seblemaguer/RobotVoice

RobotVoice: Verified Set of TTS Voices for Common Robots

RobotVoice

Verified Set of TTS Voices for Common Robots

Run synthesis server

To run locally, install Docker Desktop and run in the folder RobotVoice:

docker-compose up -d --build

You should now be able to test it: http://localhost:5000/

Run synthesis server on a server

Provision at least a m5.xlarge instance on AWS. Install Docker and Docker Compose. Then run:

git clone https://github.com/polvanrijn/RobotVoice
cd RobotVoice
docker-compose up -d --build

If you don't want to use Docker, you can also install the dependencies manually:

git clone https://github.com/polvanrijn/RobotVoice
cd RobotVoice
sh setup.sh
# Set the port and number of workers based on the number of CPUs, set max-requests = 1 to avoid memory leak
gunicorn server:app --worker-tmp-dir /dev/shm --workers=17 -b :5001 -t 600 --max-requests 1

Scraping for robots

Setup a virtual environment and install the dependencies:

pip install opencv-python matplotlib beautifulsoup4 requests

To create a json with all meta information from the robots listed at https://robots.ieee.org/robots/, run:

python scrape_ieee.py 

To select the images for the robots, run the following script.

mkdir "images"
python select_robots.py 

An array with images will pop up. Either specify the image index (starts at 1!) or reject the image by entering n.