Tone Generator with Speech Recognition iOS App
In the Spring of 2012, ECE graduate student Mathias Simeth and I created the "Tone Generator with Speech Recognition" app on the iOS platform for our final project in ECE 420: Digital Signal Processing Lab. We took the source code of Apple's "SpeakHere" application, a simple sound recorder, and modified it to allow the user to be able to say the name of a certain pitch (i.e. 'A', 'B Sharp', 'C Flat', etc.) and then have the app play the pitch back to the user.
It works by using the K-Nearest Neighbor algorithm to find the shortest distance between the Mel-frequency cepstral coefficients of the user's input with a pre-recorded library that was created with MATLAB. The biggest challenge was we were originally going to use a microcontroller to implement the calculations, but due to lack of documentation and hardware issues, we decided to start from scratch and start learning how to program for the iOS platform.
Here is a video from our demonstration:
It works by using the K-Nearest Neighbor algorithm to find the shortest distance between the Mel-frequency cepstral coefficients of the user's input with a pre-recorded library that was created with MATLAB. The biggest challenge was we were originally going to use a microcontroller to implement the calculations, but due to lack of documentation and hardware issues, we decided to start from scratch and start learning how to program for the iOS platform.
Here is a video from our demonstration:
