fairfolio
Saved Projects

Lend me your ear! - Real-time Speech Separation and Enhancement using Deep Neural Networks

CSEF · 2023 Computational Systems & Analysis Honorable_mention Award

Raghu Ananya, Raghu Anisha · 9

Overview

Hearing impairment is a serious problem affecting millions of people, making it difficult for them in crowded and noisy environments. Our grandmother, who suffered hearing loss at a young age due to some prescribed fever medication, struggles to make conversation in crowded, noisy settings, unable to isolate voices in a sea of many. Despite wearing hearing aids, she struggles to have a conversation when there is road noise as well as when all of us talk loudly at the dinner table. This is what led us to our project: developing a simple and real-time solution to help people hear better when there is interference from other speakers and noise. While research literature has shown that deep learning can be used for speech separation (remove interference from other speakers) and speech enhancement (remove interference from noise), it was unclear whether a single model can improve the signal quality when there is interference from both speakers and noise. Motivated by the work on Conv-TasNet for efficient speech separation, we investigated if it is possible to use this architecture for improving signal quality when there is interference from other speakers and noise. We developed a new training set that allowed us to train a single model that could remove interference from both sources. With this model, we are able to improve the signal quality by 3dB for interfering speakers and 7 dB for noise. We also developed a real time implementation of the model that can run on a laptop, achieving a latency of 40 ms. This is the first step towards having a solution that can be implemented on a small device like a hearing aid.

Source coverage

This record comes from a published award list, not a complete project archive. Its abstract comes from CSEF's public project showcase as archived by the Internet Archive before judging (https://web.archive.org/web/20230401224130/https://ca-csef.zfairs.com/showcase/ShowcaseInfo?f=838e60b7-ea75-46e8-865c-fde4864244b3); the version presented may differ.

Awards (1)

  • Category Award: HM

Competition history

  • CSEF 2023 Computational Systems & Analysis · Entry S0806

Resources

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

Source: California Science & Engineering Fair public projects

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google