Noise Suppression/Multiple DOA

Hello everyone,



I was just wondering if there is a way to do multiple directions of arrival (ex. if two people are talking, the ReSpeaker Core v2 will detect and do DOA for both). Also, how does one do noise suppression (ex. cancel out background music and record just the person talking)?



Thanks!

Hi there



For the first question, it does not support to detect 2 DOAs at one time.



For the second question, respeaker core v2.0 has 2 call back channels, one is from 3.5mm audio jack and other is from speaker port. so it supports cancel the background music. thanks.

Thank you very much for replying!



I am still having trouble with noise suppression. Do I have to enable it or install something for noise suppression to work? How do I get it to work?



Thanks!

Hi there,



Please install the librespeaker, for detail instructions, please refer to https://github.com/respeaker/respeakerd. THANKS.

I also installed all required software, but noise suppression does’nt work.



Respeaker run:
</s><i> </i>/usr/local/bin/respeakerd --agc_level=-5 --ref_channel=6 --snowboy_res_path="/usr/local/etc/respeakerd/resources/common.res" --snowboy_model_path="/usr/local/etc/respeakerd/resources/snowboy.umdl" --snowboy_sensitivity="0.4" --source="alsa_input.platform-sound_0.seeed-8ch" <e>

Python example run:
</s><i> </i>python demo_respeaker_v2_vep_alexa_with_light.py <e>

I’m playing music over mpd (using pulseaudio) and it still showing speech detection… (a lot of dots in console). I need to yell “snowboy” to activate it, but it records all input from speakers. When I stop playing music, it works good



UPD:

of course, I tried to change “ref_channel” to 7, agc_level from -31 to 0…