Lmst

#S2T

With Whisper-Hindi, high-performance ASR no longer demands massive compute — just a single RTX 4090 and a few smart tricks are enough to reach state-of-the-art results. https://col.la/whisperhindi2 #Transcription #S2T #ML #AI #OpenSource

After cleaning up and expanding Whisper-Hindi to 3,000 hours, we now have explicit timestamp prediction, faster I/O, and fine-tuned models across all sizes, bringing us even closer to fully reliable, production-ready Hindi ASR: https://www.collabora.com/news-and-blog/news-and-events/breaking-language-barriers-20-moving-closer-production-ready-hindi-asr.html

ICYMI ⤵️

Whisper is now available in Hindi! With 2,500 hours of Hindi speech data and innovative techniques like Indic Normalization, this model sets a new benchmark for Hindi ASR: https://www.collabora.com/news-and-blog/news-and-events/breaking-language-barriers-fine-tuning-whisper-for-hindi.html

By using techniques like Indic Normalization, Whisper now supports Hindi! With 2,500 hours of Hindi speech data, this model sets a new standard for Hindi ASR: https://www.collabora.com/news-and-blog/news-and-events/breaking-language-barriers-fine-tuning-whisper-for-hindi.html

We're proud to announce that Whisper is now available in Hindi! With 2,500 hours of Hindi speech data and innovative techniques like Indic Normalization, this model sets a new benchmark for Hindi ASR: https://www.collabora.com/news-and-blog/news-and-events/breaking-language-barriers-fine-tuning-whisper-for-hindi.html

I just tested https://github.com/natrys/whisper.el in my #Emacs: offline speech-to-text (#voice recognition) in English

Of course, it's not perfect (as any #S2T system) but I'm really impressed how good the results are. 👍

I guess this will be an integral part of my future workflows here and there. #PIM #orgmode

Version: 2025.04