Zipformer Transducer XL 290M
Sounds Good AI · released May 12, 2026 · soundsgoodai/Zipformer-transducer-XL-290M
Speech to textTranscribes a recording into words
- Type
- Open weightsCreative Commons Attribution-NonCommercial 4.0
- Languages
- 1
- Size
- 0.3B
Context measured in tokens · download allowed, licence restricts use
Our take
Written Sep 4, 2026Zipformer Transducer XL is a tiny downloadable speech-to-text model from Sounds Good AI that turns audio into written words at exceptional speed. It is built for non-commercial research and rapid local prototyping rather than production deployment.
Pick this when you need transcription done almost instantly on your own hardware and accuracy is not the priority — it processes an hour of audio in about 25 seconds. Use it for non-commercial research or for financial earnings calls, where it scores better than most alternatives. Skip it if you need commercial deployment, clean read-aloud accuracy, or any hosted provider.
The case for it
- Processes audio at 141 times real time — an hour of audio in roughly 25 seconds on the benchmark hardware.
- Strong on financial earnings calls, with a lower error rate than most models on that condition.
- Better than most on heavily accented international speech.
The case against it
- Worse than most on clean read-aloud audio, podcasts and meetings — its error rate exceeds the median on all three.
- Non-commercial licence blocks most production use: no commercial deployment or for-profit redistribution.
- No hosted inference available; you must run it yourself.
How good is it?
An open speech-to-text model that turns recordings into written text, though it trails most models on transcription overall and on podcasts.
- turning spoken English into written textOpen ASR WER · 60th of 76
- transcribing podcasts and video audioPodcasts and video · 73rd of 92
TranscriptionTurning speech into text2.5 of 5Open ASR WER · 60th of 76
93.7%
Misses roughly one word in 16, averaged over nine English test sets.
141×64th of 74
an hour of audio in 26 seconds, on the board's own hardware. Your machine will differ.
1
Listed on the model card. The accuracy above is English only.
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 model9 scoresEvery figure we hold, from 9 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. 21.4 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. 22.6 GB spare means a 10% error in the size would not change the answer.
Apple M1 (8-core GPU, 8GB unified) · 8 GB
Room to spare. 4.6 GB spare means a 10% error in the size would not change the answer.
These cards answer whether Zipformer Transducer XL 290M 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 allowsCreative Commons Attribution-NonCommercial 4.0, 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
Creative Commons Attribution-NonCommercial 4.0
Weights are downloadable but commercial use is prohibited. Research and personal use only.
Identifiers
- Hugging Face
- soundsgoodai/Zipformer-transducer-XL-290M
- Takes in, gives back
- Audio in, text out
- Catalogue slug
- soundsgoodai-zipformer-transducer-xl-290m