Installation
How to install the developer version of Parula on your Linux PC. (A Windows installer and Android version for end-users are planned for the future.)
Hardware dependencies
Microphone: You must have a good microphone. Typical computer/notebook microphones are not sufficient.
For example, the Sony PlayStation 3 Eye camera or a conference microphone.
Loud speaker, obviously.
Normal desktop CPU.
Raspberry Pi 4 is planned.
Install OS dependencies
Ubuntu, Debian etc.
# apt install yarnpkg mpg123 mpd sox libasound2-dev default-jre-headless default-jdk git
Fedora, RedHat etc.
Install RPM Fusion (for mpd)
# dnf install https://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm
# dnf install mpg123 mpd sox alsa-lib-devel java-1.8.0-openjdk-headless java-1.8.0-openjdk-devel git
Install Yarn
Complete step 3 below.
$ npm install -g yarn
Mac
# brew node yarn sox mpd mpg123 openjdk git
Skip step 3.
Install node.js
Download node version 14.x.
node version 16 and 18 do not work.
Extract the archive somewhere.
ln -s
/your/path/to/node//usr/local/bin/node
Install MaryTTS
git clone https://github.com/marytts/marytts/
Build it with
./gradlew
Install voices by running``wget
grep location.*cmu-slt download/marytts-components.xml|sed -e "s|.*href=\"||" -e "s|\".*||"```
unzip voice-cmu-slt-5.2.zipor alternatively starting the UI app
./gradlew runInstallerGui`en-US
cmu-slt
en-US
cmu-rms
en-GB
dfki-spike
en-GB
dfki-prudence
Restart the server:
./gradlew run
Open http://localhost:59125 in a browser, to check whether the server is running and the voices are listed in the dropdown.
Download data files
Download the Parula app data files
cd
to the Parula source directorymkdir data
cd data/
tar xjf /path/to/parula-data.tar.bz2
mkdir deepspeech-0.9.3-models
The DeepSpeech model version needs to match the version of DeepSpeech npm module perfectly.
cd deepspeech-0.9.3-models/
Download the DeepSpeech 0.9.3 model file and DeepSpeech 0.9.3 scorer file into this directory.
They are 1.2 GB, so while they download, you can already adjust config.json, see step 6.3. below.
Install node packages
cd
to the Parula source directoryyarn install
cp config-min.json config.json
Adapt
config.json
Run Parula
cd
to the Parula source directoryyarn start
You should see
Say "Hey Edison, what time is it?"
Find more commands to say by looking at
app/
*/intents.en.json
Configure the applications
Create your own voice apps :-)
Implement your voice commands in JavaScript, with the help of npm modules.
Last updated