Last updated 7 months ago

Microphones play a very important role on the performance of Wakeword Detection and Speech Recognition engines.

In this section we list the microphones that we think are going to suit you best to build a Voice Assistant at home.

The Respeaker 2 provides you with 2 microphones, that work up to 3 meters distance from the source. It has 3 LEDS and a button that you can configure and play with. It also offers two different interfaces to output audio: the 3.5mm Audio Jack and the JST2.0 Speaker Out.

At Snips, we have had a very good experience with this affordable microphone and it is our choice for the makers kit.

The big brother of the Respeaker 2 Mics Pi HAT is the Respeaker 4, which provides four microphone and makes it even a more powerful choice to enhance your voice assisted projects. Moreover, it has a 12 LED ring that you can configure to respond to VAD (Voice Activity Detection), and also highlight the estimated direction the voice is coming from.

Please note that the ReSpeaker 2 Mics Pi HAT and the ReSpeaker 4 Mics Pi HAT do not have any front audio processing. They are affordable Mics with good performance, but may not work as well if you are speaking from far away or/and in a noisy environment

The ReSpeaker Mic Array 2.0 is an upgraded version based on a XMOS’s XVF-3000. This new chipset includes multiple voice recognition algorithms to assist in performance. This far-field voice capture board has 4 microphones and 12 programmable RGB LED indicators which also highlight the estimated direction the voice is coming from.

Respeaker 2 and 4 install in Raspberry Pi:

You can use Sam to setup your Respeaker. Please follow these instructions to install Sam in your laptop and then just type sam setup audio and follow the instructions prompted in the console. Alternatively you can do the installation by yourself as follows:

$ git clone https://github.com/respeaker/seeed-voicecard.git
$ cd seeed-voicecard
$ sudo ./install.sh
$ sudo reboot