Overview:
To assess people with communication disorders clinicians need better insights into their daily life and it is a laborious and time-consuming task. We have created an API that extract features from conversation such as average percentage spoke in conversation, taking turns in conversation, total pauses during a conversation, and many other speaker metrics. The goal of this API is to integrate into a user-friendly system so that users can easily analyze a conversation and have an intuition of individuals' involvement in communication and daily life.
Tool installment:
Pre-requisites:
-
Programming language : Python (3.8 or above) - https://www.python.org/ftp/python/3.8.10/python-3.8.10-amd64.exe - windows , https://www.python.org/ftp/python/3.8.10/python-3.8.10-macosx10.9.pkg - MacOS.
-
Packages : Pandas,numpy,nltk
-
IDE : Any python supported IDE(we used PyCharm).
-
API Testing Tools : Postman
-
A file with 'wav','awb','m4a','mp3','mp4' format.
Steps
-Update the file location of the input file with your local directory in app.py
- Google Speech API:
- Go to your google cloud console.
- Enable google speech to text api.
- Get the credential json file by creating the key in service accounts(IAM & ADMIN -> service accounts) and replace the code in client_secret.json (Conversation-Moderator-API->resources->client_secret.json).
-
Create the virtual environment(if not created) for the application as below:
pip install virtualenv virtualenv my_env
-Then install the requiremnents:-
pip install -r requirements.txt
Execution:
• Execute the python application in the virtual environment that is created. • The application will be running in the local host.
Output:
• Try hitting the local url in postman by importing the following collection - https://www.getpostman.com/collections/8246194f7189bdfc8072
• Upload the file in the body of the postman as below -
• We can observe a successful run of the API which returns a JSON with the detailed output in the postman dashboard.
• The required result will be shown in the terminal.
