Skip to content

Comment on AiOla open-sources ultra-fast ‘multi-head’ speech recognition modelparent

Comments

Depends what you mean by “fast”.

I’ve tested WhisperLive, it’s basically real-time (i.e. low latency).

Dosent whisperlive just use faster-whisper under the hood. Witch can be way faster than real time.

There's a difference between speed & latency when it comes to performance.

Faster-whisper can process audio faster than real-time, but AFAIK vanilla Whisper needs a few seconds long audio "frame" to do inference/STT.

Whisper Live fixes that, and reduces latency to a few tens/hundreds of ms.

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.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.