Malaysian Whisper Large v3 Turbo v3 model for CTranslate2

This repository contains the conversion of mesolitica/Malaysian-whisper-large-v3-turbo-v3 to the CTranslate2 model format.

This model can be used in CTranslate2 or projects based on CTranslate2 such as faster-whisper.

Example

from faster_whisper import WhisperModel

model = WhisperModel("SeamlessX/malaysian-faster-whisper-small-v3-ct2")

segments, info = model.transcribe("audio.mp3")
for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))

Conversion Details

The original Transformers model was converted to the CTranslate2 format with the following command:

ct2-transformers-converter \
    --model mesolitica/Malaysian-whisper-large-v3-turbo-v3 \
    --output_dir malaysian-faster-whisper-large-v3-turbo-ct2 \
    --copy_files tokenizer.json preprocessor_config.json \
    --quantization float16

Note that the model weights are saved in FP16. This type can be changed when the model is loaded using the compute_type option in CTranslate2.