st_mtime)
+                    return str(latest_snapshot)
 
-    try:
-        downloaded_path = snapshot_download(repo_id=model_name_or_path, repo_type="model")
-        _create_cache_marker(cache_dir)
-        return downloaded_path
-    except LocalEntryNotFoundError:
         try:
-            os.environ["HF_ENDPOINT"] = "https://hf-mirror.com"
             downloaded_path = snapshot_download(repo_id=model_name_or_path, repo_type="model")
             _create_cache_marker(cache_dir)
             return downloaded_path
-        except Exception as e:  # pragma: no cover - bubble up for caller context
-            raise ModelLoadError(model_name_or_path, original_error=e)
-    except Exception as e:  # pragma: no cover - unexpected download issue
+        except LocalEntryNotFoundError:
+            # Fall back to modelscope if HF entry not found
+            try:
+                from modelscope.hub.snapshot_download import snapshot_download as ms_snapshot
+
+                downloaded_path = ms_snapshot(model_name_or_path)
+                return downloaded_path
+            except Exception as e:  # pragma: no cover - bubble up for caller context
+                raise ModelLoadError(model_name_or_path, original_error=e)
+        except Exception as e:  # pragma: no cover - unexpected download issue
+            import colorful
+
+            print(colorful.red | f"Error downloading from Hugging Face Hub: {e}. Trying ModelScope...")
+            from modelscope.hub.snapshot_download import snapshot_download as ms_snapshot
+
+            downloaded_path = ms_snapshot(model_name_or_path)
+            return downloaded_path
+
+    # modelscope path
+    from modelscope.hub.snapshot_download import snapshot_download as ms_snapshot
+
+    try:
+        downloaded_path = ms_snapshot(model_name_or_path)
+        return downloaded_path
+    except Exception as e:  # pragma: no cover
         raise ModelLoadError(model_name_or_path, original_error=e)
 
 

{lattifai-1.0.5.dist-info → lattifai-1.1.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lattifai
-Version: 1.0.5
+Version: 1.1.0
 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>
@@ -50,7 +50,7 @@ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: <3.15,>=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: lattifai-core>=0.5.1
+Requires-Dist: lattifai-core>=0.6.0
 Requires-Dist: lattifai-run>=1.0.1
 Requires-Dist: python-dotenv
 Requires-Dist: lhotse>=1.26.0
@@ -61,11 +61,12 @@ Requires-Dist: tgt
 Requires-Dist: onnx>=1.16.0
 Requires-Dist: onnxruntime
 Requires-Dist: msgpack
+Requires-Dist: scipy!=1.16.3
 Requires-Dist: g2p-phonemizer>=0.4.0
 Requires-Dist: av
 Requires-Dist: wtpsplit>=2.1.7
 Requires-Dist: OmniSenseVoice>=0.4.2
-Requires-Dist: nemo_toolkit_asr[asr]>=2.7.0rc3
+Requires-Dist: nemo_toolkit_asr[asr]>=2.7.0rc4
 Requires-Dist: pyannote-audio-notorchdeps>=4.0.2
 Requires-Dist: questionary>=2.0
 Requires-Dist: yt-dlp
@@ -113,9 +114,6 @@ Dynamic: license-file
 
 Advanced forced alignment and subtitle generation powered by [ 🤗 Lattice-1](https://huggingface.co/Lattifai/Lattice-1) model.
 
-> **⚠️ Note on Current Limitations**:
-> 1. **Memory Usage**: We are aware of high memory consumption and are actively working on further optimizations.
-
 ## Table of Contents
 
 - [Installation](#installation)
@@ -136,7 +134,7 @@ Advanced forced alignment and subtitle generation powered by [ 🤗 Lattice-1](h
 - [Advanced Features](#advanced-features)
   - [Word-Level Alignment](#word-level-alignment)
   - [Smart Sentence Splitting](#smart-sentence-splitting)
-  - [Speaker Diarization](#speaker-diarization-wip)
+  - [Speaker Diarization](#speaker-diarization)
   - [YAML Configuration Files](#yaml-configuration-files)
 - [Supported Formats](#supported-formats)
 - [Roadmap](#roadmap)
@@ -708,9 +706,7 @@ caption = client.alignment(
 )
 ```
 
-### Speaker Diarization (WIP)
-
-**Note:** This feature is currently under development and not yet fully available.
+### Speaker Diarization
 
 Speaker diarization automatically identifies and labels different speakers in audio. When enabled, the system will:
 - Detect speaker changes in the audio
@@ -721,6 +717,24 @@ Speaker diarization automatically identifies and labels different speakers in au
 - **Existing speaker labels in subtitles**: If your input captions already contain speaker names (e.g., `[Alice]`, `>> Bob:`, or `SPEAKER_01:`), the system will preserve them as much as possible during alignment
 - **Gemini Transcriber**: When using Gemini models for transcription (e.g., `gemini-2.5-pro`), the model can intelligently identify and extract speaker names from dialogue context, making it easier to generate speaker-aware transcripts
 
+**CLI:**
+```bash
+# Enable speaker diarization during alignment
+lai alignment align audio.wav subtitle.srt output.srt \
+    diarization.enabled=true
+
+# With additional diarization settings
+lai alignment align audio.wav subtitle.srt output.srt \
+    diarization.enabled=true \
+    diarization.device=cuda \
+    diarization.min_speakers=2 \
+    diarization.max_speakers=4
+
+# For YouTube videos with diarization
+lai alignment youtube "https://youtube.com/watch?v=VIDEO_ID" \
+    diarization.enabled=true
+```
+
 **Python SDK:**
 ```python
 from lattifai import LattifAI, DiarizationConfig

{lattifai-1.0.5.dist-info → lattifai-1.1.0.dist-info}/RECORD

@@ -1,44 +1,45 @@
 lattifai/__init__.py,sha256=7y1R5IGw0Sgvl1tfqxEK7e-ozW0wVB-q_JZgv6YyrMQ,2751
 lattifai/audio2.py,sha256=BKMCzkuEmBFAWOEnzgLxeK8TBPTFbjzr1esOfe3MQoo,17460
-lattifai/client.py,sha256=OXDGsWVeOMEjmXI795pvnK3L-ZLn_sfUwG0i7uJ1JkY,22492
+lattifai/client.py,sha256=yD8yIcw3xfvPqBzUQTJj5h9sFYbg4BSHBR5gwO-wGF8,18875
 lattifai/errors.py,sha256=LyWRGVhQ6Ak2CYn9FBYAPRgQ7_VHpxzNsXI31HXD--s,11291
 lattifai/logging.py,sha256=MbUEeOUFlF92pA9v532DiPPWKl03S7UHCJ6Z652cf0w,2860
-lattifai/mixin.py,sha256=yj3H1SSQSQrhUeqKhQmRRELRr5fp2mb2ovkK9p8Vwn4,23858
+lattifai/mixin.py,sha256=t8A3J1DVr7SBEv2FYK3JrGEFqxAV3NktfGPiOQdxTB4,24990
 lattifai/types.py,sha256=SjYBfwrCBOXlICvH04niFQJ7OzTx7oTaa_npfRkB67U,659
-lattifai/utils.py,sha256=ZYEUaoTBCwzv4PBBD-woeiDSTx8T1a1vXHIT0g1YmRI,5345
+lattifai/utils.py,sha256=N4k2cvoT2CgHsGY0sb-3l-eKUMLzGirPzVa53v8euh4,6658
 lattifai/alignment/__init__.py,sha256=ehpkKfjNIYUx7_M-RWD_8Efcrzd9bE-NSm0QgMMVLW0,178
-lattifai/alignment/lattice1_aligner.py,sha256=DpN_it7ETZgz6uH3I90Y926bvjhFRdL6dycxz5S_tkI,5142
-lattifai/alignment/lattice1_worker.py,sha256=1yYK_xLOL_xHZTVGgNb957R7HhHnl6xwrXUcN372ZIY,12407
+lattifai/alignment/lattice1_aligner.py,sha256=5c_bWv0v6jOX8H4gO3-mAKXn4Ux4vjHIUtAY4d8yMak,6110
+lattifai/alignment/lattice1_worker.py,sha256=U_EbDsbisWWKIyfxtx9q4deAyml_hni8DpznZo5zBko,12975
 lattifai/alignment/phonemizer.py,sha256=fbhN2DOl39lW4nQWKzyUUTMUabg7v61lB1kj8SKK-Sw,1761
 lattifai/alignment/segmenter.py,sha256=mzWEQC6hWZtI2mR2WU59W7qLHa7KXy7fdU6991kyUuQ,6276
-lattifai/alignment/tokenizer.py,sha256=oqgy5L9wU0_AMyUVNArEtPIDXm7WdvNNfJuB2ZJBpqI,22394
+lattifai/alignment/tokenizer.py,sha256=GNLZbkvZ066PJGUznJVgxZUwSzslD6mz8YsI2Cry6RI,22400
 lattifai/caption/__init__.py,sha256=6MM_2j6CaqwZ81LfSy4di2EP0ykvheRjMZKAYDx2rQs,477
-lattifai/caption/caption.py,sha256=Ljt-6K89AauIK05hdDqjV6G03mkTTJL2UE9ukt-tck0,52502
+lattifai/caption/caption.py,sha256=mZYobxuZ8tkJUkZMVvRTrNeGTdmIZYSXTEySQdaGQd8,54595
 lattifai/caption/gemini_reader.py,sha256=GqY2w78xGYCMDP5kD5WGS8jK0gntel2SK-EPpPKTrwU,15138
 lattifai/caption/gemini_writer.py,sha256=sYPxYEmVQcEan5WVGgSrcraxs3QJRQRh8CJkl2yUQ1s,6515
 lattifai/caption/supervision.py,sha256=DRrM8lfKU_x9aVBcLG6xnT0xIJrnc8jzHpzcSwQOg8c,905
 lattifai/caption/text_parser.py,sha256=XDb8KTt031uJ1hg6dpbINglGOTX-6pBcghbg3DULM1I,4633
-lattifai/cli/__init__.py,sha256=dIUmrpN-OwR4h6BqMhXp87_5ZwgO41ShPru_iZGnpQs,463
+lattifai/cli/__init__.py,sha256=LafsAf8YfDcfTeJ1IevFcyLm-mNbxpOOnm33OFKtpDM,523
 lattifai/cli/alignment.py,sha256=06em-Uaf6NhSz1ce4dwT2r8n56NrtibR7ZsSkmc18Kc,5954
 lattifai/cli/app_installer.py,sha256=gAndH3Yo97fGRDe2CQnGtOgZZ4k3_v5ftcUo5g6xbSA,5884
-lattifai/cli/caption.py,sha256=p0VY6orf3D77tr30NQka7A84kwEmYiZrCDB6FbTgoFM,6312
+lattifai/cli/caption.py,sha256=4qQ9DFhxcfaeFMY0TB5I42x4W_gOo2zY6kjXnHnFDms,6313
+lattifai/cli/diarization.py,sha256=cDz1p6RUC-ySPzzGYHWff0L6EWHTUPrci7DaVxwZrVc,3933
 lattifai/cli/server.py,sha256=sXMfOSse9-V79slXUU8FDLeqtI5U9zeU-5YpjTIGyVw,1186
-lattifai/cli/transcribe.py,sha256=W42SVhnOQ0EndMk-Lu38BiG1LuMcJnzre9X83M6kBZ4,8137
+lattifai/cli/transcribe.py,sha256=_vHzrdaGiPepQWATqvEDYDjwzfVLAd2i8RjOLkvdb0w,8218
 lattifai/cli/youtube.py,sha256=-EIDSS1Iel3_6qD9M2CZZHwKOvgdkIa1cMY4rX7xwVo,5331
 lattifai/config/__init__.py,sha256=Z8OudvS6fgfLNLu_2fvoXartQiYCECOnNfzDt-PfCN4,543
-lattifai/config/alignment.py,sha256=v6SuryAVNET9hgH_ZidYN2QhZqpEDnNhR-rogSSSfAg,4039
+lattifai/config/alignment.py,sha256=vLiH150YWvBUiVkFOIO-nPXCB-b8fP9iSZgS79k1Qbg,4586
 lattifai/config/caption.py,sha256=AYOyUJ1xZsX8CBZy3GpLitbcCAHcZ9LwXui_v3vtuso,6786
 lattifai/config/client.py,sha256=I1JqLQlsQNU5ouovTumr-PP_8GWC9DI_e9B5UwsDZws,1492
 lattifai/config/diarization.py,sha256=cIkwCfsYqfMns3i6tKWcwBBBkdnhhmB_Eo0TuOPCw9o,2484
 lattifai/config/media.py,sha256=cjM8eGeZ7ELhmy4cCqHAyogeHItaVqMrPzSwwIx79HY,14856
-lattifai/config/transcription.py,sha256=bzghOGgcNWzTnDYd_cqCOB7GT8OnzHDiyam7LSixqxM,2901
+lattifai/config/transcription.py,sha256=_gPJD6cob_jWNdf841nBHhAqJGCxS6PfSyvx2W_vPcM,3082
 lattifai/diarization/__init__.py,sha256=MgBDQ1ehL2qDnZprEp8KqON7CmbG-qaP37gzBsV0jzk,119
-lattifai/diarization/lattifai.py,sha256=SE2BpIZ3_deKyhXdBqe77bsDLXIUV9AQV34gfINv7_s,2657
+lattifai/diarization/lattifai.py,sha256=tCnFL6ywITqeKR8YoCsYvyJxNoIwoC6GsnI9zkXNB-Q,3128
 lattifai/server/app.py,sha256=wXYgXc_yGQACtUJdhkfhLsTOQjhhIhDQRiVRny7Ogcs,15455
-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
-lattifai/transcription/lattifai.py,sha256=h0nhXST0qljhyndf80IEddM7Y_N1jiS28YoaE536eME,3483
+lattifai/transcription/__init__.py,sha256=vMHciyCEPKhhfM3KjMCeDqnyxU1oghF8g5o5SvpnT_4,2669
+lattifai/transcription/base.py,sha256=v_b1_JGYiBqeMmwns0wHCJ7UOm6j9k-76Uzbr-qmzrs,4467
+lattifai/transcription/gemini.py,sha256=LJSQt9nGqQdEG6ZFXoHWltumyMEM7-Ezy8ss0iPJb7k,12414
+lattifai/transcription/lattifai.py,sha256=EKEdCafgdRWKw_084eD07BqGh2_D-qo3ig3H5X3XYGg,4621
 lattifai/transcription/prompts/README.md,sha256=X49KWSQVdjWxxWUp4R2w3ZqKrAOi6_kDNHh1hMaQ4PE,694
 lattifai/transcription/prompts/__init__.py,sha256=G9b42COaCYv3sPPNkHsGDLOMBuVGKt4mXGYal_BYtYQ,1351
 lattifai/transcription/prompts/gemini/README.md,sha256=rt7f7yDGtaobKBo95LG3u56mqa3ABOXQd0UVgJYtYuo,781
@@ -48,9 +49,9 @@ lattifai/workflow/agents.py,sha256=yEOnxnhcTvr1iOhCorNvp8B76P6nQsLRXJCu_rCYFfM,3
 lattifai/workflow/base.py,sha256=8QoVIBZwJfr5mppJbtUFafHv5QR9lL-XrULjTWD0oBg,6257
 lattifai/workflow/file_manager.py,sha256=IUWW838ta83kfwM4gpW83gsD_Tx-pa-L_RWKjiefQbQ,33017
 lattifai/workflow/youtube.py,sha256=ON9z0UUk16ThQzdhdgyOiwBmewZOcxfT05dsl3aKYqw,23840
-lattifai-1.0.5.dist-info/licenses/LICENSE,sha256=xGMLmdFJy6Jkz3Hd0znyQLmcxC93FSZB5isKnEDMoQQ,1066
-lattifai-1.0.5.dist-info/METADATA,sha256=cTg6ivcixFAv-464qk0R2v19LdEgGkETcNvRzycFSKk,26117
-lattifai-1.0.5.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
-lattifai-1.0.5.dist-info/entry_points.txt,sha256=F8Akof3VtKtrbnYSav1umgoo9Xbv34rUcKn-ioRfeGQ,474
-lattifai-1.0.5.dist-info/top_level.txt,sha256=tHSoXF26r-IGfbIP_JoYATqbmf14h5NrnNJGH4j5reI,9
-lattifai-1.0.5.dist-info/RECORD,,
+lattifai-1.1.0.dist-info/licenses/LICENSE,sha256=xGMLmdFJy6Jkz3Hd0znyQLmcxC93FSZB5isKnEDMoQQ,1066
+lattifai-1.1.0.dist-info/METADATA,sha256=MXg4IXWA38Y9c_RT9S_6NnCZFc_4-Yic_zZDHlS1TeY,26400
+lattifai-1.1.0.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
+lattifai-1.1.0.dist-info/entry_points.txt,sha256=nHZri2VQkPYEl0tQ0dMYTpVGlCOgVWlDG_JtDR3QXF8,545
+lattifai-1.1.0.dist-info/top_level.txt,sha256=tHSoXF26r-IGfbIP_JoYATqbmf14h5NrnNJGH4j5reI,9
+lattifai-1.1.0.dist-info/RECORD,,

{lattifai-1.0.5.dist-info → lattifai-1.1.0.dist-info}/entry_points.txt

@@ -1,6 +1,7 @@
 [console_scripts]
 lai-align = lattifai.cli.alignment:main
 lai-app-install = lattifai.cli.app_installer:main
+lai-diarize = lattifai.cli.diarization:main
 lai-server = lattifai.cli.server:main
 lai-transcribe = lattifai.cli.transcribe:main
 lai-youtube = lattifai.cli.youtube:main
@@ -11,4 +12,5 @@ laicap-shift = lattifai.cli.caption:main_shift
 [lai_run.cli]
 alignment = lattifai.cli
 caption = lattifai.cli
+diarization = lattifai.cli
 transcribe = lattifai.cli