Hacker News Reader

Comment by dharma1 | Hacker News Reader

I've been trying to find a usb array mic/software combo that works better for far field sound capture than a single mic - to use with either OK Google on old android phones, or Jasper running on Rpi.. Any recommendations?

View on HN

I would also love having something like this. I found the Playstation Eye camera for PS3 has an array of four mics, but there's no accompanying software, so it's not as useful. I'd buy a hardware/software combo that did far-field sound capture well.

Essentially, I want a better mic so I can run this:

https://www.youtube.com/watch?v=8eiHO7uqccs

Yeah, same here, PS3 Eye but no drivers/software for array mics or beamforming. I wanted to create something like in that video too. Looks promising with OK Google - a little laggy/glitchy but the idea definitely has legs.

I think in a couple of years this stuff will become ubiquitous

I hope so, and I certainly hope that we'll see hardware-only solutions with open software so I know where my sound is going.

Yeah, I'm pretty sure we'll have state-of-the-art open source deep learning based speech recognition libraries this year that will run on local hardware without significant latency.

Baidu open sourced their warp CTC a few weeks ago, give it another few months before someone will release a trained English network for it

https://github.com/baidu-research/warp-ctc

Maybe you want an Acoustic Magic Voice Tracker... http://www.acousticmagic.com

Not cheap and a little ugly some would say...

Try teleconference microphones. There used to be some great array mics for that workload

Look into the Kinect - it has open APIs and good microphones

Yep, I build my own jarvis with kinect and it works great.

Would you be able to go into some detail about the setup? Are you using Rpi or another board? What speech recognition library/APIs you are using? Are you using the array mic setup on the kinect?

Reply on news.ycombinator.com