Noisy environment / Long distance

I do not worry you will make the natural language interpretation nice and add wide a variety of skills for MyCroft.
All of that look very cool.

I’m much more worried about loud environmen and long distance (not near the mycroft) recognition.
(= effortless usage !)
Have you developed some algorithm for that ? What about the micro quality ?

Amazon Echo, Cubic and other project seem to be strong on those points and I think they actually are the most important point for such a project success (i.e. I think hardware issues are as important as software issues here).

I would like to hear more about it, do you have any information ?

Thanks :slight_smile:

1 Like

I’m happy to see some information on the blog about the mic.

I’ve found some info about the range in the faq :

What is the range of Mycroft’s microphone?
Depending on the background noise level, Mycroft can accurately interpret speech at 15 feet. In a noisy environment, the range may be smaller.

Yeah, we are still experimenting with the mic to see what the maximum range we can get is. Obviously we’d like that number to be as far as possible.

There are so many factors that go into this as well, turns out a “room” can represent a ton of very different environments. :slight_smile:

Are your first tests encouraging ?

What about noise ? The echo can understand you even with music on, is there some kind of algorithm for that ?

I’m testing Mycroft Core on Xubuntu 16.04. He can’t understand speech through any built-in laptop mics that I’ve tried or with a decent quality Samson USB mic. He works very will with a corded headset, so room acoustics and echo seem to be the culprit. He works best with a Logitech webcam (C615) which has multiple mics and built-in noise/echo cancellation. With this, he can understand me clearly across the room.