The aim of this project is to compress speech using a fixed frequency comb algorithm that is inspired in part by the manner in which cochlear implants relay audio information. The technique is capable of high compression ratios; however, the quality and intelligibility of speech reduces significantly. You can experience the change in quality in the two example sentences below. Listen first to the compressed sentence. Can you understand what is being said? The original uncompressed sentence is provided for comparison.

• 'Example 1' Original file:
  
• 'Example 1' Compressed:
  
• 'Example 2' Original file:
  
• 'Example 2' Compressed:
  

After listening to the example sentences try the survey below which plays the compressed sentence and asks you to type in what you hear. We will be using this sort of test to refine the algorithm. Our goal is to find a scheme that balances high compression ratio with good intelligibility.

This project is part of an NSF sponsored REU project entitled "Computational Modeling and Simulation in Applied Sciences" at Middle Tennessee State University.

Go Back