Audio to Text Classifier

A deep learning model based on Whisper that converts non-speech audio into textual descriptions.

Overview

This project uses OpenAI's Whisper model, fine-tuned to classify and describe non-speech audio signals into text. It can be used for various applications such as environmental sound classification, machine sound analysis, and audio event detection.

Installation

git clone https://github.com/rubemalmeida/audio-classifier
cd audio-classifier
pip install -r requirements.txt

Usage

Training

Place your audio files in folders named after their categories under the sounds directory:

sounds/
├── categoryX/
│ ├── sound1.wav
│ └── sound2.wav
└── category2/
  ├── sound3.wav
  └── sound4.wav

Then run:

python src/train.py

Inference

python src/infer.py

Requirements

• Python 3.8+
• PyTorch 2.0+
• See requirements.txt for full list

License

MIT

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.