Speech2Text — Local Whisper Transcription

A small desktop GUI that transcribes audio/video files into text using faster-whisper (open-source Whisper). Runs 100% locally — no cloud APIs required.

Features

• 🎙️ Transcription of .m4a / .mp3 / .mp4 / .wav / .webm / ...
• ⚡ Two modes:
  • Precision — Whisper large-v3
  • Speed — deepdml/faster-whisper-large-v3-turbo-ct2
• 🖥️ CPU (int8) or GPU (CUDA float16) inference
• 🪟 Simple PySimpleGUI desktop UI
• 📝 Local logging to ./logs/app.log

Architecture

.
├── app.py            # GUI + transcription logic
├── requirements.txt
└── logs/             # auto-created (gitignored)

Setup

python -m venv venv
source venv/bin/activate   # Windows: venv\Scripts\activate
pip install -r requirements.txt

For GPU, install the CUDA-enabled wheels of PyTorch matching your CUDA version (see https://pytorch.org/get-started/locally/).

Usage

python app.py

Then in the GUI:

1. Pick a mode (Precision / Speed)
2. Pick CPU or GPU
3. Browse to your audio/video file
4. Pick an output .txt path
5. Click Run

Stack

• Python 3.10+
• faster-whisper
• PyTorch
• PySimpleGUI

License

MIT