VoiceCoachReact

Realtime AI Voice Coach is the ReactJS web application that can stream microphone audio from the browser to a NodeJS server and transmit the DeepSpeech results back to the browser.

Video Tutorial

For Mac users or developers, here is a new version tutorial

For Mac users or developers, here is the tutorial

For Windows users or developers, here is the tutorial

Setup

To launch to Realtime AI Voice Coach to your local machine, you need to do set up the following

1. Download the GitHub repository to your local machine

In our github repository, you can either download the zip file or open with GitHub Desktop.

2. Download the following pre-trained models through Mozilla DeepSpeech:

• deepspeech-0.9.3-models.pbmm
• deepspeech-0.9.3-models.scorer

You need to move these 2 files into the VoiceCoachReact file on you local machine.

3. Install Node js

Download the correct version of node.js

4. Install npm and yarn

Search Terminal(Mac) / Command Prompt(Windows), copy the following prompt line by line and hit enter to install

Install npm

npm install

Install yarn. Depends on your environment, you might need to try the following arguments until it can successfully download

npm install -g yarn

sudo npm install -g yarn

npm install --global yarn

---

Run the Website

1. Open

You need to open TWO Terminal(Mac) / Command Prompt(Windows) windows

2. Go to the directory of the VoiceCoachReact file

The following is the example if your VoiceCoachReact file on you local machine is in Downloads

cd Downloads/VoiceCoachReact

3. Run ReactJS Client website

On one of the terminal / command prompt, run ReactJS Client to start the website:

yarn start

4. Run NodeJS Server

On another terminal / command prompt, Run NodeJS Server to activate the server to interact with the client website:

node server.js

5. Remember to TURN ON you microphone

---

Overview of the VoiceCoachReact file

• server: server.js in VoiceCoachReact
• The UI(User Interface) with html, css, JavaScript in VoiceCoachReact/src
• .pbmm and .scorer are the DeepSpeech pre-trained model

---

Additional / Further Infomation

• Train your own model with DeepSpeech
• DeepSpeech Documentation