Two years ago, Seeed released their popular ReSpeaker Core platform, giving users the ability to add voice interaction to their projects. It featured seven digital microphones that supported Beamforming, acoustic source localization, noise suppression, and echo cancellation- all powered by an XMOS XVF-2000 running OpenWRT. Don’t get me wrong, it’s still a great board, but it’s been improved upon with revision 2.0.
Seeed’s new, more powerful ReSpeaker Mic Array v2.0 replaces the 2000 WiSoC with XMOS’s latest XVF-3000- AKA the Rockchip RK3229 quad-core Cortex A7 processor with 1Gb of RAM. The board is also outfitted with six microphone array with a host of speech algorithms, including Direction of Arrival (DOA), Beamforming, and AEC (Acoustic Echo Cancellation) among a host of others.
With as much hardware as there is on the board, Seeed split the ReSpeaker V2.0 into two sections with the first being the center core module, which hosts the SoC (CPU, RAM, and PMU) while the other sports the connectors, eMMC and wireless components.
The full specs are as follows:
- SoC: Rockchip RK3229 quad-core Cortex-A7 processor (up to 1.5GHz) with Arm Mali-400MP2 GPU and 1Gb of DDR3 RAM
- Storage: 4Gb eMMC, micro SD slot
- Microphone: 8-channel ADC for 6-microphone array and two loopbacks
- Headset Connector: 3.5mm jack
- Speaker Connector: JST 2.0 output connector for an 8-Ohm speaker
- Video Output: HDMI 2.0
- Connectivity: Wi-Fi 802.11 b/g/n, Bluetooth 4.0/LE, Ethernet
- Peripherals: 3 X USB Host/OTB, 1 X USB Power, 1 X Groove Socket (support for I2C/Digital Port), 6 X GPIO, 1 X Debug UART, and 12 X programmable LEDs
The board ships with an impressive kit, which includes a speech recognition SDK (complete with documentation), and a C++ SDK with Python wrapper as well as the other mic accouterments mentioned earlier, such as Beamforming, DOA, and a myriad of others. Seeed is currently taking pre-orders for the ReSpeaker V2.0 for $99.