I just uploaded an MP3 file and the accuracy of the transcription is much worse than usual. Usually, I find a couple of minor mistakes over 20 pages when this is a lot of background noise but this is returning big mistakes every page . The background noise is very low and much lower then the voice. It is almost like it is trying to guess the words from context but the audio is pretty clear. I don’t know if anyone else has had this experience.
Is there an issue with Whisper transcription right now?
I have personally had problems with Whisper AI - despite transcribing clear, well-spoken podcasts only. The transcription would miss entire paragraphs, often. I’ve started using Whisper AI locally on my machine since then, but it would be nice if LingQ fixed the transcription.
Thanks. I may have to do the same. Something seems to be wrong. I tried again today.
LingQ just provided the transcription and it failed to transcribe the last minute of the audio. Did the same again today - I had an 11 minute audio file and it stopped transcribing at 9 minutes.