I just got Mycroft up and running on the Google AIY Voice Kit as well.
It’s much easier to install Mycroft on the Google Voice Kit image rather than fiddle with the Voice HAT drivers in picroft. For reference, here is the procedure I used after flashing the Google Voice Kit image and booting it up:
sudo apt-get update
Install Mycroft for Linux –
cd ~/
git clone https://github.com/MycroftAI/mycroft-core.git
cd mycroft-core
bash dev_setup.sh
Start the PulseAudio daemon (might need to reboot after doing this the first time) –
pulseaudio -D
Start Mycroft –
cd ~/mycroft-core
./start-mycroft.sh debug
Obviously you’ll want to have a bash script to run PulseAudio and Mycroft at system startup if you plan on having your Voice Kit be a standalone Mycroft box.
If anyone is ambitious enough to try to get the Voice HAT drivers working on picroft, here are some steps I took before I gave up and did it the other way around:
sudo apt-get update
sudo apt-get upgrade
Get drivers from the raspberry pi kernel –
sudo apt-get install raspberrypi-kernel
Clone the Google AIY Voice project and use it to configure the Voice HAT driver –
cd ~/
git clone https://github.com/google/aiyprojects-raspbian.git voice-recognizer-raspi
cd ~/voice-recognizer-raspi
sudo scripts/configure-driver.sh
sudo scripts/install-alsa-config.sh
sudo reboot
** Not sure if this is necessary **
cd ~/voice-recognizer-raspi
sudo scripts/install-deps.sh
sudo scripts/install-services.sh
List PulseAudio sources –
pacmd list-sources
Set the default source to whichever one is the microphone (index 1 for me) –
pacmd set-default-source 1
At this point Mycroft was able to output TTS through the speaker using the “speak” command, e.g. “speak hi”. He could also process speech through the mic, and his responses to my commands were output to the debug screen. However, his responses were not output to the speaker for whatever reason. It’s probably just a configuration file that needs fixing, but I wasn’t able to figure it out.