OpenAI / whisper-large-v3-turbo
Whisper Large v3 Turbo
Universal multilingual transcription
Languages
99
Parameters
809M
License
MIT
OpenVox fit
Multilingual STT
Model overview
Built into OpenVox for private local speech transcription and subtitles.
Whisper Large v3 Turbo is a high-speed, pruned multilingual speech recognition foundation model developed by OpenAI, optimized for fast offline transcription and timestamped subtitle generation.
Whisper Large v3 Turbo reduces decoder depth from 32 layers to 4 layers, dramatically speeding up inference while retaining near Large v3 accuracy.
Supports 99 languages with automatic language identification, robust punctuation, capitalization, and English speech translation.
Perfect for transcribing interviews, podcasts, video audio, lectures, and meetings into plain text or timestamped SRT/VTT subtitles.
Runs 100% locally on your computer in OpenVox with zero microphone audio, recordings, or transcripts sent over the internet.
All speech transcription runs 100% locally on your computer with offline model weights. No microphone audio, meeting recordings, or transcripts are ever sent to external cloud servers.
Open-source model: OpenAI / whisper-large-v3-turbo
View on Hugging Face