High Voice & Audio · 1 min read
Whisper open source: audio transcription becomes a commodity
In one sentence OpenAI releases Whisper under MIT license: a speech-to-text model trained on 680,000 hours of multilingual audio, near commercial-grade quality, runs locally.
Reading level
OpenAI drops a model on GitHub that transcribes audio to text. Free, open-source, works in 90+ languages.
You install it as a command: pip install openai-whisper, feed it an mp3, get a .txt. Quality matches paid commercial services, and beats them on many languages.
Changes everything for podcasters, journalists, audio archives, captioning, accessibility. Sysadmins stop paying APIs to transcribe meetings.
Companies
OpenAI
Tools
Whisper
Tags
OpenAIWhisperASRSpeech-to-TextOpen SourceMultilingual
Sources