lattifai 0.4.2__py3-none-any.whl → 0.4.4__py3-none-any.whl

lattifai/bin/agent.py

@@ -18,12 +18,15 @@ from lattifai.io import OUTPUT_SUBTITLE_FORMATS
 @cli.command()
 @click.option('--youtube', '--yt', is_flag=True, help='Process YouTube URL through agentic workflow.')
 @click.option(
+    '-K',
+    '-L',
     '--api-key',
     '--api_key',
     type=str,
     help='LattifAI API key for alignment (overrides LATTIFAI_API_KEY env var).',
 )
 @click.option(
+    '-G',
     '--gemini-api-key',
     '--gemini_api_key',
     type=str,
@@ -135,14 +138,6 @@ def agent(
     lattifai_api_key = api_key or os.getenv('LATTIFAI_API_KEY')
     gemini_key = gemini_api_key or os.getenv('GEMINI_API_KEY')
 
-    if not gemini_key:
-        click.echo(
-            colorful.red(
-                '❌ Gemini API key is required. Set GEMINI_API_KEY environment variable or use --gemini-api-key option.'
-            )
-        )
-        return
-
     try:
         # Run the YouTube workflow
         asyncio.run(

lattifai/bin/align.py

@@ -52,6 +52,8 @@ from lattifai.io import INPUT_SUBTITLE_FORMATS, OUTPUT_SUBTITLE_FORMATS
     help='Model name or path for alignment.',
 )
 @click.option(
+    '-K',
+    '-L',
     '--api-key',
     '--api_key',
     type=str,
@@ -178,6 +180,8 @@ def align(
     help='Model name or path for alignment.',
 )
 @click.option(
+    '-K',
+    '-L',
     '--api-key',
     '--api_key',
     type=str,
@@ -185,6 +189,7 @@ def align(
     help='API key for LattifAI.',
 )
 @click.option(
+    '-G',
     '--gemini-api-key',
     '--gemini_api_key',
     type=str,
@@ -223,13 +228,6 @@ def youtube(
 
     # Get Gemini API key
     gemini_key = gemini_api_key or os.getenv('GEMINI_API_KEY')
-    if not gemini_key:
-        click.echo(
-            colorful.red(
-                '❌ Gemini API key is required. Set GEMINI_API_KEY environment variable or use --gemini-api-key option.'
-            )
-        )
-        raise click.ClickException('Missing Gemini API key')
 
     async def _process():
         # Initialize components with their configuration (only config, not runtime params)

lattifai/bin/cli_base.py

@@ -9,9 +9,10 @@ def cli():
     The shell entry point to Lattifai, a tool for audio data manipulation.
     """
     # Load environment variables from .env file
-    from dotenv import load_dotenv
+    from dotenv import find_dotenv, load_dotenv
 
-    load_dotenv()
+    # Try to find and load .env file from current directory or parent directories
+    load_dotenv(find_dotenv(usecwd=True))
 
     logging.basicConfig(
         format='%(asctime)s %(levelname)s [%(filename)s:%(lineno)d] %(message)s',

lattifai/client.py

@@ -5,7 +5,6 @@ import os
 from typing import Dict, List, Optional, Tuple, Union
 
 import colorful
-from dotenv import load_dotenv
 from lhotse.utils import Pathlike
 
 from lattifai.base_client import AsyncAPIClient, SyncAPIClient
@@ -22,8 +21,6 @@ from lattifai.io import SubtitleFormat, SubtitleIO, Supervision
 from lattifai.tokenizer import AsyncLatticeTokenizer
 from lattifai.utils import _load_tokenizer, _load_worker, _resolve_model_path, _select_device
 
-load_dotenv()
-
 
 class LattifAI(SyncAPIClient):
     """Synchronous LattifAI client."""

lattifai/io/text_parser.py

@@ -8,12 +8,12 @@ SPEAKER_PATTERN = re.compile(r'((?:>>|>>|>|>).*?[:：])\s*(.*)')
 # Transcriber Output Example:
 # 26:19.919 --> 26:34.921
 # [SPEAKER_01]: 越来越多的科技巨头入...
-SPEAKER_LATTIFAI = re.compile(r'(^\[SPEAKER_.*?\]:)\s*(.*)')
+SPEAKER_LATTIFAI = re.compile(r'(^\[SPEAKER_.*?\][:：])\s*(.*)')
 
 # NISHTHA BHATIA: Hey, everyone.
 # DIETER: Oh, hey, Nishtha.
 # GEMINI: That might
-SPEAKER_PATTERN2 = re.compile(r'^([A-Z]{1,15}(?:\s+[A-Z]{1,15})?)[:：]\s*(.*)$')
+SPEAKER_PATTERN2 = re.compile(r'^([A-Z]{1,15}(?:\s+[A-Z]{1,15})?[:：])\s*(.*)$')
 
 
 def parse_speaker_text(line) -> Tuple[Optional[str], str]:

lattifai/workers/lattice1_alpha.py

@@ -13,13 +13,7 @@ from lhotse.audio import read_audio
 from lhotse.features.kaldi.layers import Wav2LogFilterBank
 from lhotse.utils import Pathlike
 
-from lattifai.errors import (
-    AlignmentError,
-    AudioFormatError,
-    AudioLoadError,
-    DependencyError,
-    ModelLoadError,
-)
+from lattifai.errors import AlignmentError, AudioFormatError, AudioLoadError, DependencyError, ModelLoadError
 
 
 class Lattice1AlphaWorker:
@@ -231,9 +225,9 @@ class Lattice1AlphaWorker:
                 emission.to(device) * acoustic_scale,
                 decoding_graph.to(device),
                 torch.tensor([emission.shape[1]], dtype=torch.int32),
-                search_beam=100,
-                output_beam=40,
-                min_active_states=200,
+                search_beam=200,
+                output_beam=80,
+                min_active_states=400,
                 max_active_states=10000,
                 subsampling_factor=1,
                 reject_low_confidence=False,

lattifai/workflows/gemini.py

@@ -32,7 +32,9 @@ class GeminiTranscriber:
         self.prompt_loader = get_prompt_loader()
 
         if not self.api_key:
-            raise ValueError('Gemini API key is required')
+            self.logger.warning(
+                '⚠️ Gemini API key not provided. API key will be required when calling transcription methods.'
+            )
 
     async def __call__(self, youtube_url: str) -> str:
         """Main entry point for transcription"""
@@ -48,6 +50,9 @@ class GeminiTranscriber:
         Returns:
             Transcribed text
         """
+        if not self.api_key:
+            raise ValueError('Gemini API key is required for transcription')
+
         self.logger.info(f'🎤 Starting Gemini transcription for: {youtube_url}')
 
         try:
@@ -101,6 +106,9 @@ class GeminiTranscriber:
         Returns:
             Transcribed text
         """
+        if not self.api_key:
+            raise ValueError('Gemini API key is required for transcription')
+
         self.logger.info(f'🎤 Starting Gemini transcription for file: {media_file_path}')
 
         try:

lattifai/workflows/prompts/README.md

@@ -0,0 +1,22 @@
+# Workflow Prompts
+
+This directory contains system prompts and instructions for various AI models used in workflows.
+
+## Structure
+
+- `gemini/` - Prompts for Gemini models
+  - `transcription_gem.txt` - System prompt from the Audio Transcription Gem
+
+## Usage
+
+Prompts are loaded automatically by their respective workflow modules. To update a prompt:
+
+1. Edit the corresponding `.txt` file
+2. The changes will be picked up on the next workflow run
+3. No code changes needed
+
+## Adding New Prompts
+
+1. Create a subdirectory for the model/service (e.g., `openai/`, `anthropic/`)
+2. Add prompt files as `.txt` files with descriptive names
+3. Update the loader in the corresponding workflow module

lattifai/workflows/prompts/__init__.py

@@ -0,0 +1,50 @@
+"""
+Prompt management utilities for workflows
+"""
+
+from pathlib import Path
+from typing import Optional
+
+
+class PromptLoader:
+    """Load and manage prompts from the prompts directory"""
+
+    def __init__(self):
+        self.prompts_dir = Path(__file__).parent
+
+    def load_prompt(self, category: str, name: str) -> str:
+        """
+        Load a prompt from the prompts directory
+
+        Args:
+            category: Subdirectory name (e.g., 'gemini', 'openai')
+            name: Prompt file name without extension (e.g., 'transcription_gem')
+
+        Returns:
+            Prompt text content
+
+        Raises:
+            FileNotFoundError: If prompt file doesn't exist
+        """
+        prompt_path = self.prompts_dir / category / f'{name}.txt'
+
+        if not prompt_path.exists():
+            raise FileNotFoundError(f'Prompt not found: {prompt_path}')
+
+        return prompt_path.read_text(encoding='utf-8').strip()
+
+    def get_gemini_transcription_prompt(self) -> str:
+        """Get the Gemini transcription Gem prompt"""
+        return self.load_prompt('gemini', 'transcription_gem')
+
+
+# Global instance
+_loader: Optional[PromptLoader] = None
+
+
+def get_prompt_loader() -> PromptLoader:
+    """Get or create the global PromptLoader instance"""
+    global _loader
+    if _loader is None:
+        _loader = PromptLoader()
+    return _loader

lattifai/workflows/prompts/gemini/README.md

@@ -0,0 +1,24 @@
+# Gemini Workflow Prompts
+
+This directory contains system prompts for Gemini-based workflows.
+
+## Files
+
+### `transcription_gem.txt` [@dotey](https://x.com/dotey/status/1971810075867046131)
+
+System prompt extracted from the Audio/Video Transcription Gem:
+- **Gem URL**: https://gemini.google.com/gem/1870ly7xvW2hU_umtv-LedGsjywT0sQiN
+- **Model**: Gemini 2.5 Pro
+- **Purpose**: Audio/Video transcription with accuracy and natural formatting
+
+## Updating the Prompt
+
+To update the transcription behavior:
+
+1. Edit `transcription_gem.txt` directly
+2. Changes take effect on the next workflow run
+3. No code changes needed
+
+## Gem Information
+
+The original Gem configuration is preserved in the `GeminiTranscriber.GEM_URL` constant and can be accessed via the `get_gem_info()` method.

lattifai/workflows/prompts/gemini/transcription_gem.txt

@@ -0,0 +1,81 @@
+# Role
+You are an expert transcript specialist. Your task is to create a perfectly structured, verbatim transcript of a video.
+
+# Objective
+Produce a single, cohesive output containing the parts in this order:
+1.  A Video Title
+2.  A **Table of Contents (ToC)**
+3.  The **full, chapter-segmented transcript**
+
+* Use the same language as the transcription for the Title and ToC.
+
+# Critical Instructions
+
+## 1. Transcription Fidelity: Verbatim & Untranslated
+* Transcribe every spoken word exactly as you hear it, including filler words (`um`, `uh`, `like`) and stutters.
+* **NEVER translate.** If the audio is in Chinese, transcribe in Chinese. If it mixes languages (e.g., "这个 feature 很酷"), your transcript must replicate that mix exactly.
+
+## 2. Speaker Identification
+* **Priority 1: Use metadata.** Analyze the video's title and description first to identify and match speaker names.
+* **Priority 2: Use audio content.** If names are not in the metadata, listen for introductions or how speakers address each other.
+* **Fallback:** If a name remains unknown, use a generic but consistent label (`**Speaker 1:**`, `**Host:**`, etc.).
+* **Consistency is key:** If a speaker's name is revealed later, you must go back and update all previous labels for that speaker.
+
+## 3. Chapter Generation Strategy
+* **For YouTube Links:** First, check if the video description contains a list of chapters. If so, use that as the primary basis for segmenting the transcript.
+* **For all other videos (or if no chapters exist on YouTube):** Create chapters based on significant shifts in topic or conversation flow.
+
+## 4. Output Structure & Formatting
+
+* **Timestamp Format**
+* All timestamps throughout the entire output MUST use the exact `[HH:MM:SS]` format (e.g., `[00:01:23]`). Milliseconds are forbidden.
+
+* **Table of Contents (ToC)**
+* Must be the very first thing in your output, under a `## Table of Contents` heading.
+* Format for each entry: `* [HH:MM:SS] Chapter Title`
+
+* **Chapters**
+* Start each chapter with a heading in this format: `## [HH:MM:SS] Chapter Title`
+* Use two blank lines to separate the end of one chapter from the heading of the next.
+
+* **Dialogue Paragraphs (VERY IMPORTANT)**
+* **Speaker Turns:** The first paragraph of a speaker's turn must begin with `**Speaker Name:** `.
+* **Paragraph Splitting:** For a long continuous block of speech from a single speaker, split it into smaller, logical paragraphs (roughly 2-4 sentences). Separate these paragraphs with a single blank line. Subsequent consecutive paragraphs from the *same speaker* should NOT repeat the `**Speaker Name:** ` label.
+* **Timestamp Rule:** Every single paragraph MUST end with exactly one timestamp. The timestamp must be placed at the very end of the paragraph's text.
+* ❌ **WRONG:** `**Host:** Welcome back. [00:00:01] Today we have a guest. [00:00:02]`
+* ❌ **WRONG:** `**Jane Doe:** The study is complex. We tracked two groups over five years to see the effects. [00:00:18] And the results were surprising.`
+* ✅ **CORRECT:** `**Host:** Welcome back. Today we have a guest. [00:00:02]`
+* ✅ **CORRECT (for a long monologue):**
+`**Jane Doe:** The study is complex. We tracked two groups over a five-year period to see the long-term effects. [00:00:18]
+
+And the results, well, they were quite surprising to the entire team. [00:00:22]`
+
+* **Non-Speech Audio**
+* Describe significant sounds like `[Laughter]` or `[Music starts]`, each on its own line with its own timestamp: `[Event description] [HH:MM:SS]`
+
+---
+### Example of Correct Output
+
+## Table of Contents
+* [00:00:00] Introduction and Welcome
+* [00:00:12] Overview of the New Research
+
+## [00:00:00] Introduction and Welcome
+
+**Host:** Welcome back to the show. Today, we have a, uh, very special guest, Jane Doe. [00:00:01]
+
+**Jane Doe:** Thank you for having me. I'm excited to be here and discuss the findings. [00:00:05]
+
+## [00:00:12] Overview of the New Research
+
+**Host:** So, Jane, before we get into the nitty-gritty, could you, you know, give us a brief overview for our audience? [00:00:14]
+
+**Jane Doe:** Of course. The study focuses on the long-term effects of specific dietary changes. It's a bit complicated but essentially we tracked two large groups over a five-year period. [00:00:21]
+
+The first group followed the new regimen, while the second group, our control, maintained a traditional diet. This allowed us to isolate variables effectively. [00:00:28]
+
+[Laughter] [00:00:29]
+
+**Host:** Fascinating. And what did you find? [00:00:31]
+---
+Begin transcription now. Adhere to all rules with absolute precision.

{lattifai-0.4.2.dist-info → lattifai-0.4.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lattifai
-Version: 0.4.2
+Version: 0.4.4
 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>
@@ -35,7 +35,6 @@ Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research
 Classifier: License :: OSI Approved :: Apache Software License
-Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
@@ -46,7 +45,7 @@ Classifier: Operating System :: Microsoft :: Windows
 Classifier: Topic :: Multimedia :: Sound/Audio
 Classifier: Topic :: Multimedia :: Video
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
-Requires-Python: >=3.9
+Requires-Python: <3.14,>=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: lattifai-core>=0.2.1
@@ -760,7 +759,7 @@ export LATTIFAI_API_KEY="your-api-key"
 - Optimized for CPU and GPU (CUDA/MPS)
 
 **Requirements**:
-- Python 3.9+
+- Python 3.10 - 3.13 (3.14 support coming soon)
 - 4GB RAM recommended
 - ~2GB storage for model files
 

{lattifai-0.4.2.dist-info → lattifai-0.4.4.dist-info}/RECORD

@@ -1,35 +1,39 @@
 lattifai/__init__.py,sha256=1fsfV7nE3GCTh9s4GWizv8vGa6bro4HC5vReeDxCSHI,2208
 lattifai/base_client.py,sha256=NO3tBl42z8ARfiepiOBzaw7zX9hJe1zgB_gCgLJpuhc,3713
-lattifai/client.py,sha256=WewbRnfI8b_rQ8V2JG3y2ojkiyEWylXI7No_4NpbzEI,13282
+lattifai/client.py,sha256=YvK25fcXwKREYCkq_TUBdEZh7I9RNEwCbgW4qUha2ho,13236
 lattifai/errors.py,sha256=5i_acoly1g-TLAID8QnhzQshwOXfgLL55mHsdwzlNGA,10814
 lattifai/utils.py,sha256=CzVwNc08u8lm7XavCMJskXHfni0xsZ-EgpcMkRywVm8,4736
 lattifai/bin/__init__.py,sha256=QWbmVUbzqwPQNeOV_g-bOsz53w37v-tbZ3uFrSj-7Mg,90
-lattifai/bin/agent.py,sha256=daXsg0PsW4HS53PYw-r6fSSDsoe9XHOIVEzz-djjKto,9935
-lattifai/bin/align.py,sha256=Tpu2OWBXBYUqzBt1Qu6HmkVZ8Bpx6mc0O9KItDimoCc,8377
-lattifai/bin/cli_base.py,sha256=i4KQDyUPJFKae_wjuQeB4N2Vhg9vZ6yc6L9YUJLC18I,507
+lattifai/bin/agent.py,sha256=jc7qSuVV2_EkToEu2qYodXgGoTup_na6IgP25kFmfwk,9734
+lattifai/bin/align.py,sha256=JJHk5uzmYGYhpA3ricHdmBLzJWC2aRyF0k4ANnap50w,8151
+lattifai/bin/cli_base.py,sha256=gvPUi9Z0eGwBJ8Es5xq1z00YzFPlocYiR2zpL7ekyhw,626
 lattifai/bin/subtitle.py,sha256=UZMPh71O2X1UwbfZ9VWlhzxkz78viz8KWwoVsDpewK0,6577
 lattifai/io/__init__.py,sha256=mGnQkGm6BClRcogRK3J1nJJfPOk5z4o0sZI2nnyJVOU,1087
 lattifai/io/gemini_reader.py,sha256=WDZA93MSrUAsa5j-ZDXLdPXzEIoREymEy-rMAED_6f4,15152
 lattifai/io/gemini_writer.py,sha256=rlXO9zx6kQhqTi9K9izE69-8S-2GPOIiJHPwZyebpiM,6515
 lattifai/io/reader.py,sha256=h4T8dveLHXqSonma0J50iyjqkxH26tujeoPbnLx05nA,3333
 lattifai/io/supervision.py,sha256=iBDRiDJ0hddo__SoEZau2cdEIBFnXZNLgSWFjtJd-lM,871
-lattifai/io/text_parser.py,sha256=LQHgcEYXaSdhwUo9rP6P_31Z6RMv_BTP1YSKzXji4bk,2386
+lattifai/io/text_parser.py,sha256=5rO2qgFLfWIcoU0K-FGBgs2qG7r6Uz2J2LklGTgSUyA,2391
 lattifai/io/utils.py,sha256=4drRwcM1n7AYhdJcF51EZxMTy_Ut_1GKtdWpRhPuVmg,686
 lattifai/io/writer.py,sha256=8n9ZBuXuVOCFwzr1hqrnXpZ-fARTsepebwjKgRuueWE,3872
 lattifai/tokenizer/__init__.py,sha256=y-FyfO7tLga9b46pkCC6jdSBKOFZS-jFfHcqUieGEyU,120
 lattifai/tokenizer/phonemizer.py,sha256=SfRi1KIMpmaao6OVmR1h_I_3QU-vrE6D5bh72Afg5XM,1759
 lattifai/tokenizer/tokenizer.py,sha256=6XlHehhwahVWgUMj79LqMs13xaRTa17beOZqico4vLE,19186
 lattifai/workers/__init__.py,sha256=s6YfkIq4FDIAzY9sPjRpXnJfszj2repqnMTqydRM5Zw,83
-lattifai/workers/lattice1_alpha.py,sha256=1lCq0-bgWMXvYslAbCTFgHC0p6UWPto1y0wkTw9WrmQ,10177
+lattifai/workers/lattice1_alpha.py,sha256=826U2D5UJMnrxqswF6N1mlSs-3xRB9TgsiQe-Z89Ie4,10152
 lattifai/workflows/__init__.py,sha256=mTOdwQQs2YY1s0JsVGsATb2TWPhpNo7bRiAAJW92740,830
 lattifai/workflows/agents.py,sha256=ZnxyEb-Li8gQw3Z7V3K7jjnT7sQAtG1uCFuXGX77IcE,227
 lattifai/workflows/base.py,sha256=ZSePq2O2hB59K5CMbk_iGiOM9FSHOVQdG3t8Oyz9gwE,6264
 lattifai/workflows/file_manager.py,sha256=5km59cB4s9PcJgMMqZ6YqRY8sTqvTzd1qyQ2T5bHlfI,31985
-lattifai/workflows/gemini.py,sha256=zxLYqRolTUiUV3fj32sH91a_nCYUe7HtTxNKhlueZyI,5827
+lattifai/workflows/gemini.py,sha256=kpp3GiHyui8RZLWRQMx7vojBscCi2lZSxrjM1VHo_mc,6144
 lattifai/workflows/youtube.py,sha256=y1ez32G40CtHAo7Kz7J_DO96ljOwW-3SP9Sm2uPe5hY,38785
-lattifai-0.4.2.dist-info/licenses/LICENSE,sha256=LNuoH5jpXXNKgjQ3XLwztFq8D3O7kZI-LSg81o4ym2M,1065
-lattifai-0.4.2.dist-info/METADATA,sha256=LDwKsTMhAz6UBwbgAG0tkBg74W2kXttiwGvLHGlO128,26719
-lattifai-0.4.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-lattifai-0.4.2.dist-info/entry_points.txt,sha256=fCgo8-LKA_9C7_jmEGsZPJko0woXHtEh0iRbpO7PYzI,69
-lattifai-0.4.2.dist-info/top_level.txt,sha256=tHSoXF26r-IGfbIP_JoYATqbmf14h5NrnNJGH4j5reI,9
-lattifai-0.4.2.dist-info/RECORD,,
+lattifai/workflows/prompts/README.md,sha256=X49KWSQVdjWxxWUp4R2w3ZqKrAOi6_kDNHh1hMaQ4PE,694
+lattifai/workflows/prompts/__init__.py,sha256=i3kMT5sg_W9LbPcda0xmZWLg0tPjXGVI3iKtHrBng3o,1351
+lattifai/workflows/prompts/gemini/README.md,sha256=rt7f7yDGtaobKBo95LG3u56mqa3ABOXQd0UVgJYtYuo,781
+lattifai/workflows/prompts/gemini/transcription_gem.txt,sha256=cljzZ--BDgnnKzqVCakr-fTp2Xk38UOsUquvruNX-LU,4600
+lattifai-0.4.4.dist-info/licenses/LICENSE,sha256=LNuoH5jpXXNKgjQ3XLwztFq8D3O7kZI-LSg81o4ym2M,1065
+lattifai-0.4.4.dist-info/METADATA,sha256=wW1ZAk7_WdKxHpvgucnJA2xXye1vB-NiKAfpTT56hDk,26710
+lattifai-0.4.4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+lattifai-0.4.4.dist-info/entry_points.txt,sha256=fCgo8-LKA_9C7_jmEGsZPJko0woXHtEh0iRbpO7PYzI,69
+lattifai-0.4.4.dist-info/top_level.txt,sha256=tHSoXF26r-IGfbIP_JoYATqbmf14h5NrnNJGH4j5reI,9
+lattifai-0.4.4.dist-info/RECORD,,