The moonshine GitHub repository is a public open-source automatic speech recognition project that advances scientific knowledge and democratizes access to AI/ML technology for edge devices. The content strongly advocates for Articles 19 (freedom of expression and information), 26 (education), and 27 (scientific and cultural advancement) through transparent code publication, unrestricted global distribution, and removal of access barriers. Structural support for accessibility features and collaborative development further amplifies alignment with UDHR principles of universal participation and equitable resource access.
This is awesome, well done guys, I’m gonna try it as my ASR component on the local voice assistant I’ve been building https://github.com/acatovic/ova. The tiny streaming latencies you show look insane
No idea why 'sudo pip install --break-system-packages moonshine-voice' is the recommended way to install on raspi?
The authors do acknowledge this though and give a slightly too complex way to do this with uv in an example project (FYI, you dont need to source anything if you use uv run)
haven't tested yet but I'm wondering how it will behave when talking about many IT jargon and tech acronyms. For those reason I had to mostly run LLM after STT but that was slowing done parakeet inference. Otherwise had problems to detect properly sometimes when talking about e.g. about CoreML, int8, fp16, half float, ARKit, AVFoundation, ONNX etc.
For those wondering about the language support, currently English, Arabic, Japanese, Korean, Mandarin, Spanish, Ukrainian, Vietnamese are available (most in Base size = 58M params)
According to the OpenASR Leaderboard [1], looks like Parakeet V2/V3 and Canary-Qwen (a Qwen finetune) handily beat Moonshine. All 3 models are open, but Parakeet is the smallest of the 3. I use Parakeet V3 with Handy and it works great locally for me.
Any plans regarding JavaScript support in the browser?
There was an issue with a demo but it's missing now. I can't recall for sure but I think I got it working locally myself too but then found it broke unexpectedly and I didn't manage to find out why.
Accuracy is often presumed to be english, which is fine, but it's a vague thing to say "higher" because does it mean higher in English only? Higher in some subset of languages? Which ones?
The minimum useful data for this stuff is a small table of language | WER for dataset
I've helped many Twitch streamers set up https://github.com/royshil/obs-localvocal to plug transcription & translation into their streams, mainly for German audio to English subtitles.
I'd love a faster and more accurate option than Whisper, but streamers need something off-the-shelf they can install in their pipeline, like an OBS plugin which can just grab the audio from their OBS audio sources.
I see a couple obvious problems: this doesn't seem to support translation which is unfortunate, that's pretty key for this usecase. Also it only supports one language at a time, which is problematic with how streamers will frequently code-switch while talking to their chat in different languages or on Discord with their gameplay partners. Maybe such a plugin would be able to detect which language is spoken and route to one or the other model as needed?
Claiming higher accuracy than Whisper Large v3 is a bold opening move. Does your evaluation account for Whisper's notorious hallucination loops during silences (the classic 'Thank you for watching!'), or is this purely based on WER on clean datasets? Also, what's the VRAM footprint for edge deployments? If it fits on a standard 8GB Mac without quantization tricks, this is huge.
Nice work. One metric I’d really like to see for streaming use cases is partial stability, not just final WER.
For voice agents, the painful failure mode is partials getting rewritten every few hundred ms. If you can share it, metrics like median first-token latency, real-time factor, and "% partial tokens revised after 1s / 3s" on noisy far-field audio would make comparisons much more actionable.
If those numbers look good, this seems very promising for local assistant pipelines.
My startup is making software for firefighters to use during missions on tablets, excited to see (when I get the time) if we can use this as a keyboard alternative on the device. It's a use case where avoiding "clunky" is important and a perfect usecase for speech-to-text.
Due to the sector being increasingly worried about "hybrid threats" we try to rely on the cloud as little as possible and run things either on device or with the possibility of being self-hosted/on-premise. I really like the direction your company is going in in this respect.
We'd probably need custom training -- we need Norwegian, and there's some lingo, e.g., "bravo one two" should become "B-1.2". While that can perhaps also be done with simple post-processing rules, we would also probably want such examples in training for improved recognition? Have no VC funding, but looking forward to getting some income so that we can send some of it in your direction :)
The streaming architecture looks really promising for edge deployments. One thing I'm curious about: how does the caching mechanism handle multiple concurrent audio streams? For example, in a meeting transcription scenario with 4-5 speakers, would each stream maintain its own cache, or is there shared state that could create bottlenecks?
I vibe-trained moonshine-tiny on amateur radio morse code last weekend, and was surprised at the ~2% CER I was seeing in evals and over the air performance was pretty acceptable for a couple hour run on a 4090.
Congrats on the results. The streaming aspect is what I find most exciting here.
I built a macOS dictation app (https://github.com/T0mSIlver/localvoxtral) on top of Voxtral Realtime, and the UX difference between streaming and offline STT is night and day. Words appearing while you're still talking completely changes the feedback loop. You catch errors in real time, you can adjust what you're saying mid-sentence, and the whole thing feels more natural. Going back to "record then wait" feels broken after that.
Curious how Moonshine's streaming latency compares in practice. Do you have numbers on time-to-first-token for the streaming mode? And on the serving side, do any of the integration options expose an OpenAI Realtime-compatible WebSocket endpoint?
Which program does support it to allow streaming? Currently using spokenly and parakeet but would like to transition to a model that is streaming instead of transcribing chunk wise.
Open-weight STT models hitting production-grade accuracy is huge for privacy-sensitive deployments. Whisper was already impressive, but having competitive alternatives means we're not locked into a single model family. The real test will be multilingual performance and edge device efficiency—has anyone benchmarked this on M-series or Jetson?
By the way, I've been using a Whisper model, specifically WhisperX, to do all my work, and for whatever reason I just simply was not familiar with the Handy app. I've now downloaded and used it, and what a great suggestion. Thank you for putting it here, along with the direct link to the leaderboard.
I can tell that this is now definitely going to be my go-to model and app on all my clients.
I find it an even more weird practice for anyone working with speech or text models not in the first paragraph name the language it is meant for (and I do not mean the programming language bindings). How many English native speakers are there 5% of the world population?
Interesting. Can we get in touch? I just sold my webapp/saas where I used NB-Whisper to transcribe Norwegian media (podcast, radio, TV) and offer alerts and search by indexing it using elasticsearch.
Edit: It was https://muninai.eu (I shut down the backend server yesterday so the functionality is disabled).
I'm building a local-first transcription iOS app and have been on Whisper Medium, switching to Parakeet V3 based on this.
One note for anyone using Handy with codex-cli on macOS: the default "Option + Space" shortcut inserts spaces mid-speech. "Left Ctrl + Fn" works cleanly instead. I'm curious to know which shortcuts you're using.
Tangentially, have you got any idea what the equivalent "partial tokens revised" rate for humans is? I know I've consciously experienced backtracking and re-interpreting words before, and presumably it happens subconsciously all the time. But that means there's a bound on how low it's reasonable to expect that rate to be, and I don't have an intuition for what it is.
Project explicitly contributes to cultural and scientific advancement: moonshine is open-source ASR technology advancing speech recognition science; project enables participation in scientific progress.
FW Ratio: 43%
Observable Facts
Project title explicitly identifies it as advancing speech recognition science: 'Fast and accurate automatic speech recognition (ASR) for edge devices'.
The repository is published under open-source terms enabling scientific collaboration.
GitHub's collaborative infrastructure supports scientific discussion and peer contribution.
Inferences
Open-source ASR project directly advances scientific knowledge and practice.
Public repository enables participation in scientific development.
Removal of paywall/licensing barriers supports universal scientific culture.
Open-source model embodies principles of scientific freedom and knowledge sharing.
Repository title and description explicitly address information sharing about speech recognition technology: 'Fast and accurate automatic speech recognition (ASR) for edge devices' demonstrates commitment to communicating technical innovation.
FW Ratio: 50%
Observable Facts
The page title is 'GitHub - moonshine-ai/moonshine: Fast and accurate automatic speech recognition (ASR) for edge devices', directly stating the project's information focus.
The repository is publicly visible, indexed, and shareable across networks.
GitHub's platform design enables forking, sharing, and dissemination of code and knowledge.
Inferences
Public repository publication directly advances freedom of opinion and expression.
Open-source distribution model actively promotes information freedom and knowledge access.
Technical documentation enables informed debate and public understanding of AI/ML systems.
Project contributes to education: open-source ASR implementation provides educational resource for machine learning and speech processing; technical documentation supports learning.
FW Ratio: 50%
Observable Facts
The repository is freely accessible for anyone to study and learn from.
Open-source code enables educational inspection of ASR implementation.
GitHub's platform supports documentation and educational materials without cost barriers.
Inferences
Public source code serves educational function in machine learning and AI literacy.
Free access supports education rights for global learners.
Technical transparency enables informed understanding of AI systems.
Project description addresses health and welfare: speech recognition technology enables accessibility for persons with disabilities and supports inclusive communication.
FW Ratio: 50%
Observable Facts
The project explicitly targets 'edge devices', supporting healthcare use cases in resource-constrained settings.
Speech recognition technology enables communication access for persons with disabilities.
GitHub page includes ARIA labels and keyboard navigation support visible in HTML structure.
Inferences
ASR technology directly supports health and wellness for disabled persons.
Open-source model enables adaptation for diverse health contexts.
Platform accessibility features ensure equitable access to project information.
Public repository enables free expression of information; documentation and code are openly accessible and shareable; README and technical content are freely distributed without gatekeeping.
GitHub's public repository model makes moonshine freely accessible for educational use and skill development; source code transparency enables learning-by-examination; no paywalls restrict educational access.
GitHub's accessibility features (keyboard navigation, ARIA labels, responsive design) enable equal access to the moonshine repository; edge device optimization supports health accessibility for resource-limited populations.
GitHub's platform structure enables collaborative development and open-source contributions, supporting collective human advancement through technology sharing.
GitHub's platform does not visibly restrict participation based on protected characteristics; the moonshine repository maintains open contribution policies.
GitHub provides privacy controls for repository data; public repository mode allows user control over disclosure, with privacy settings available for sensitive information.
GitHub retains platform control; user contributions are subject to GitHub's terms of service, creating conditional rather than absolute intellectual property ownership. Moonshine project is open-source, which distributes rather than concentrates property rights.
Open-source repository enables developers to express technical ideas and design philosophy without censorship; GitHub community guidelines protect expression within bounds.
GitHub's issue trackers and discussions enable collaborative organization and collective action on technical projects; moonshine repository can host community organizing around ASR development.
Open-source community governance models enable democratic participation in project decisions; moonshine can implement governance aligned with collective decision-making.
Open-source projects like moonshine provide free access to technology that supports economic and social welfare; accessibility of speech recognition tools advances equity.
Open-source project participates in global digital commons supporting international cooperation; GitHub's global infrastructure enables worldwide collaboration aligned with Article 28 principles of international order.
GitHub's terms and open-source licensing prevent nullification of UDHR rights; public repository cannot be used to restrict others' fundamental freedoms.
Supplementary Signals
How this content communicates, beyond directional lean. Learn more
build 1ad9551+j7zs · deployed 2026-03-02 09:09 UTC · evaluated 2026-03-02 10:41:39 UTC
Support HN HRCB
Each evaluation uses real API credits. HN HRCB runs on donations — no ads, no paywalls.
If you find it useful, please consider helping keep it running.