I’ve been using Whisper Open AI API for automatic speech recognition across multiple languages, specifically Malayalam, Nepali, and Telugu. However, starting about two weeks ago, I’ve noticed that Whisper has stopped correctly recognizing these languages. When I input speech in any of these languages, the resulting transcription appears in a different language entirely.
I’ve tested this across different audio samples and various use cases, but the problem persists. Prior to this, Whisper API was performing well with these languages. I’d like to understand if there have been any recent changes or known issues affecting language recognition, especially for regional languages like Malayalam, Nepali, and Telugu.
Is anyone else experiencing similar issues? Any insights or guidance on how to resolve this would be much appreciated.
Thank you in advance.