Here are the steps required to setup a TalkiePi with a Raspberry Pi Zero W and SeeedStudio ReSpeaker 2-Mic Pi HAT. These instructions assume you are using a Windows PC to configure the RPi.
By default, TalkiePi will run without any arguments, it will autogenerate a username and then connect to Daniel Chote's mumble server.
You can change this behavior by appending command line arguments of -server YOUR_SERVER_ADDRESS, -username YOUR_USERNAME to the ExecStart line in /etc/systemd/system/mumble.service once installed.
TalkiePi will also accept arguments for -password, -insecure, -certificate and -channel, all defined in cmd/talkiepi/main.go, if you run your own mumble server, these will be self-explanatory.
- (1) Raspberry Pi Zero W
- (1) SeeedStudio ReSpeaker 2-Mic Pi HAT
- (1) Stereo speaker (3W, 8Ω) with JST connector
- (1) MicroSD card, 16GB, class 10 with SD card adapter
- (1) Power adapter with micro USB cable
- (4) M2.5 Brass Standoffs
- (4) M2.5 x 6 mm screws
Download the latest version of Raspbian Lite; older versions available here
These instructions have been tested with:
Based on: Debian Stretch Lite
Version: November 2018
Release date: 2018-11-13
Kernel version: 4.14
Write the image to a miroSD card with Etcher. Note, after the write is complete Windows will complain that the microSD card needs to be formatted. This message can be ignored. It is just that Windows cannot read the Linux partitions, but Windows can read the Boot partition which is the only one we need access to.
Create two files on the Boot partition of the microSD card:
wpa_supplicant.conf
ssh
Below is a template for the wpa_supplicant.conf file. Please edit these details to match one or more wireless networks you intend to use. You can define as many networks as you like, if using only one network please delete the second network definition. Also, the two-letter country code should match the country you are operating in.
Note, the file must be formatted with Unix style line end characters (i.e. LF not CR LF). Notepad++ is an example of a Windows based text editor that supports this.
ctrl_interface=DIR=/var/run/wpa_supplicant GROUP=netdev
update_config=1
country=AU
network={
id_str="descriptive name for network 2"
ssid="SSID2"
psk="password2"
key_mgmt=WPA-PSK
priority=2
}
network={
id_str="descriptive name for network 1"
ssid="SSID1"
psk="password1"
key_mgmt=WPA-PSK
priority=1
}
The ssh file should be empty.
Insert the microSD card into the RPi and apply power. The RPi will boot and attempt to connect one of the WiFi networks defined above with SSH enabled.
Use Advanced Port Scanner or your router’s web interface to find the IP address of the RPi.
Use PuTTY to make an SSH connection to the RPi.
The default login credentials are:
raspberrypi login: pi
password: raspberry
Update repositories and upgrade software with:
$sudo apt update
$sudo apt upgrade
Start the RPi configuration tool with this command:
$sudo raspi-config
Make the follow configuration changes:
- Change User Password > Ok > YOUR_PASSWORD > YOUR_PASSWORD
- Network Options > N1 Hostname > Ok > YOU_HOSTNAME > Ok
- Localisation Options > I2 Change Timezone > YOUR_COUNTRY > YOUR_CITY
- Localisation Options > I4 Change WiFi Country > YOUR_COUNTRY_CODE > Ok
- Interfacing Options > P4 SPI > Yes > Ok
- Advanced Options > A1 Expand Filesystem > Ok
- Finish > Reboot? Yes
To save power the HDMI output can be disabled at boot.
$sudo nano /etc/rc.local
Add the following lines above exit 0:
# Disable HDMI
/usr/bin/tvservice -o
Install Git with this command:
$sudo apt install git
I was not able to get TalkiePi to work with the latest version (3.1) of seeed-voicecard. If anybody has a solution to this please let me know. In the meantime, we need to install an older version.
Install drivers with these commands:
$wget https://github.com/respeaker/seeed-voicecard/archive/v3.0.tar.gz
$tar -xvzf v3.0.tar.gz
$mv seeed-voicecard-3.0 seeed-voicecard
$cd seeed-voicecard
$sudo ./install.sh 2mic
Edit the ALSA configuration:
$cp asound_2mic.conf ~/.asoundrc
$cd
$nano ~/.asoundrc
Change the two occurrences of "hw:0,0" to "hw:1,0".
The edited file should 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
pcm.!default {
type asym
playback.pcm "playback"
capture.pcm "capture"
}
pcm.playback {
type plug
slave.pcm "dmixed"
}
pcm.capture {
type plug
slave.pcm "array"
}
pcm.dmixed {
type dmix
slave.pcm "hw:1,0"
ipc_key 555555
}
pcm.array {
type dsnoop
slave {
pcm "hw:1,0"
channels 2
}
ipc_key 666666
}
Save (CTRL-O) and exit (CTRL-X).
Reboot:
$sudo reboot
Test the speaker and mic with:
$arecord -f cd | aplay
Test speakers with the following command (CTRL-C to quit).
$speaker-test
Test the microphone with:
$arecord test.wav
$aplay test.wav
$rm test.wav
Audio configuration is done with:
$alsamixer
Install prerequisites with:
$sudo apt install libopenal-dev libopus-dev golang
You can check you Golang version with:
$go version
go version go1.7.4 linux/arm
Setup Golang with:
$cd
$mkdir ~/go
$cd go
$mkdir src
$mkdir bin
$cd
Edit the ~/.profile file with:
$nano ~/.profile
Add the following text to the end of the file:
export GOPATH=$HOME/go
export GOBIN=$HOME/go/bin
export PATH=$PATH:$GOPATH/bin
Reload the configuration with:
$source ~/.profile
Install TalkiePi with:
$cd $GOPATH/src
$go get periph.io/x/periph/cmd/...
$go get github.com/dchote/gopus
$go get github.com/CustomMachines/talkiepi
$go build -o $GOPATH/bin/talkiepi $GOPATH/src/github.com/CustomMachines/talkiepi/cmd/talkiepi/main.go
The RPi Zero needs libopenal complied without ARM NEON support. These packages can be found in the workarounds directory of this repo. They can be installed over the existing libopenal libraries.
$cd ~/go/src/github.com/CustomMachines/talkiepi/workarounds
$sudo apt install ./libopenal-dev_1.17.2-4_armhf.deb
$sudo apt install ./libopenal1_1.17.2-4_armhf.deb
$sudo apt install ./libopenal-data_1.17.2-4_all.deb
$sudo apt install ./libopenal1-dbg_1.17.2-4_armhf.deb
This is optional, mainly if you want to register your talkiepi against a mumble server and apply ACLs.
$cd
$openssl genrsa -aes256 -out key.pem
Enter a simple passphrase, its ok, we will remove it shortly...
$openssl req -new -x509 -key key.pem -out cert.pem -days 1095
Enter your passphrase again, and fill out the certificate info as much as you like, its not really that important if you're just hacking around with this.
$openssl rsa -in key.pem -out nopasskey.pem
Enter your password for the last time.
$cat nopasskey.pem cert.pem > mumble.pem
First test the TalkiePi application with the following command, replacing YOUR_SERVER:PORT and YOUR_USERNAME with your Mumble server credentials:
$$GOPATH/bin/talkiepi -server YOUR_SERVER:PORT -username YOUR_USERNAME -certificate /home/pi/mumble.pem -channel Root
To quit the application press (CTRL+C).
If all is well the Systemd service can be created. Copy the service file to the required folder.
$cd
$sudo cp /home/pi/go/src/github.com/CustomMachines/talkiepi/conf/systemd/mumble.service /etc/systemd/system/mumble.service
Edit the service file with:
$sudo nano /etc/systemd/system/mumble.service
Replace YOUR_SERVER:PORT and YOUR_USERNAME with your Mumble server credentials.
[Unit]
Description = Mumble Client
Requires = systemd-user-sessions.service network.target sound.target
After = multi-user.target
[Service]
User = pi
Group = pi
Type = simple
ExecStart = /home/pi/go/bin/talkiepi -server YOUR_SERVER:PORT -username YOUR_USERNAME -certificate /home/pi/mumble.pem -channel Root
Restart = always
RestartSec = 5
[Install]
WantedBy = multi-user.target
Enable the Systemd service with:
$sudo systemctl enable mumble.service
To start the service without rebooting run:
$sudo systemctl daemon-reload
$sudo systemctl restart mumble.service