Moonshine Streaming Small
Useful Sensors · released Jan 6, 2026 · usefulsensors/moonshine-streaming-small
Speech to textTranscribes a recording into words
- Type
- Open weightsMIT License
- Languages
- 1
- Size
- 0.1B
Context measured in tokens
Our take
Written Aug 2, 2026Moonshine Streaming Small is a tiny English speech-to-text model from Useful Sensors, weighing roughly 0.1 billion parameters under a permissive MIT licence. It is built for speed above all else, processing an hour of audio in about a second on benchmark hardware, though its accuracy falls sharply on anything beyond clean read-aloud recordings.
Choose this when you need real-time or batch transcription on minimal hardware, or when deploying to edge devices where every megabyte counts. It suits clean read-aloud scenarios well. Skip it if you are transcribing meetings, podcasts, accented speech, or anything in a language other than English, or if you need hosted inference rather than self-hosting.
The case for it
- Extremely fast: processes roughly one hour of audio in about 1.1 seconds on benchmark hardware.
- Permissive MIT licence allows commercial use, modification, and redistribution.
- Strong on clean read-aloud English, with a word error rate far below its own average.
- Tiny parameter count of approximately 0.12 billion enables broad hardware compatibility.
The case against it
- Accuracy collapses on challenging audio: more than six times worse on accented speech than on clean read-aloud, and more than five times worse on recorded meetings.
- English only; no measurements held for any other language, and no hosted inference options in our data.
- Overall English accuracy is modest even at its best, with roughly one word in 15 wrong on average.
How good is it?
TranscriptionTurning speech into text2 of 5Open ASR WER · 59th of 74
93.1%
Misses roughly one word in 15, averaged over nine English test sets.
3,206×13th of 62
an hour of audio in 1 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, not to the board.
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.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done, which is why they get no rating.
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?
- Fits in memory
- weights load entirely on the card
- Spills to system RAM
- some weights offload; much slower
- Too large
- will not load even with offload
- est
- size is calculated; the verdict could change by 10%
Comfortable fit
GeForce RTX 4090 · 24 GB
Room to spare. 21.5 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. 22.7 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. 4.7 GB spare means a 10% error in the size would not change the answer.
Memory use by level
Against a 24 GB card.
All 3 sizes here are calculated, not measured. We hold no measured file for this model, so each size comes from the parameter count and every verdict above inherits that uncertainty.
Check against your own machine → · All 71 devices, with every size →
When we formed this view
Dates behind this page
Prices last checked 6h ago
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 allowsMIT License, 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
MIT License
Fully permissive: do anything with attribution. No patent grant, unlike Apache-2.0.
Identifiers
- Hugging Face
- usefulsensors/moonshine-streaming-small
- Architecture
- Dense
- Modality record
- audio->text
- Catalogue slug
- usefulsensors-moonshine-streaming-small