Skip to content

Latest commit

 

History

History
46 lines (34 loc) · 1.89 KB

README.md

File metadata and controls

46 lines (34 loc) · 1.89 KB
Raw

ReSpeaker for Raspberry Pi

The repository contains some examples to use ReSpeaker series mic arrays on Raspberry Pi.

The examples are included in the custom image with pre-installed seeed-voicecard, snowboy, voice-engine and etc. You can flash the custom image to get started easily.

Hardware

ReSpeaker 2 Mic Hat, ReSpeaker 4 Mic Array or ReSpeaker 6 Mic Array (they are all pi hats)

Software

  • snowboy for KWS (Keyword Search / Keyword Spotting)
  • webrtc audio processing for NS (Noise Suppression)
  • speexdsp for AEC (Acoustic Echo Cancellation)
  • GCC-PHAT for DOA (Direction Of Arrial)
  • avs for alexa voice service
  • voice-engine for connecting all the elements together
pip install webrtc-audio-processing speexdsp voice-engine avs

and go to kitt-ai/snowboy to install snowboy.

Files

├── 2mic                               # ReSpeaker 2 Mic Hat
│   └── ns_kws_doa_alexa.py                hands-free alexa with NS, KWS and DOA (0 ~ 180 degree)
├── 4mic                               # ReSpeaker 4 Mic Array
│   ├── kws_doa.py                         KWS and then DOA
│   ├── ns_kws_doa_alexa.py                hands-free alexa with NS, KWS and DOA (0 ~ 360 degree)
│   └── ns_kws_doa.py
├── 6mic                               # ReSpeaker 6 Mic Array
│   ├── aec_ns_kws_doa_alexa.py            hands-free alexa with AEC, NS, KWS and DOA (0 ~ 360 degree)
│   ├── aec_ns_kws_doa.py                  has 2 loopback channels for AEC
│   └── kws_doa.py
├── ns_kws_alexa.py
├── ns_kws.py
└── ns_vs_raw.py                       Compare KWS between raw audio and audio with NS