FICS 2024 Registrations are now open! Register Now
Idea - FICS

Beyond Words

Team and Contact Details

Student Name School Degree Year Email
Huma AmeerSEECSMS/M.PhilThird[email protected]
Saher SohailSEECSMS/M.PhilThird[email protected]

Inter School Idea ? No
Do you need expertises from another area: No
If Yes please provide details of expertises you need:

Idea Details

Idea Name: Beyond Words
Slogan: Because Every Utterance Counts
Supervisor Name: Dr. Seemab Latif
Supervisor Designation: Associate Professor
Supervisor School: SEECS
Supervisor Department: Department of Computing
Contact number: 03235555451
Email ID: [email protected]
Abstract:
    In Pakistan, approximately 23 mil people are suffering from communication and articulation disorders [1]. Furthermore, going to a therapist is time consuming and costly . We propose an Ai based solution that detects speech disfluencies using audio of 5 to 10 minutes sessions.
What is the unmet need in society that your idea will fulfill ?
    The assessment of stuttering is a time consuming task for the therapists. Moreover, for patients, either therapists are not available in their vicinity, or they are not affordable. Hence, the idea will provide a solution to the above problems and target Urdu and English speakers.
Who needs it ? How many would benefit ?
   Considering the bigger picture of this project that will eventually become a service product for therapists and patients. The product will aid the therapists for time efficient and accurate diagnosis of the disorder. A progress monitoring application which will be more accessible to the patients.
How will the solution works
    The respective problem is solved using deep learning models. The multi-stuttering dataset has been curated from a publicly available dataset i.e. SEP-28K. Secondly, the speech representation of the audio input has been investigated using state-of-the-art models . The dataset is fine-tuned on pre-trained models. Thirdly, the experimentation includes freezing/unfreezing layers of wav2vec2 on English, German and Urdu audios to identify the most accurate speech representation.
Who are your competitors ? How is your solution different
    Our competitors are such application development teams, that can assist the therapists in diagnosis of stuttering. As in Pakistan, the sessions are mostly conducted in Urdu language, we aim to develop an intelligent system that will cater the needs of our people.
Status: new
Entry Date & Time: 2022-12-15 (1740)