Moonshine
Moonshine Voice is an open-source speech toolkit under MIT built for live voice interfaces that run on the device with no account or API key. Its speech-to-text models are trained from scratch in sizes down to 1 MB and are designed to work while the person is still talking. One library covers Python, the browser, iOS, Android, macOS, Linux, Windows and Raspberry Pi. Every streaming model and every English model is MIT.
What does Moonshine do?
Moonshine Voice is "an open source AI toolkit for developers building real-time voice agents and applications". Everything runs on the device with no account or API key. The README says it is "optimized for live streaming, with low latency by doing work while the user is still talking", and that its models are "trained from scratch, from higher accuracy than Whisper Large V3 down to tiny 1MB models". That accuracy claim is the project's own.
One library covers Python, JavaScript through WebAssembly, iOS, Android, macOS, Linux, Windows and Raspberry Pi. The toolkit also covers text to speech and conversational agents; those models are licensed separately from the speech-to-text ones.
Key facts
- Licence: MIT on the code, apart from a third-party folder, and MIT by default on the models.
- The LICENSE names the exceptions as exhaustive: legacy non-streaming models in Arabic, Japanese, Korean, Mandarin, Spanish, Ukrainian and Vietnamese, thirteen models in all, under the Moonshine Community License.
- That community licence is free for research and non-commercial use, requires registration for commercial use, and terminates once the licensee's annual revenue passes one million US dollars.
- Text-to-speech and grapheme-to-phoneme models follow their own readmes, since they derive from third-party work.
- The README's accuracy claim: models "from higher accuracy than Whisper Large V3 down to tiny 1MB models", the project's own figure.
- Optimised for live streaming, doing work while the user is still talking.
- Platforms: Python, JavaScript/WASM, iOS, Android, macOS, Linux, Windows and Raspberry Pi; install with pip install moonshine-voice.
- Latest tagged release when read: v0.1.5 on 2026-08-24.
What does it replace, and where does it fall short?
Moonshine replaces Deepgram and AssemblyAI for live audio on a device, in a browser or on small hardware, where a round trip to a server is too slow or not allowed.
The licence is worth reading in full. The code, apart from a third-party folder, is MIT. The LICENSE file says "Moonshine models are released under the MIT License by default, in every language and at every size", including every streaming model and every English model. The exceptions, which the file calls exhaustive, are legacy non-streaming models in Arabic, Japanese, Korean, Mandarin, Spanish, Ukrainian and Vietnamese, thirteen models in all, under the Moonshine Community License: free for research and non-commercial use, commercial use by registration, and terminating once the licensee's annual revenue passes one million US dollars. Text-to-speech and grapheme-to-phoneme models follow their own readmes.
Where it falls short: word timestamps, speaker labelling and the full language list are not in the README; they are in the documentation.
How does Moonshine run?
pip install moonshine-voice, then moonshine-voice mic --language en for a live test. A C API, JavaScript/WASM and mobile SDKs are listed. GPU requirements are not stated, which matches a toolkit designed for CPUs and phones.
Who is Moonshine for?
A team building a voice interface that must respond while the person is still speaking, on hardware it does not control. A team doing bulk transcription of recordings on a GPU should look at Whisper or WhisperX.
What limits does the README state?
None beyond the licence exceptions above.
Questions people ask
Is Moonshine open source?
Yes, for the code and for the models most people will use: the LICENSE says models are MIT by default in every language and at every size, including every streaming model and every English model. Thirteen legacy non-English, non-streaming models are the exception, under a non-commercial community licence, and the file lists them by name.
Can Moonshine be used commercially?
Yes, under MIT, for the code and for every model except the thirteen legacy ones. Those need registration for commercial use and their licence ends when annual revenue passes one million US dollars. If you stay on the streaming models you never touch that clause.
Does Moonshine work offline?
Yes. The README's first point is that everything runs on the device with no account or API keys. The models are built for that, with sizes down to 1 MB for small hardware, and the same library runs on phones, in the browser through WebAssembly, and on a Raspberry Pi.
How does Moonshine compare with Deepgram?
Deepgram sells a hosted, conversation-aware speech API. Moonshine gives you streaming recognition on the device with no round trip, which matters for a voice interface that has to respond while the person speaks. The alternatives to Deepgram page places it as the pick for live streaming on the device.
Which languages does Moonshine support?
The README does not list them; the documentation does. What the LICENSE tells you is that non-English models exist at least for Arabic, Japanese, Korean, Mandarin, Spanish, Ukrainian and Vietnamese, since those are the languages whose legacy non-streaming models carry the community licence, and that streaming models are MIT in every language.
Sources
- Moonshine README: github.com/moonshine-ai/moonshine, read 2026-09-16.
- Moonshine LICENSE, with the model licence list: LICENSE, read 2026-09-16.
Compared with the others
On the open-source alternatives to Deepgram page, Moonshine is the pick for live streaming on the device. MIT code, MIT streaming models, built to work while the person is still talking, with models down to 1 MB and one library across Python, the browser, phones and desktops.
Also on that page: Whisper for the reference model on a gpu, whisper.cpp for a cpu, a mac, or a phone, WhisperX for word timestamps and speaker labels.
On the open-source alternatives to AssemblyAI page, Moonshine is the pick for live streaming on the device. MIT code, MIT streaming models, built to work while the person is still talking, with models down to 1 MB and one library across Python, the browser, phones and desktops.
Also on that page: WhisperX for word timestamps and speaker labels, Speaches for an openai-compatible server, Whisper for the reference model on a gpu.
More speech-to-text engines
OpenAI's general-purpose speech recognition model, released with code and weights under MIT, in six sizes from 39 million to 1.55 billion parameters, with multilingual transcription, translation to English and language detection.
A dependency-free C and C++ port of Whisper under MIT that runs on CPUs, Apple Silicon and phones, with quantised models, an HTTP server and bindings for a dozen languages.
Batched Whisper transcription under BSD-2-Clause with word-level timestamps from forced alignment and speaker labels from pyannote, the closest open-source match to a hosted transcription API's output.
NVIDIA's speech models and toolkit: Apache-2.0 code, and the Parakeet TDT 0.6B v3 model under CC-BY-4.0 with 25 European languages, automatic language detection, punctuation and word timestamps.
A runtime under Apache-2.0 from the next-generation Kaldi team that runs speech to text, text to speech, speaker diarization and voice activity detection locally on CPUs, NPUs and phones, with no internet connection.
An offline speech recognition toolkit under Apache-2.0 with streaming, models of about 50 MB, 20+ languages, a reconfigurable vocabulary and speaker identification, from small boards to clusters.
Added September 16, 2026. Every claim above comes from the project's README, LICENSE or model card, read on September 16, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/moonshine-ai/moonshine.
Want Moonshine running on your own servers?
Reveneau sets it up on your hardware, connects it to your product, and builds the software around it.