EMNS-DCT: An Audio Remote Creation and Validation Tool for Researchers

EMNS-DCT is a Django-based web tool that enables researchers to remotely create and validate audio utterances. The tool is designed to simplify the process of generating high-quality audio data for research projects, reducing the need for expensive recording equipment and extensive in-person collaboration. The tool uses Docker for deployment, allowing for easy installation and configuration across different environments.

Prerequisite

The EMNS-DCT tool requires the following prerequisites to be installed:

• Docker
• docker-compose

Getting Started

Enviroment Variables

Before running the tool, you need to set the environment variables in the env_variables.conf file. The following environment variables are required:

• SECRET_KEY: A secret key for Django
• ALLOWED_HOSTS: A comma-separated list of allowed hosts
• EMAIL_USER: Email ID for sending email notifications
• EMAIL_PASS: Password for email account

Docker

Docker is used for deploying the tool. Once you have Docker and docker-compose installed, you can build and run the tool using the provided Makefile. Here are some useful commands to get started:

• make build: Build the Docker container
• make run: Start the Docker container
• make restart: Restart the Docker container
• make down: Stop and remove the Docker container

Debug Mode modifying the tool

To make debugging and modifying the tool easier, we have created a separate compose file. You can access this mode by executing the following command:

make run_debug

This command will enable the debug mode and start the tool. With this mode, you can easily modify the tool's code and see the changes in real-time. It is important to note that this mode is intended for development purposes only and should not be used in a production environment.

Reset migrations

If you want to reset the migrations, you can use the following commands:

find . -path "*/*/migrations/*.py" -not -name "__init__.py" -delete
find . -path "*/*/migrations/*.pyc"  -delete
find . -path "*/db.sqlite3"  -delete
docker exec -it emns-dct_app_1 bash
python3 manage.py makemigrations
python3 manage.py migrate
python3 manage.py createsuperuser

Collect static file

If there are changes to static files, run the following command to collect them:

python3 manage.py collectstatic

Create Utterances

To create utterances for recording, validation, etc., use the following commands. The function expects a .tsv file containing the transcript and emotion.

docker exec -it emns-dct_app_1 bash
python3 manage.py shell

from audio_recorder.models import Utterances
from django.contrib.auth.models import User
import random
import pandas as pd


def create_utterance(user, utterance, prosody):
 post = Utterances( utterance=utterance,
      prosody=prosody,
      author=user)
 post.save()

def main(csv_dir, username, sep="\t", emotions=None, header=None):
 if emotions==None:
  emotions = ['Happy', 'Sad', 'Angry', 'Excited', 'Sarcastic', 'Neutral', 'Disgust', 'Surprised']
 
 user = User.objects.filter(username=username).first()
 df = pd.read_csv(csv_dir, sep=sep, header=header)
 
 for i, row in df.iterrows():
  if not row.isnull().any():
   create_utterance(user, row[0], row[1])
  else:
   create_utterance(user, row[0], random.choice(emotions))
   print("Found nan: ", i)

main("sample.tsv")

Start container

After you have created the utterances, you can start the container with the following command:

make run

If you want to run in debug mode you can use the following command:

docker-compose -f docker-compose.yml up

Create ssl certificate

Change proxy/default.conf to your domain name. you must also rebuild the proxy:

cd proxy
make no-cache

To create an SSL certificate, you can use the following commands:

docker exec -it emns-dct_proxy_1 sh

certbot --nginx --noninteractive --agree-tos -m knoriy72@gmail.com -d dct.openempathic.ai -d www.dct.openempathic.ai --redirect --test-cert

Remove --test-cert after confirming that everything loaded as expected.