Modularity and separation of concerns, in hardware. Of course those with embedded/microcontroller skills are also very welcome. The Espressif docs are a reasonable spec for something quite good on the voice side.īut then we would also need a good dataset of background noise. In addition to ML skills, the project would need someone that is good at organizing volunteer outreach, in order to build a good sized dataset. Model architecture wise the problem is rather well understood, there are several good papers available from ARM etc.ĭeployement infrastructure also pretty good these days, for example with Tensorflow Lite Micro. And then tackling low SNR and far-field like with Alexa is yet another level. Getting to PoC level is quite quick, but then getting robust performance in nearfield / high SNR cases across diverse background noise is a lot more work.

It will be a multi-month endeavour though. Ping me there and we can create a channel etc. One can find me in the Sound of AI Slack. Microcontroller-level audio ML is my speciality. In case anyone with practical skills in ML+deep learning (but not in audio or embedded) wants to tackle this project, I am willing to mentor on those things. Let’s just say I won’t be waiting for it. The more likely outcome (as you dismissively said) will be “yet another open source project” that goes in the graveyard of completely unusable open source wake word implementations. You will have created the first of its kind, and you’ll be famous! If you can create a completely open source wake implementation that gets even remotely close to the performance and reliability of those from Espressif, etc (while running on a microcontroller) we would be thrilled to use it. Yes, in the 15 seconds it took you to write that comment you came up with something no one in the field has ever thought of before. It’s getting very, very old and I’ve all but given up engaging on it because it always ends with the classic “Well couldn’t you just…”. Everyone has an opinion and exactly zero of them have any idea what they are talking about (that I’ve encountered). That said, wake word comes up more than I could have ever imagined. Someone else linked to the Espressif training dataset requirements. Yet it requires a not-completely dissimilar level of effort. You’re not going to see OpenAI, Meta, etc release a microcontroller optimized wake implementation because literally no one cares (by comparison). This isn’t Whisper or an LLM that has applications in practically anything. Short answer: it’s a tiny niche that is still very difficult, expensive, and time consuming.

Arduino relay 1 channel