Respeaker Core v2 DOA, Echo Cancel, and Pixel ring without AVS

I will be running my own voice recognition program on this respeaker. I imaged my SD card to respeaker-debian-9-lxqt-sd. and installed my voice recognition program. I say the wake word, it answers, seems to work well.



However, I want to make sure all the benefits of the board are being used. I want to make sure its doing the below



Beam forming

Detection of Arrival

Noise suppression

Echo cancelation

auto gain control



I do not care about wake word, because that is programmed into my voice assistant.



I also would like the LED ring to show the direction of arrival. The only info I have seen has described how to make all this work with Amazon voice services. How can I get all of this to work without.



Does the BF, DOA, NS, EC, AGC all automatically work regardless of what you are running on it for the voice assistant, or how do I achieve my goal? I’m very new, and have no idea where to start.

Hi there,



A. For pixel ring, please refer to https://github.com/respeaker/pixel_ring



B. For DOA,

  1. Python : please refer to <LINK_TEXT text=“https://github.com/voice-engine/voice-e … aker_v2.py”>https://github.com/voice-engine/voice-engine/blob/master/examples/kws_doa_alexa_respeaker_v2.py</LINK_TEXT>
  2. C++: please refer to <LINK_TEXT text=“http://respeaker.io/librespeaker_doc/cl … sNode.html”>http://respeaker.io/librespeaker_doc/classrespeaker_1_1VepDoaKwsNode.html</LINK_TEXT>



    C. ReSpeakerd: Audio front-end processing algorithms including AEC, Beamforming, NS and KWS

    1. http://respeaker.io/librespeaker_doc/
  3. https://github.com/respeaker/respeakerd



    you can install respeakerd to process the audio no matter what voice assistant you use. thanks.



    Bill

I was looking through those links.



I installed pixel ring, but when I get to the line “python examples”, I get an error importError: no module named gpiozero



for DOA, I see in the files that it is detecting a wake word of Alexa. My setup the voice assistant has its own wake word. I want to avoid having to say alexa, wakeword 2, command… is there a way around that?



I will try and read through that other documentation again, most of it doesn’t make any sense to me. lol, I’m very much a noob.

Ok, I went through the turtorial listed here <LINK_TEXT text=“http://wiki.seeedstudio.com/ReSpeaker_C … e-solution”>http://wiki.seeedstudio.com/ReSpeaker_Core_v2.0/#closed-source-solution</LINK_TEXT>



And got everything installed correctly. I’m now on Part 3. Compile and Run AVS C++ SDK. Do I skip this part since I’m not linking to AVS? and do I need to do anything specific to make sure my AI is listening POST processing, or is that all done automatically since everything is installed??

No, you do not need to install avs, for the respeakerd audio processing, Please refer to <LINK_TEXT text=“https://github.com/respeaker/respeakerd … _MANUAL.md”>https://github.com/respeaker/respeakerd/blob/master/doc/DEVELOPER_MANUAL.md</LINK_TEXT>. thanks.