News for hackers. Stuff that matters.

A comment on Show HN: CPU-only transcription for YouTube, TikTok, X, Instagram videos. Parent comment

Re:Show HN: CPU-only transcription for YouTube, TikTok, X, I... by mscdex on Wednesday May 20, 2026 @11:22PM (#48217481)
Why the choice of Kroko over something like parakeet-tdt-0.6b-v3, which is also faster than realtime on CPU?
[ Parent | on HN ]
Re:Show HN: CPU-only transcription for YouTube, TikTok, X, I... by nshm on Thursday May 21, 2026 @01:59AM (#48218477)
Kroko models are more accurate and their size is just a hundred megabytes compared to parakeet (2.5 gigabytes in default fp32)
[ Parent | on HN ]
Re:Show HN: CPU-only transcription for YouTube, TikTok, X, I... by mscdex on Thursday May 21, 2026 @02:40AM (#48218742)
Do you have a link to results confirming this? Kroko does not seem to be on the Open ASR Leaderboard. Parakeet has an average WER of 6.32 across several common datasets.
[ Parent | on HN ]
Re:Show HN: CPU-only transcription for YouTube, TikTok, X, I... by mrkn1 on Thursday May 21, 2026 @11:17AM (#48224245)
Kroko's website says benchmarks aren't formalized yet. FWIW, this url says 5% WER for English [0]. though it doesn't specify the dataset, so not directly comparable to Parakeet's 6.32 on the Open ASR Leaderboard

Best way to judge is to try it on your own audio

[0] https://huggingface.co/hudaiapa88/sherpa-stt-onnx

[ Parent | on HN ]