Fact · Speech recognition
The Whisper paper says many different tasks can be performed on the same input audio, and lists transcription, translation, voice activity detection and alignment among them.
Checked against its source on October 17, 2026
In the source's own words
“There are many different tasks that can be performed on the same input audio signal: transcription, translation, voice activity detection, alignment”
Robust Speech Recognition via Large-Scale Weak Supervision (full text), Alec Radford et al., OpenAI (arXiv), 2022, Section 2.3, Multitask Format.
Open the original (DOI 10.48550/arXiv.2212.04356)- What it measures
- The authors want one model to perform the whole speech processing pipeline, and they specify the task in the input with special tokens, such as transcribe or translate.
- What it does not show
- A task being supported by the model does not mean every task is reliable in every language. Per-language accuracy varies a lot.