Bachelor thesis: Development and evalution of a voice assistant system based on open source compoments for Raspberry Pi.
Berta is a Raspberry Pi local voice assistant system based on open source components.
See the following instructions on how to work with it.
Deployement on Linux or Raspberry Pi Hosting. Raspberry Pi with Raspbian, official distribution from the Raspberry Pi Foundation. Use a fresh setup SD Card with Raspbian for using Berta.
At the moment, berta requires a very specific set of hardware. To use berta you will need:
- a Raspbery Pi (tested on a model 4 b with 4GB)
- a Seeed ReSpeaker raspberry pi hat (test with ReSpeaker 4-Mic Array for Raspberry pi)
- python 3+
Documentation on the microphone array can be found here: [https://wiki.seeedstudio.com/ReSpeaker_4_Mic_Array_for_Raspberry_Pi]
The drivers on the seeedstudio website are outdated, please refer to the install script if manually installing any dependencies.
Installing Berta on Raspberry Pi by cloning the application directly from its git repository:
$ git clone https://github.com/PassionPlus/berta.git
after cloning berta, please run the install script as root (sudo will do)
$ ./install.sh
To start berta, there is a script inside berta folder, just execute the run.sh
$ ./run.sh
The default user name and password that are created in the database are:
admin
123
all dependencies are located in a folder during installation named bertaDependancies (installed to the pi users home directory). If for any reason the installation fails, or berta cannot be started, in most cases the the daughter board drivers did not install correctly. To troubleshoot, please try the following steps:
- Restart the raspberry and try starting berta again
- if it still fails, rerun the following scripts found inside the bertaDependancies folder:
cd path/to/bertaDependancies
./seeed-voicecard/ubuntu-prerequisite.sh
./seeed-voicecard/install.sh
after another reboot, berta should be up and running
Helpfull Links: Learn Python: It's a Python 2 tutorial, but for the overview it's enough. [https://developers.google.com/edu/python]
Learn Flask: [https://blog.miguelgrinberg.com/post/the-flask-mega-tutorial-part-i-hello-world]
DeepSpeech: [https://deepspeech.readthedocs.io/en/r0.9/] [https://www.hackster.io/dmitrywat/offline-speech-recognition-on-raspberry-pi-4-with-respeaker-c537e7]