How to configure
To start playing with Adrian the core modules need to be configured. There are three core modules:
- Google Speech to Text Service Module
- Ivona Text to Speech Module
- SnowBoy hotword file configuration (optional)
There are then three main commands to start, stop and update Adrian:
All module configuration changes can be done in the constants.js file
First thing first. The application is located in the folder below
1. Configuring Google Speech to Text Service Module
All users need a Google Speech Service Account which can be obtained online. If you don’t have one yet get it from https://cloud.google.com/speech . Google Speech Service is free up to a certain monthly limit.
Once you have registered you will be issued a: – project_id – google application credential file
The google application credential file can be downloaded at https://console.developers.google.com/project/_/apis/credentials. Once you have your google application credentials file copy it to the ‘/Library/googleSpeech/Account/’ folder. Then Change the constants.js file replacing the values starting with ‘YOUR_’ below:
define("GOOGLE_APPLICATION_CREDENTIALS", __dirname + "/Library/googleSpeech/Account/YOUR_PROJECT_FILE_NAME_COMES_HERE") define("GCLOUD_PROJECT", "YOUR_PROJECTN_NAME_COMES_HERE")
Detailed steps about how to register to set up Google Speech on the link below
Configuring Google Speech
2. Configuring Ivona Text to Speech Module
All users also need an Ivona account. If you dont have one get it at https://www.ivona.com. It is a very easy process, just pick the free plan option and register. Once you have your Ivona credentials change the constants.js repacling the values starting with ‘YOUR_’ below:
define("IVONA_ACCESSKEY", "YOUR_IVONA_ACCESSKEY_COMES_HERE"); define("IVONA_SECRETKEY", "YOUR_IVONA_SECRETKEY_COMES_HERE");
3. SnowBoy hotword file configuration
The great Snowboy offline hotword recognition gives Adrian the ability to recognize and wake up when you say the name Adrian. Every persons pronunciation is different but Adrian must recognize all of us. For this Adrian needs a good configuration. The configuration is an audio module file created by snoboy. if Adrian is not able to recognize you when you say Adrian then you need to record your voice on the showboy website. https://snowboy.kitt.ai/dashboard The whole process takes about 3 minutes but gives almost 100% recognition rate.
Go to the dashboard and search for “Adrian”. Set up your personal model with the microphone icon. Follow the steps and download the output file and replace the file below with the new file. Keep the name of the file.
4. Start Adrian
cd ~/AdrianSmartAssistant ./adrian.sh
5. Stop Appilication
The application can be stopped any time pressing CTRL+C . To then clear memory & kill all depedencies the below command can be run:
cd ~/AdrianSmartAssistant ./stop.sh
6. Update Adrian
in the AdrianSmartAssistant folder execute the below command. Before execution commit all of your local changes.
cd ~/AdrianSmartAssistant ./update.sh