You can find models here: huggingface.co/UsefulSensors Paper: arxiv.org/pdf/2509.02523…

Preprint: arxiv.org/abs/2509.02523 HF: huggingface.co/UsefulSensors/…

.Useful Sensors' speech-to-text model is faster than Whisper for the same accuracy, and it's in Torre, Useful's real-time translation device. The team is also researching speech-to-intent models, which Pete Warden said are the missing link for agentic AI: eetimes.com/pete-warden-sp…

We (Useful Sensors) are trying really really hard to make this a reality.

Naval David We are working on it: usefulsensors.com/torre with the eventual goal of making real time translation as simple, efficient, and ubiquitous as noise cancelling headphones 🎧.

kwindla Not finding any explicit benchmarking for TTFT, but Xenova posted some nice benchmarking stats (in general) here: github.com/usefulsensors/…

usefulsensors/moonshine: Fast and accurate automatic speech recognition (ASR) for edge devices github.com/usefulsensors/…

yuta suito この辺に頑張ってる話は出てるんですが、更新が止まってしまってますね github.com/usefulsensors/…

Andrew Carr 🤸 The numbers I saw and the vibes I heard about Useful Sensors' Moonshine (cc: keveman) were stunning as well -- do take a look!

hellooooo github.com/usefulsensors/…

rk3588のnpuのwhisperあるんですね github.com/usefulsensors/…

Moonshine (github.com/usefulsensors/…) is now on Open ASR leaderboard huggingface.co/spaces/hf-audi… Thanks Vaibhav (VB) Srivastav .

🤖 From this week's issue: Fast and accurate automatic speech recognition (ASR) for edge devices. github.com/usefulsensors/…

Moonshine-web: Run Moonshine models in the browser. Based on onnxruntime-web. Huggingface Space here: huggingface.co/spaces/UsefulS…. Source here: github.com/usefulsensors/…

Moonshine Fast and accurate automatic speech recognition (ASR) for edge devices. github.com/usefulsensors/…

And support the great work from Useful Tensors github.com/usefulsensors Thank you Pete Warden for the open contributions!