Lite Whisper Large v3 Acc
Efficient Speech · released Feb 26, 2025 · efficient-speech/lite-whisper-large-v3-acc
Speech to textTranscribes a recording into words
- Type
- Open weightsApache License 2.0
- Languages
- 99
- Size
- 1.4B
Context measured in tokens
Our take
Written Sep 4, 2026Lite Whisper Large v3 Acc is a 1.4-billion-parameter speech-to-text model from Efficient Speech with open weights under an Apache licence. It covers 99 languages and transcribes at 204 times real time, though it trades some accuracy on clean and everyday audio for speed and accented-speech performance.
Pick this for open-weights transcription in 99 languages without vendor lock-in, or when throughput matters — an hour of audio in about 18 seconds on benchmark hardware. It also suits accented international earnings calls, where it outperforms most rivals. Skip it if your audio is clean read-aloud, podcasts, or meeting-room recordings with crosstalk, where it lags the field; or if you need a hosted provider, as none are currently tracked.
The case for it
- 204 times real-time speed on benchmark hardware — an hour of audio in roughly 18 seconds.
- Better than most on accented earnings-call audio, with a 10.69% word error rate against a field median of 10.8%.
- Apache 2.0 licence allows commercial use, modification and redistribution.
- 99 languages supported, far broader than most transcription models.
The case against it
- Worse than most on clean read-aloud audio, the easiest test case, at 1.63% word error rate versus a field median of 1.4%.
- Worse than most on podcasts and everyday video, at 8.48% versus a field median of 8.2%.
- Worse than most on meeting-room audio with crosstalk, at 13.69% versus a field median of 10.6%.
How good is it?
An open transcription model for turning speech into text, though meeting recordings are where it struggles most.
- transcribing recordings of meetings in a roomRecorded meetings · 73rd of 92
TranscriptionTurning speech into text
204×57th of 74
an hour of audio in 18 seconds, on the board's own hardware. Your machine will differ.
99
Stated by the leaderboard; we do not hold the list itself.
Percentage of words wrong on each set, lower better. Bars are scaled to this model's own worst case; the placing beneath each rate is against every model measured on that set.
The figures above come from the Open ASR Leaderboard, an independent public test that runs every model on the same recordings. It is the only measurement of transcription quality we know of, so there are no other scores to show.
Each of these is the same transcription job on a different kind of recording, so together they say where it holds up and where it slips — not how closely it follows an instruction.
Every published score for this model7 scoresEvery figure we hold, from 7 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
Comfortable fit
GeForce RTX 4090 · 24 GB
Room to spare. 20.7 GB spare means a 10% error in the size would not change the answer.
Apple M1 Pro (16-core GPU) · 32 GB
Room to spare. 21.9 GB spare means a 10% error in the size would not change the answer.
Apple M2 (8-core GPU, 8GB unified) · 8 GB
Room to spare. 3.9 GB spare means a 10% error in the size would not change the answer.
These cards answer whether Lite Whisper Large v3 Acc loads, not how fast it transcribes. Throughput figures for a transcription model come from its text decoder, so treat this as a fit answer rather than a speed one.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first. This is a fit answer: whether it loads, not how fast it transcribes.
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
Licence and identifiers
What the licence allowsApache License 2.0, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Apache License 2.0
Fully permissive: commercial use, redistribution, and derivatives allowed. Requires attribution and a copy of the license. Includes an express patent grant.
Identifiers
- Hugging Face
- efficient-speech/lite-whisper-large-v3-acc
- Architecture
- Dense
- Takes in, gives back
- Audio in, text out
- Catalogue slug
- efficient-speech-lite-whisper-large-v3-acc