Comment on AiOla open-sources ultra-fast ‘multi-head’ speech recognition modelparentComments−gcr2yWhat sort of performance are you needing?My Apple M1 MacBook (2021) can infer whisper-medium at roughly 10x realtime, for comparison. Takes about 20min to process three hours of audio.
Comments
What sort of performance are you needing?
My Apple M1 MacBook (2021) can infer whisper-medium at roughly 10x realtime, for comparison. Takes about 20min to process three hours of audio.