ReSpeaker Core v2.0

The ReSpeaker Core v2.0 allows developers create powerful and impactful voice and sound interfaces. Suitable to use as a base for smart assistance as well as responding to environmental sounds, the ReSpeaker Core v2.0 was designed with the idea that developers deserve to have many options available to them.

We received much feedback from ReSpeaker Core v1 users and have added many requested features. Different from the original board this SBC is running Debian* with an onboard microphone array and software based voice enhancement algorithms.

The Debian and Android is supported now, for more detail please check our wiki.

The ReSpeaker Core v2.0 is powered by the Axol Core module, featuring the Rockchip RK3229 Quad-Core Cortex-A7 with a Mali 400MP GPU, 1GB of DDR3 RAM, and a PMU. The core runs the Debian OS. This provides developers with much more flexibility and resources to create their projects.

Around the core are peripherals including a Kingston 4GB eMMC to enable the OS to run onboard, a WiFi/BLE module, USB, Grove, and HDMI connectors, the on board 6-mic array, LED ring, and more.

An Onboard 6-Mic Array with Professional Voice Enhancement Algorithms

New in our ReSpeaker product line is a software based approach to voice enhancement. The more powerful Axol Core allows the Core v2.0 to process the audio received through the 6 far field microphones in real-time. The ReSpeaker Core v2.0 supports Noise Suppression (NS), Keyword Wakeup, Direction of Arrival (DoA), Beam-forming (BF), Acoustic Echo Cancellation (AEC). Our BF and AEC algorithms are provided by Alango, a professional audio DSP company used in the automotive industry.

An Open Software Ecosystem

We recognize that each project has different requirements and goals. As such we have worked to provide as much flexibility to developers as possible. Our quad-core solution allows us to run numerous software solutions. From the selection of algorithms, NLP services, smart assistants, operating systems, and more, we wish to support our users. If you have suggestions on what services you would like to see us officially support let us know on our forums, or if you enjoy the work yourself, submit a merge request to our github.

Audio and Video Output*

Audio output is available via the 3.5mm audio jack (output only), as well as a JST two pin connector and Bluetooth. The ReSpeaker Core v2.0 supports video output through HDMI* enabling developers to create rich visual interfaces to pair with a voice interface.

*GPU not fully implemented on Debian: 1080p video playback will be available July 2018. Android 8.1 will support 4k video on release.

Customization
The ReSpeaker Core v2.0 is ready for customization. To allow quick customization we separated the core and peripherals. If a project required a custom mic array solution we are able to quickly design and manufacture to help scale. Contact us at [email protected] to learn more.

Features

• Debian-Based Linux System

• SDK for Speech Algorithms with Full Documents

• C++ SDK and Python Wrapper
  

• Speech Algorithms and Features

  • Keyword Spotting (Wake-Up)

  • BF (Beamforming)

  • DoA (Direction of Arrival)

  • NS (Noise Suppression)

  • AEC (Acoustic Echo Cancellation) and AGC (Automatic Gain Control)

• All-in-One Solution with High Performance SoC

• 8 Channel ADC for 6 Microphone Array and 2 Loopbacks (Hardware Loopback)

Applications

• Smart Speaker

• Intelligent Voice Assistant System
  

• Voice Recorder

• Voice Conferencing System
  

• Meeting Communication Equipment

• Voice Interacting Robot

• Car Voice Assistant

• Other Scenarios Utilizing Voice Interfaces

ReSpeaker Core v2.0 - Alexa Demo

Document

Wiki

Technical details

Part List

ECCN/HTS