Multiple Voice Separation with Speaker Diarization

Speech Separation is a special scenario of source separation problem and a challenging task. We will present a method to separate a mixed audio sequence, in which multiple speakers speak simultaneously. The classification model will estimate the number of speakers and will train different models for

2025-06-28 16:28:38 - Adil Khan

Project Title

Multiple Voice Separation with Speaker Diarization

Project Area of Specialization Artificial IntelligenceProject Summary

Speech Separation is a special scenario of source separation problem and a challenging task. We will present a method to separate a mixed audio sequence, in which multiple speakers speak simultaneously. The classification model will estimate the number of speakers and will train different models for each speaker. We will evaluate our model under both clean and noisy data. The expected results will be that our model will separate multiple voices and then additionally it will transcribe the speech into text and display the transcribed results with speaker diarization on our website.

Project Objectives

The project is aimed at developing a software that will help individuals and businesses separate and identify voices from any audio. It is also aimed at converting the audio of speakers in form of text. We are expecting our blind source separation model to predict separate multiple voices with at least 50% accuracy and we will try to increase its accuracy as well.

Project Implementation Method

Major application components include

Benefits of the Project

In terms of business, our goals are:

Technical Details of Final Deliverable

This project aims to build a model that will separate multiple voices/signals from a mixed-signal and transcribe the text with speaker diarization.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Education Core Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources
Item Name Type No. of Units Per Unit Cost (in Rs) Total (in Rs)
Total in (Rs) 70000
Gtx1650 Equipment17000070000

More Posts