search

SYSTRAN / faster-whisper

Faster Whisper transcription with CTranslate2
MIT License
11.24k stars 937 forks source link

why faster-whisper is slower than the origin whisper model #922

Open ASHLEYDX opened 1 month ago

ASHLEYDX commented 1 month ago

For the same whisper model (small), faster-whisper is even slower than origin whisper model. To load the model, my code is : """ from faster_whisper import WhisperModel model=WhisperModel("small", device='cpu',compute_type='int8' """ the CPU is 8 core Intel(R) Xeon(R) Gold 6346@3.1GHZ. Does anyone encounter the same issue?

ooobo commented 1 month ago

How did you install faster-whisper and which version - can you share the command? If it was latest main might be same issue #917. Otherwise can you share testing code.