Repository: ochen1/insanely-fast-whisper-cli
Branch: main
Commit: ab934dd330df
Files: 7
Total size: 11.1 KB

Directory structure:
gitextract_2amsqsvz/

├── .gitignore
├── LICENSE
├── README.md
├── insanely-fast-whisper.py
├── install-gfx1010.sh
├── requirements-gfx1010.txt
└── requirements.txt

================================================
FILE CONTENTS
================================================

================================================
FILE: .gitignore
================================================
# Byte-compiled / optimized / DLL files
__pycache__/
*.py[cod]
*$py.class

# C extensions
*.so

# Distribution / packaging
.Python
build/
develop-eggs/
dist/
downloads/
eggs/
.eggs/
lib/
lib64/
parts/
sdist/
var/
wheels/
share/python-wheels/
*.egg-info/
.installed.cfg
*.egg
MANIFEST

# PyInstaller
#  Usually these files are written by a python script from a template
#  before PyInstaller builds the exe, so as to inject date/other infos into it.
*.manifest
*.spec

# Installer logs
pip-log.txt
pip-delete-this-directory.txt

# Unit test / coverage reports
htmlcov/
.tox/
.nox/
.coverage
.coverage.*
.cache
nosetests.xml
coverage.xml
*.cover
*.py,cover
.hypothesis/
.pytest_cache/
cover/

# Translations
*.mo
*.pot

# Django stuff:
*.log
local_settings.py
db.sqlite3
db.sqlite3-journal

# Flask stuff:
instance/
.webassets-cache

# Scrapy stuff:
.scrapy

# Sphinx documentation
docs/_build/

# PyBuilder
.pybuilder/
target/

# Jupyter Notebook
.ipynb_checkpoints

# IPython
profile_default/
ipython_config.py

# pyenv
#   For a library or package, you might want to ignore these files since the code is
#   intended to run in multiple environments; otherwise, check them in:
# .python-version

# pipenv
#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
#   However, in case of collaboration, if having platform-specific dependencies or dependencies
#   having no cross-platform support, pipenv may install dependencies that don't work, or not
#   install all needed dependencies.
#Pipfile.lock

# poetry
#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
#   This is especially recommended for binary packages to ensure reproducibility, and is more
#   commonly ignored for libraries.
#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
#poetry.lock

# pdm
#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
#pdm.lock
#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
#   in version control.
#   https://pdm.fming.dev/#use-with-ide
.pdm.toml

# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
__pypackages__/

# Celery stuff
celerybeat-schedule
celerybeat.pid

# SageMath parsed files
*.sage.py

# Environments
.env
.venv
env/
venv/
ENV/
env.bak/
venv.bak/

# Spyder project settings
.spyderproject
.spyproject

# Rope project settings
.ropeproject

# mkdocs documentation
/site

# mypy
.mypy_cache/
.dmypy.json
dmypy.json

# Pyre type checker
.pyre/

# pytype static type analyzer
.pytype/

# Cython debug symbols
cython_debug/

# PyCharm
#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
#  and can be added to the global gitignore or merged into this file.  For a more nuclear
#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
#.idea/


================================================
FILE: LICENSE
================================================
MIT License

Copyright (c) 2023 ochen1

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.


================================================
FILE: README.md
================================================
# Insanely Fast Whisper (CLI)

[![GitHub License](https://img.shields.io/badge/license-MIT-blue.svg)](https://github.com/ochen1/insanely-fast-whisper-cli/blob/main/LICENSE)
[![Python Version](https://img.shields.io/badge/python-3.10-blue)](https://www.python.org/downloads/)

Powered by 🤗 *Transformers* & *Optimum* and based on **[Vaibhavs10/insanely-fast-whisper](https://github.com/Vaibhavs10/insanely-fast-whisper)**.

**TL;DR** - 🎙️ Transcribe **300** minutes (5 hours) of audio in less than **10** minutes - with [OpenAI's Whisper Large v2](https://huggingface.co/openai/whisper-large-v2). Blazingly fast transcription is now a reality!⚡️

## Features

✨ **ASR Model**: Choose from different 🤗 Hugging Face ASR models, including all sizes of [openai/whisper](https://github.com/openai/whisper) and even use an English-only variant (for non-large models).

🚀 **Performance**: Customizable optimizations ASR processing with options for batch size, data type, and BetterTransformer, all from the comfort of your terminal! 😎

📝 **Timestamps**: Get an SRT output file with accurate timestamps, allowing you to create subtitles for your audio or video content.

## Installation

- Clone git repository with `git clone https://github.com/ochen1/insanely-fast-whisper-cli`
- Switch to that folder with `cd insanely-fast-whisper-cli/`
- (optional) Create a new Python environment with `python -m venv venv`
- (optional) Activate environment with `source venv/bin/activate`
- Install packages from requirements with `pip install -r requirements.txt`
- Run program with `python insanely-fast-whisper.py`

## Usage

```bash
insanely-fast-whisper --model openai/whisper-base --device cuda:0 --dtype float32 --batch-size 8 --better-transformer --chunk-length 30 your_audio_file.wav
```

- `model`: Specify the ASR model (default is "openai/whisper-base").
- `device`: Choose the computation device (default is "cuda:0").
- `dtype`: Set the data type for computation ("float32" or "float16").
- `batch-size`: Adjust the batch size for processing (default is 8).
- `better-transformer`: Use BetterTransformer for improved processing (flag).
- `chunk-length`: Define audio chunk length in seconds (default is 30).

## Example

Transcribing an audio file with English-only Whisper model and returning timestamps:

```bash
insanely-fast-whisper --model openai/whisper-base.en your_audio_file.wav
```

## Output

The tool will save an SRT transcription of your audio file in the current working directory.

## License

This project is licensed under the [MIT License](https://github.com/ochen1/insanely-fast-whisper-cli/blob/main/LICENSE).

## Acknowledgments

- This tool is powered by Hugging Face's ASR models, primarily Whisper by OpenAI.
- Optimizations are developed by [Vaibhavs10/insanely-fast-whisper](https://github.com/Vaibhavs10/insanely-fast-whisper).
- Developed by [@ochen1](https://github.com/ochen1).

## 📞 Contact

Have questions or feedback? Feel free to create an issue!

🌟 **Star this repository if you find it helpful!**

[![Star History Chart](https://api.star-history.com/svg?repos=ochen1/insanely-fast-whisper-cli&type=Date)](https://star-history.com/#ochen1/insanely-fast-whisper-cli&Date)

---

[![GitHub Issues](https://img.shields.io/github/issues/ochen1/insanely-fast-whisper-cli.svg)](https://github.com/ochen1/insanely-fast-whisper-cli/issues)
[![GitHub Pull Requests](https://img.shields.io/github/issues-pr/ochen1/insanely-fast-whisper-cli.svg)](https://github.com/ochen1/insanely-fast-whisper-cli/pulls)

🚀 Happy transcribing with Insanely Fast Whisper! 🚀


================================================
FILE: insanely-fast-whisper.py
================================================
#!/usr/bin/env python3

import click
import os
import time

@click.command()
@click.option('--model', default='openai/whisper-base', help='ASR model to use for speech recognition. Default is "openai/whisper-base". Model sizes include base, small, medium, large, large-v2. Additionally, try appending ".en" to model names for English-only applications (not available for large).')
@click.option('--device', default='cuda:0', help='Device to use for computation. Default is "cuda:0". If you want to use CPU, specify "cpu".')
@click.option('--dtype', default='float32', help='Data type for computation. Can be either "float32" or "float16". Default is "float32".')
@click.option('--batch-size', type=int, default=8, help='Batch size for processing. This is the number of audio files processed at once. Default is 8.')
@click.option('--better-transformer', is_flag=True, help='Flag to use BetterTransformer for processing. If set, BetterTransformer will be used.')
@click.option('--chunk-length', type=int, default=30, help='Length of audio chunks to process at once, in seconds. Default is 30 seconds.')
@click.argument('audio_file', type=str)
def asr_cli(model, device, dtype, batch_size, better_transformer, chunk_length, audio_file):
    from transformers import pipeline
    import torch

    # Initialize the ASR pipeline
    pipe = pipeline("automatic-speech-recognition",
                    model=model,
                    device=device,
                    torch_dtype=torch.float16 if dtype == "float16" else torch.float32)

    if better_transformer:
        pipe.model = pipe.model.to_bettertransformer()

    # Perform ASR
    click.echo("Model loaded.")
    start_time = time.perf_counter()
    outputs = pipe(audio_file, chunk_length_s=chunk_length, batch_size=batch_size, return_timestamps=True)

    # Output the results
    click.echo(outputs)
    click.echo("Transcription complete.")
    end_time = time.perf_counter()
    elapsed_time = end_time - start_time
    click.echo(f"ASR took {elapsed_time:.2f} seconds.")

    # Save ASR chunks to an SRT file
    audio_file_name = os.path.splitext(os.path.basename(audio_file))[0]
    srt_filename = f"{audio_file_name}.srt"
    with open(srt_filename, 'w', encoding="utf-8") as srt_file:
        prev = 0
        for index, chunk in enumerate(outputs['chunks']):
            prev, start_time = seconds_to_srt_time_format(prev, chunk['timestamp'][0])
            prev, end_time = seconds_to_srt_time_format(prev, chunk['timestamp'][1])
            srt_file.write(f"{index + 1}\n")
            srt_file.write(f"{start_time} --> {end_time}\n")
            srt_file.write(f"{chunk['text'].strip()}\n\n")

def seconds_to_srt_time_format(prev, seconds):
    if not (isinstance(seconds, int) or isinstance(seconds, float)):
        seconds = prev
    else:
        prev = seconds
    hours = seconds // 3600
    seconds %= 3600
    minutes = seconds // 60
    seconds %= 60
    milliseconds = int((seconds - int(seconds)) * 1000)
    hours = int(hours)
    minutes = int(minutes)
    seconds = int(seconds)
    return (prev, f"{hours:02d}:{minutes:02d}:{int(seconds):02d},{milliseconds:03d}")

if __name__ == '__main__':
    asr_cli()


================================================
FILE: install-gfx1010.sh
================================================
echo "Python <= 3.10 only!"
uv pip install -r requirements-gfx1010.txt --extra-index-url https://download.pytorch.org/whl/rocm5.2
echo
echo "Example usage:"
echo "HSA_OVERRIDE_GFX_VERSION=10.3.0 python3 insanely-fast-whisper.py --model distil-whisper/distil-medium.en audio.mp3"


================================================
FILE: requirements-gfx1010.txt
================================================
#accelerate
#optimum
click
transformers
torch==1.13.1+rocm5.2


================================================
FILE: requirements.txt
================================================
torch
torchvision
torchaudio
git+https://github.com/huggingface/transformers
accelerate
optimum
click