Software
The primary goal of the testing and experiments is to have our system produce the highest STI and SII values possible. Mohammad Raza developed developed various tools with MATLAB that work in conjuction with STI measuring software to produce the STI and frequency response of our system based on recorded audio files that have been fed through the various components of the system in the Spring 2009 semester. The programs work by breaking the signal up into its various frequency components, and using key frequencies to determine whether or not the signal is likely to be intelligible by a human.



The first program that was developed is able to take various audio files and automatically generate the frequency spectrum of each file. This is useful for our team because it allows us to look at various noise sources and see what frequencies might be filtered out to improve intelligibility.

The second program was developed to help with our testing process. An interesting error happened with third party testing software where the files that were produced by our system were out of sync by fractions of a second, and this caused the software to fail when trying to produce the STI of the signal. This program used an algorithm to reposition the audio signal within the file, which allowed the third party software to successfully compute the STI.

To analyse our recorded audio files to determine the STI we used LexSTI

We used TrueRTA for sound level reference for test stability. TrueRTA determines real SPL (sound pressure levels) to set a standard.

To record and edit our test files we used the audio software Audacity. MATLAB scripts were used for ANAA (ambient noise aquisition and analysis) testing and stimulus response capture (sends out and records stimuli for analysis).