Transcribe audio/video using xezpeleta/whisper-large-v3-eu — the best Whisper model for Basque (Euskera).
# Install dependencies
pip install -r requirements.txt
# Transcribe Basque audio (default)
python transcribe.py audio.wav
# Transcribe Spanish
python transcribe.py video.mp4 --language es
# Use medium model (faster, less accurate)
python transcribe.py recording.mov --model medium
# 1-second segments for animation sync
python transcribe.py podcast.mp3 --max-dur 1- Python 3.8+
faster-whisper(CTranslate2-based Whisper)- ~3GB disk for large-v3 model (auto-downloads on first run)
| Language | Code |
|---|---|
| Basque | eu |
| Spanish | es |
| French | fr |
| English | en |
SRT subtitle file created next to input file:
audio.wav → audio.srt
video.mp4 → video.srt
MIT