Whisper
Definition
Whisper includes multilingual models from OpenAI for speech recognition and speech translation.
A family of multilingual encoder-decoder models for speech recognition and speech translation published by OpenAI.
Simply explained
Whisper listens to an audio recording and can produce text or a linguistically translated output from it.
Why it matters in practice
The term helps to select, develop, evaluate AI systems in a well-founded manner and realistically assess their technical risks.
Example from business
A company evaluates Whisper locally on technical language and marks uncertain passages.
Limits and pitfalls
- Whisper is a product model, not a synonym for ASR; hallucinations, language, latency, and licensing must be checked.
Origin & history
Whisper was described in 2022 on 680,000 hours of weakly supervised audio data and published as weights.
Use in mid-sized companies
- Possible example: A company evaluates Whisper locally on technical language and marks uncertain passages.
Frequently asked questions
What does Whisper mean?
A family of multilingual encoder-decoder models for speech recognition and speech translation released by OpenAI. What should you watch out for? Whisper is a product model, not a synonym for ASR; hallucinations, language, latency, and licensing need to be checked.
Sources
- Robust Speech Recognition via Large-Scale Weak Supervision · Radford et al., Robust Speech Recognition via Large-Scale Weak Supervision
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: