1. ICNS Home
  2. Projects
  3. Open PhD Projects
  4. Machine Learning-Based Tool for Therapists to Monitor Speech Progress in Late Talkers

Machine Learning-Based Tool for Therapists to Monitor Speech Progress in Late Talkers

Supervisors:

Primary supervisor Dr Saeed Afshar

Description:

Working with late talkers, therapists often rely on subjective observations and manual tracking methods to monitor speech development. This approach can be time-consuming and may lack the necessary precision and consistency. This project proposes the development of an automated machine learning-based tool designed to analyze and track the speech of both the therapist and the child during therapy sessions.

The proposed system will use speech recognition and natural language processing algorithms to transcribe and analyze the content, clarity, and complexity of the child's speech. It will also track the speech patterns and strategies used by the therapist. Over time, the tool will monitor changes and improvements in the child's speech, providing data-driven insights into their development.

With the aid of this tool, therapists can better understand the effectiveness of their therapeutic strategies, adjust their approaches based on the child's progress, and provide more informed updates to families.

Outcomes:

The project's aim is to develop an automated tool, based on machine learning principles, which can be used by therapists to track and monitor the speech of late talkers and assess improvements over time. This system will provide quantitative, objective feedback to therapists and families about the child's speech progress. This will include the following tasks:

  • Literature Review: Conduct a thorough review of current research in late talker speech therapy, speech recognition, natural language processing, and machine learning models for speech analysis.
  • Tool Design: Design the machine learning-based tool for automated speech tracking and analysis. This will involve defining the key speech metrics to track and developing suitable machine learning models for speech recognition and analysis.
  • Tool Implementation: Implement the designed tool using appropriate machine learning frameworks and software tools.
  • Validation and Optimization: Test the tool in real therapy sessions with late talkers, and adjust and optimize the machine learning models based on their performance.
  • Evaluation: Assess the performance of the developed tool in terms of accuracy of speech tracking, sensitivity to improvements, and its utility to therapists.
  • Communication and Publication: Document the results in a format suitable for publication in a scientific journal. Also, present the findings at relevant conferences and workshops.

Eligibility criteria:

Experience in Python or other languages commonly used in machine learning is necessary. Familiarity with speech recognition models, natural language processing techniques, and machine learning frameworks is beneficial. Knowledge of late talker speech therapy and methods of speech analysis would also be advantageous.