This playbook prepares the Raspberry Pi for use as voice assistant with Rhasspy and Node-RED. Both applications will be running on Docker CE. It also stages driver files for a ReSpeaker voice card and starts an online radio Docker container.
Raspberry Pi OS (aka Raspbian) is the preferenced distribution as it is more stable and offers better driver support for the ReSpeaker, but Ubuntu Server should also be possible. Please not that conventional Debian won't work.
If you want to slim-down your Ubuntu installation, there is also an Ansible Playbook.
The following Ansible roles are used:
Copy your SSH key to your Raspberry Pi:
$ ssh-copy-id pi@<ip-address>Update the inventory file to include your Raspberry Pi IP addresses:
[assistants]
192.168.178.200When using Rasbian instead of Ubuntu, change the remote_user variable to pi.
Run the playbook:
$ ansible-playbook -i inventoryWait - this can take up to 45 minutes. Once finishes, login to the Raspberry Pi and compile the seeed-voicecard drive (if you're using a ReSpeaker voice card):
$ ssh ubuntu@<ip_address>
$ sudo bash
# cd seeed-voicecard
# ./ubuntu-prerequisite.sh
# ./install.shOnce finished, reboot the system:
# rebootYou should now be able to access the applications via the following links:
- radio_api: http://ip-address:5000
- Rhasppy: http://ip-address:12101
- Node-RED: http://ip-address:1880
- Settings
- Audio Recording > PyAudio
- Wake Word > Rhasspy Raven or Porcupine
- Speech to Text > Pocketsphinx
- Intent Recognition > Fsticuffs
- Text to Speech > NanoTTS
- for MaryTTS: http://marytts:59125/process
- Audio Playing > aplay
- Intent Handling > Remote HTTP
- Remote URL: http://ip-address:1880/intent
- Sounds
- Wake WAV:
/usr/lib/rhasspy/.venv/lib/python3.7/site-packages/snowboy/resources/ding.wav - Recorded WAV:
/usr/lib/rhasspy/.venv/lib/python3.7/site-packages/snowboy/resources/dong.wav
- Wake WAV:
- Save Settings
- F5
- Download
- Sentences
- Import from sentences.ini
- Save
MaryTTS