Whisper Large — Welsh & English — CTranslate2 (int8)

CTranslate2 int8 quantised version of techiaith/whisper-large-ft-cy-en.

This model provides faster inference with lower memory usage. See the source model for full details on training data, evaluation results, and usage.

Usage

import faster_whisper

model = faster_whisper.WhisperModel("techiaith/whisper-large-ft-cy-en-ct2")

# Welsh transcription
segments, info = model.transcribe("welsh_audio.wav", language="cy", task="transcribe")

# Welsh to English translation
segments, info = model.transcribe("welsh_audio.wav", language="cy", task="translate")