ReSpeaker 2-Mics Pi HAT

The reSpeaker 2-Mics Pi HAT has been equipped with 2 analogue microphones and WM8960 Audio Codec, for high-definition voice capture and process, designed to be the voice user interface for Raspberry Pi Series. Attach with NLU software algorithms, it has the abilities of Voice Activity Detection, Direction of Arrival and Key Word Spotting, widely used in Voice Interaction applications. It is a fine solution for local Neutral Language Understanding.

Feature

• Voice User Interface for Raspberry Pi Series: Placed by 4 analogue microphones, which come with the Voice Interaction algorithms of VAD(Voice Activity Detection), DOA(Direction of Arrival), KWS(Key Word Spotting)
• WM8960 Audio Codec: Low power, high-quality stereo Codec designed for portable digital audio applications
• Controllable Peripherals: Provide 3 programable APA102 RGB LEDs and 1 User Button for attached Speech Recognition usage
• Easy Assembly: Simply pop up on the Raspberry Pi to use

Description

The reSpeaker 2-Mics Pi HAT is the voice user interface for Raspberry Pi Series, which has been equipped 2 analogue microphones and WM8960 Audio Codec for high-definition voice capture. Attach with NLU software algorithms and noise reduction algorithms, it has the abilities of Voice Activity Detection, Direction of Arrival and Key Word Spotting, widely used in Voice Interaction applications.

How does reSpeaker work?

There are analogue microphones mounted on each corner of the square reSpeaker, to receive audio data around then transmit them to the codec. The codec integrates synchronized ADCs with mic boost amplifier, to deliver valid channel data to the Raspberry Pi/ Pi Zero, where it transits them over I2S ports by standard I2S or PCM format, along with a single port by TDM format.

The NLU algorithms will be applied to the Raspberry Pi once the audio data has arrived. The main functions include Voice Activity Detection(VAD), Direction of Arrival(DOA), and Key Word Spotting(WKS), suitable for various Voice Interaction applications such as voice-based remote control cars, voice user interface, keywords wake up and so on.

Since the audio data transmitted may contain the noise as well, the reSpeaker comes with the functions of Auto Echo Cancellation (AEC), Beamforming, Webrtc Noise Suppression (NS), and Automatic Gain Control (AGC) to reduce noise and perform better audio data to the Raspberry Pi.

Application

• Voice Interaction Learning
• Secondary software Development
• Voice-based remote control cars

Get started with Google Assistant

How to use button to trigger Google Assistant

Part List

LEARN AND DOCUMENTS

Documentations

• Resources

  • Respeaker_2_Mics_Pi_HAT_SCH
  • Respeaker_2_Mics_Pi_HAT_PCB
  • Respeaker_2_Mics_Pi_HAT_SCH
  • Respeaker_2_Mics_Pi_HAT_PCB
  • ReSpeaker 2 Mics Pi HAT 3D
  • Seeed-Voice Driver
  • Algorithms includes DOA, VAD, NS
  • Voice Engine project, provides building blocks to create voice enabled objects
  • AEC