Hacker News Reader

That's pretty good. Amazon Echo integration is one of the main Cloud dependencies I have.

I also like that you are directly supporting ESPHome and I'll definitely take that route and build something nice.

However, for the more casual users, are you thinking about releasing an off-the-shelf "voice assistant" device?

Yes, that's the plan.

We're still exploring different ML on the edge chips to help with wake word and audio processing. One hard requirement that we have is that the tooling is openly available, so that anyone can convert an AI model and run it on the chip. We don't accept tooling to be available as a website or under NDA. That challenge is harder than we expected!

What does an ML-on-the-edge system chip do differently from other chips? Does it have a gpu or other accelerator on it?

Some labs are developing analogue chips instead of digital

Domain specific fixed function hardware could consume 10-100x less power than a gpu for the same workload. Think video decode or ‘YouTube battery life’ benchmarks but for speech recognition.

Reply on news.ycombinator.com