What Is Whisper AI? OpenAI's Speech Recognition Model
Whisper is OpenAI's open-source speech recognition model, trained on 680,000 hours of audio across 99+ languages, available free under MIT license.
Definition
Whisper is OpenAI's open-source automatic speech recognition (ASR) model, trained on 680,000 hours of multilingual audio, supporting transcription and translation across 99+ languages.
Full Explanation
Whisper is a speech recognition model released by OpenAI in September 2022 under the MIT open-source license. It was trained on 680,000 hours of multilingual and multitask supervised data collected from the web. Whisper performs speech recognition (audio to same-language text), speech translation (audio to English text), language identification, and voice activity detection, all within a single model. It supports 99+ languages including Hindi, Mandarin, Spanish, and Arabic, with varying accuracy by language. Whisper comes in multiple sizes (tiny, base, small, medium, large, large-v3, turbo), trading off accuracy for speed and memory. Because it is open source, Whisper can be self-hosted via WebGPU/WASM in the browser, on a GPU server, or used through hosted APIs (OpenAI, Groq, Replicate, Fal.ai). Whisper popularized accessible, free, high-quality transcription and is the engine behind many modern caption generators.
Examples
- -Running `whisper audio.mp3 --model large-v3` in a terminal.
- -Calling OpenAI's hosted Whisper API to transcribe a podcast.
Frequently Asked Questions
Everything you need to know before you start.
Can't find what you're looking for? Contact us