Voice-Analysis-Pipeline

Production ready and scalable pipeline built for Voice(Audio) Analysis which includes Speech-to-Text(S2T) and then Enrichment(Summary, Sentiment, Word Cloud etc.) using Chat GPT.

Project is highly customizable and can be deployed as a standalone microservice on Azure Durable Function App.

A simple illustration of how this Microservice can be used for performing Voice Analysis

Setup

Project was created in Python Version 3.10.12, any version above 3.10 should work fine.

Reference: Quickstart - Python Durable Functions app in V1 programming model

1. python -m venv venv
2. python -m pip install -U pip wheel setuptools uv
3. uv pip install -r requirements.txt
4. Open the project in VS Code and run F5 (Debug command) to Test the function locally

Reusability

1. To add or edit the Analysis Pipeline, make changes in analysis/init.py.
2. Google Speech APIs, Azure Speech APIs, Assembly AI is supported by default for S2T, you can add more S2T services or edit existing from the transcribe folder.
3. Encrichment includes generating Voice Conversation Summary, Overall & Sentences(utterances) sentiment, Wordcloud(excluding more than enough stop words), to generate something else or edit existing by making changes in the metrics folder.

Postman Collection