Preconditions

Install these software first:

• Node.js: Follow the instructions on the page https://pimylifeup.com/raspberry-pi-nodejs/
• Python 3: execute this commands sudo apt update && sudo apt install python3 idle3

Setup Software

Login to your Raspberry Pi and solve the following steps.

1. Configure Audio

Install ac108 codec

Your Raspberry Pi needs the ac108 codec to make the ReSpeaker work. Copy and paste the commands to your console.

Update and upgrade Raspberry Pi

sudo apt-get update
sudo apt-get upgrade -y

This may take a while.

Go to pi’s home directory

cd ~

Clone seeed-voicecard

git clone https://github.com/respeaker/seeed-voicecard.git

Change directory

cd seeed-voicecard

Install seeed-voicecard

sudo ./install.sh

Reboot

sudo reboot

The driver should now be installed.

Set Default Audio Device

Next you need to edit the file /etc/asound.conf to make the ReSpeaker as your default audio device.

Use nano or any editor: sudo nano /etc/asound.conf

Edit pcm.!default:

1. change type asym to type pulse
2. add hint.description "Default Audio Device"

Add this code:

ctl.!default {
    type pulse
}

The file /etc/asound.conf should now look like this:

# The IPC key of dmix or dsnoop plugin must be unique
# If 555555 or 666666 is used by other processes, use another one

# use samplerate to resample as speexdsp resample is bad
defaults.pcm.rate_converter "samplerate"

pcm.!default {
    type pulse
    playback.pcm "playback"
    capture.pcm "ac108"
    hint.description "Default Audio Device"
}

pcm.playback {
    type plug
    slave.pcm "hw:ALSA"
}

# pcm.dmixed {
#     type dmix
#     slave.pcm "hw:0,0"
#     ipc_key 555555
# }

pcm.ac108 {
    type plug
    slave.pcm "hw:seeed4micvoicec"
}                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         # pcm.multiapps {
#     type dsnoop
#     ac108-slavepcm "hw:1,0"
#     ipc_key 666666
# }

ctl.!default {
    type pulse
}

Set output device

To be able to hear the voice output on the speaker, set the 3.5mm jack to default.

sudo raspi-config
# Select 1 System Options
# Select S2 Audio
# Select 0 Headphones
# Select Finish

Read more about audio configuration here

Test mic functionally

Check if the mic is installed correctly:

arecord -L

Please check if there is an ac108 device and a device named seeed4micvoicec. You can make a test record of 5 seconds with this command:

arecord -Dac108 -f S32_LE -r 16000 -c 4 -d 5 hello.wav

Play the command with:

aplay hello.wav

If you could hear your recording everything works! Great!

2. Install a MQTT-Server

Install Mosquitto and Mosquitto-Clients:

sudo apt-get install mosquitto mosquitto-clients -y

3. Update Node.js and install Node-RED

Make sure you have installed Node.js. If not, follow these instructions first: https://pimylifeup.com/raspberry-pi-nodejs/

Then you need to download and execute a script to update Node.js and install Node-RED.

Go to pi’s home directory

cd ~

Download Script to update Node.js and install Node-RED

wget https://raw.githubusercontent.com/node-red/linux-installers/master/deb/update-nodejs-and-nodered

Make the downloaded file executable

sudo chmod 755 ./update-nodejs-and-nodered

Execute the file

./update-nodejs-and-nodered

Notes

• You will be asked some questions. Always answer with y (for YES)
• Wait until the script is done
• If you get the message “rm: cannot remove ‘/usr/bin/update-nodejs-and-nodered’: No such file or directory” the folder could be in your actually location. You can delete the folder by yourself.

Enable Node-RED at boot

To start Node-RED at boot run this command:

sudo systemctl enable nodered.service

Now start Node-RED by restarting your Raspberry Pi or with this command:

sudo systemctl start nodered.service

Finished!

The Node-RED server is now accessible at the IP of your Raspberry PI under port 1880:

http://<ip-adress>:1880

Test it!

4. Install Zigbee2MQTT

To install Zigbee2MQTT follow this script:

Clone Zigbee2MQTT repository

sudo git clone https://github.com/Koenkk/zigbee2mqtt.git /opt/zigbee2mqtt

Set rights to user pi

sudo chown -R pi:pi /opt/zigbee2mqtt

Enter folder

cd /opt/zigbee2mqtt

Install python env

python3 -m venv .

Activate environment

source /opt/zigbee2mqtt/bin/activate

Upgrade pip, wheel and setuptools

pip install --upgrade pip wheel setuptools

Install node environment

pip install nodeenv

Init node environment

nodeenv -p -n 16.15.0

Deactivate and activate environment to be sure

deactivate
source /opt/zigbee2mqtt/bin/activate

Install dependencies

cd /opt/zigbee2mqtt
npm ci

Deactivate environment

Enter deactivate to deactivate the environment.

Start and test Zigbee2MQTT

Start Zigbee2MQTT and test if the connection works.

Tip: Use the MQTT explorer to see incoming messages.

Enter folder

cd /opt/zigbee2mqtt

Activate environment

source /opt/zigbee2mqtt/bin/activate

Start

npm start

Now you can turn on and off your Zigbee device to see if everything works. An output could look like this:

...
Zigbee2MQTT:info  2020-11-17 11:32:44: MQTT publish: topic 'zigbee2mqtt/0x00158d000520f47d', payload '{"brightness":255,"color_temp":153,"linkquality":118,"state":"ON"}'
...

MQTT error

If you get the error MQTT error: Connection refused: Not authorized you probably set credentials for the mqtt broker.

Go to the data folder of zigbee2mqtt:

cd /opt/zigbee2mqtt/data

Open the file configuration.yml with an editor of your choice like nano, vi or vim.

Remove the comment parameter # in the line of user and password and add your credentials there.

Quit

Press ctrl + c to quit.

Deactivate environment

Enter deactivate to deactivate the environment.

Enable Zigbee2MQTT at boot

To start Zigbee2MQTT at boot create a service:

Execute sudo nano /etc/systemd/system/zigbee2mqtt.service and copy and paste the following content:

[Unit]
Description=zigbee2mqtt
After=network.target

[Service]
ExecStart=/bin/bash -c 'source /opt/zigbee2mqtt/bin/activate; /opt/zigbee2mqtt/bin/npm start'
WorkingDirectory=/opt/zigbee2mqtt
StandardOutput=inherit
StandardError=inherit
Restart=always
User=pi

[Install]
WantedBy=multi-user.target

!Ensure that you use the correct user!

Enable the new service with

sudo systemctl enable zigbee2mqtt.service

Now start the service by restarting your Raspberry Pi or with this command:

sudo systemctl start zigbee2mqtt.service

Frontend

Zigbee2MQTT brings an own optional frontend. You can activate it in the configuration.yml.

• Go to the folder /opt/zigbee2mqtt/data
• Edit the file configuration.yml
• Add the line frontend: true at the end of the file
• Save and exit the editor
• Restart the service with sudo systemctl restart zigbee2mqtt.service

The frontend will run on port 8080. If you want to change the port, use this code instead of frontend: true:

frontend:
  # Optional, default 8080
  port: 8080
  # Optional, default 0.0.0.0
  host: 0.0.0.0
  # Optional, enables authentication, disabled by default
  auth_token: your-secret-token
  # Optional, url on which the frontend can be reached, currently only used for the Home Assistant device configuration page
  url: 'https://zigbee2mqtt.myhouse.org'

Read more about the frontend configuration on the zigbee2mqtt documentation: https://www.zigbee2mqtt.io/guide/configuration/frontend.html

5. Optional install Hermes Led Control for ReSpeaker LEDs

Hermes Led Control controls the LEDs of the ReSpeaker device. For example, when you say the Wake Word, the LEDs turn on. With this setting you always can see when the ReSpeaker is active.

Go to pi’s home directory

cd ~

Download script

wget https://gist.github.com/Sh4der/8476895a0863ca6062103315ac122954/raw/f5c83b34eaafbd0b315d17282c1fc52b672e008a/hlc_download.sh

Make the script executable

sudo chmod +x hlc_download.sh

Execute the script

sudo ./hlc_download.sh

Answer the questions like following:

• What assistant engine are you using? Enter 2 for Rhasspy
• What’s the path to your assistant config file? Press Enter (Path Should be like (/home/pi/.config/rhasspy/profiles/de/profile.json)
• What device do you wish to control with SLC? Enter 2 for respeaker4
• What pattern do you want to use? Choose your preferred Led pattern. For example 1 for google
• Do you need to install / configure your respeaker4? Enter 2 for no

Later you will configure Rhasspy to work with Hermes Led Control.

Read more about Hermes Led Control

6. Install Rhasspy native

Go to pi’s home directory

cd ~

Download Rhasspy V2.5.7 to

You can get the latest version from github

Download with command:

wget https://github.com/rhasspy/rhasspy/releases/download/v2.5.7/rhasspy_2.5.7_armhf.deb

Make the downloaded file executable

Install Rhasspy with command:

sudo apt install ./rhasspy_2.5.7_armhf.deb -y

Edit profile.json

Get the open the profile.json file with the command:

sudo nano /home/pi/.config/rhasspy/profiles/de/profile.json

Find the entry “microphone” and change it to:

"microphone": {
  "system": "arecord",
  "arecord": {
    "device": "sysdefault:CARD=seeed4micvoicec"
  }
}

Start Rhasspy

Start Rhasspy and check if rhasspy runs without problems.

rhasspy --profile de

If an error occurs like PermissionError: [Errno 13] Permission denied: '/home/pi/.config/rhasspy/profiles/de' you need to give your user permission to the folder /home/pi/.config/rhasspy

sudo chown -R pi:pi /home/pi/.config/rhasspy

Start Rhasspy as service (english)

Create the file /etc/system.d/system/rhasspy.service and put the following code inside:

[Unit]
Description=Rhasspy
After=syslog.target network.target seed-voicecard.service sound.target

[Service]
Type=simple
WorkingDirectory=/home/pi/.config/rhasspy
ExecStart=/bin/bash -lc 'rhasspy --profile en 2>&1 | cat'
User=pi
Group=pi

RestartSec=1
Restart=on-failure

StandardOutput=syslog
StandardError=syslog

SyslogIdentifier=rhasspy

[Install]
WantedBy=multi-user.target

Start Rhasspy as service (german)

Create the file /etc/system.d/system/rhasspy.service and put the following code inside:

[Unit]
Description=Rhasspy
After=syslog.target network.target seed-voicecard.service sound.target

[Service]
Type=simple
WorkingDirectory=/home/pi/.config/rhasspy
ExecStart=/bin/bash -lc 'rhasspy --profile de 2>&1 | cat'
User=pi
Group=pi

RestartSec=1
Restart=on-failure

StandardOutput=syslog
StandardError=syslog

SyslogIdentifier=rhasspy

[Install]
WantedBy=multi-user.target

Have a look on the rhasspy.service file

• english rhasspy.service
• german rhasspy.service

Now enable the service:

systemctl enable rhasspy

Start it with:

systemctl start rhasspy

You can stop it with:

systemctl stop rhasspy

After saving settings in Rhasspy you need to restart rhasspy manual in the console with the command:

systemctl restart rhasspy

(6). Install Rhasspy with Docker (not recommend)

To install Rhasspy pull the Docker image with the command

docker pull rhasspy/rhasspy

This may take a while.

Run Rhasspy Docker image

Run the Rhasspy docker image with the command

docker run -d -p 12101:12101 \
      --name rhasspy \
      --restart unless-stopped \
      -v "$HOME/.config/rhasspy/profiles:/profiles" \
      -v "/etc/localtime:/etc/localtime:ro" \
      --device /dev/snd:/dev/snd \
      rhasspy/rhasspy \
      --user-profiles /profiles \
      --profile de

Have a look to more useful docker commands

Finished!

Rhasspy is now accessible at the IP of your Raspberry PI under port 12101. http://<ip-adress>:12101. Test it!

7. Install DeepSpeech 0.7.4

DeepSpeech is the Speech to Text program for Rhasspy in this project. You need to install it differently than the Rhasspy documentation says. Otherwise, it will not work.

Install choices

You have different choices to install DeepSpeech:

• Use the install_rhasspy_deepspeech.sh script
• Alternative installation step-by-step

Auto installation script

You can install DeepSpeech with a pretrained english or german model.

English pretrained model

The DeepSpeech installation script for an english pretrained model is available in our git repository: install_rhasspy_deepspeech_en.sh

Download script:

wget -N https://raw.githubusercontent.com/th-koeln-intia/ip-sprachassistent-team4/master/docs/scripts/install_rhasspy_deepspeech_en.sh -O $HOME/tmp/install_rhasspy_deepspeech.sh

Make script executable:

sudo chmod +x $HOME/tmp/./install_rhasspy_deepspeech.sh

Execute script: $HOME/tmp/./install_rhasspy_deepspeech.sh

German pretrained model

The DeepSpeech installation script for a german pretrained model is available in our git repository: install_rhasspy_deepspeech_de.sh

Download script:

wget -N https://raw.githubusercontent.com/th-koeln-intia/ip-sprachassistent-team4/master/docs/scripts/install_rhasspy_deepspeech_de.sh -O $HOME/tmp/install_rhasspy_deepspeech.sh

Make script executable:

sudo chmod +x $HOME/tmp/./install_rhasspy_deepspeech.sh

Execute script: $HOME/tmp/./install_rhasspy_deepspeech.sh

Alternative installation step-by-step

We are using german pretrained model files for deepspeech 0.7.4 from this repository: AASHISHAG/deepspeech-german.

You need the files output_graph.tflite and kenlm.scorer. You can download them manually from this google drive folder. Save then into the folder .config/rhasspy/profiles/de/deepspeech/models/.

If the folders don’t exist, create them:

mkdir ~/.config/rhasspy/profiles/de/deepspeech
mkdir ~/.config/rhasspy/profiles/de/deepspeech/models

You also can download the files with the following commands:

wget -N --no-check-certificate 'https://docs.google.com/uc?export=download&id=1lyOFCfrxiTwXotmeWs1hdm_Amg3J_y1T' -O ~/.config/rhasspy/profiles/de/deepspeech/models/output_graph.tflite
wget -N --load-cookies /tmp/cookies.txt "https://docs.google.com/uc?export=download&confirm=$(wget -N --quiet --save-cookies /tmp/cookies.txt --keep-session-cookies --no-check-certificate 'https://docs.google.com/uc?export=download&id=1mrfMSYp_mYrsLswttY_fvfAHaJ7azahC' -O- | sed -rn 's/.*confirm=([0-9A-Za-z_]+).*/\1\n/p')&id=1mrfMSYp_mYrsLswttY_fvfAHaJ7azahC" -O ~/.config/rhasspy/profiles/de/deepspeech/models/kenlm.scorer && rm -rf /tmp/cookies.txt
wget --no-check-certificate 'https://docs.google.com/uc?export=download&id=1zIvpdJ0YOreg_HwVwoyt_6BEEusH5g4w' -O ~/.config/rhasspy/profiles/de/deepspeech/models/alphabet.txt

Now you need to install a custom version of rhasspy-asr-deepspeech and rhasspy-asr-deepspeech-hermes.

Execute the following commands:

sudo git clone https://github.com/Sh4der/rhasspy-asr-deepspeech /opt/rhasspy-asr-deepspeech
sudo git clone https://github.com/Sh4der/rhasspy-asr-deepspeech-hermes /opt/rhasspy-asr-deepspeech-hermes
sudo chown -R pi:pi /opt/rhasspy-asr-deepspeech
sudo chown -R pi:pi /opt/rhasspy-asr-deepspeech-hermes

wget https://github.com/th-koeln-intia/ip-sprachassistent-team4/raw/master/docs/tools/deepspeech_train_tools.tar.gz -O /opt/rhasspy-asr-deepspeech-hermes/deepspeech_train_tools.tar.gz
tar xvfz /opt/rhasspy-asr-deepspeech-hermes/deepspeech_train_tools.tar.gz -C /opt/rhasspy-asr-deepspeech-hermes
rm /opt/rhasspy-asr-deepspeech-hermes/deepspeech_train_tools.tar.gz

sudo apt install llvm libfst-tools libngram-tools

mkdir /opt/rhasspy-asr-deepspeech-hermes/.venv
virtualenv -p python3 /opt/rhasspy-asr-deepspeech-hermes/.venv
source /opt/rhasspy-asr-deepspeech-hermes/.venv/bin/activate

pip install /opt/rhasspy-asr-deepspeech/
pip install /opt/rhasspy-asr-deepspeech-hermes/
pip install progressbar2==3.47.0
pip install /opt/rhasspy-asr-deepspeech-hermes/tools/packages/*.whl

sudo ln /opt/rhasspy-asr-deepspeech-hermes/bin/rhasspy-asr-deepspeech-hermes /usr/bin/rhasspy-asr-deepspeech-hermes

sudo wget https://raw.githubusercontent.com/th-koeln-intia/ip-sprachassistent-team4/master/docs/scripts/rhasspy-asr-deepspeech-hermes.service -P /etc/systemd/system/
sudo systemctl enable rhasspy-asr-deepspeech-hermes.service
sudo systemctl start rhasspy-asr-deepspeech-hermes.service
deactivate

After going through these commands, you have a service named rhasspy-asr-deepspeech-hermes.

8. Install Snips-NLU

Install choices

You have different choices to install Snips-NLU:

• Use the snips-nlu-install.sh script
• Alternative installation step-by-step

Auto installation script

The Snips-NLU auto installation script is available in english and german.

English auto installation script

You can install Snips-NLU with the english install_rhasspy_snips_nlu_en.sh or with a virtual environment install-rhasspy-snips-nlu_venv_en.sh script. Change directory to home folder: cd $HOME

Download script:

wget https://github.com/th-koeln-intia/ip-sprachassistent-team4/raw/master/scripts/rhasspy_snips_nlu_hermes_venv_en.service -O $HOME/tmp/install_rhasspy_snips_nlu_venv.sh

Make script executable:

sudo chmod +x ./install_rhasspy_snips_nlu_venv.sh

Execute script:

./snips-nlu-install.sh

German auto installation script

You can install Snips-NLU with the german install_rhasspy_snips_nlu_de.sh or with a virtual environment install-rhasspy-snips-nlu_venv_de.sh script.

Change directory to home folder: cd $HOME

Download script:

wget https://github.com/th-koeln-intia/ip-sprachassistent-team4/raw/master/scripts/rhasspy_snips_nlu_hermes_venv_de.service -O $HOME/tmp/install-rhasspy-snips-nlu_venv.sh

Make script executable:

sudo chmod +x ./install-rhasspy-snips-nlu_venv.sh

Execute script:

./snips-nlu-install.sh

Alternative installation step-by-step

1. Install Rust:

   sudo apt install rustc -y
   

2. Install Setuptools plugin for Rust extensions:

   pip3 install setuptools-rust
   

3. Install Snips-NLU:

   pip install snips-nlu
   

4. Do Snips-NLU rebirth

4.1 sudo apt install libatlas3-base libgfortran5

1. Install Rhasspy Snips-NLU Hermes pip3 install rhasspy-snips-nlu-hermes

2. Start Client for MQTT-Server rhasspy-snips-nlu-hermes --language de --engine-path /home/pi/.config/rhasspy/profiles/de/snips/engine --dataset-path /home/pi/.config/rhasspy/profiles/de/snips/dataset.yaml --host 127.0.0.1 --port 1883

Resources:

• https://github.com/snipsco/snips-nlu
• https://github.com/jr-k/snips-nlu-rebirth
• https://rhasspy.readthedocs.io/en/latest/intent-recognition/#snips-nlu

Setup completed!

Congratulations! You finished the installation of all necessary software. Next you will configure Rhasspy.