Speech to text · OpenAI
Whisper large-v3 turbo
Whisper with a 4-layer decoder: much faster, slightly less accurate.
Datasheet
Maker | OpenAI |
|---|---|
Parameters | 809 MVerifiedsafetensors.total from the HF API Source |
Licence | MITVerified |
Access | Public download |
Task (HF pipeline) | automatic-speech-recognition |
First published on HF | 2024-10-01 |
Last updated on HF | 2024-10-04 |
Weights in memory | 1.6 GB at 16-bit · 0.5 GB at ~4.85 bitsEstimateParameters × bits ÷ 8. Excludes context cache and runtime overhead; not every model has a 4-bit build. |
Benchmark
8.39%
Open ASR Leaderboard, average WER (English short-form) · #7 of 7 scored in this category
avg 8.3871, RTFx 810.5. Source
Used in
The description summarises the model card at https://huggingface.co/openai/whisper-large-v3-turbo. Parameters and licence are read from the Hugging Face API; the weights-in-memory line is our arithmetic, not a measurement.