Goal: Analyze a Wave File Generating and Saving WAV files Sound Recorder; Volume Controls (watch out for mute); Play back to test quality… Setting up folders Work in pairs; 1 microphone Record all vowels for each person Suggested filenames: fred_a.wav in folder named fred Math involved here: Sampling rate, duration => filesize Should be able to predict file size 3.5 seconds, 22,050 samples per sec, mon0 => what size? 3.5 x 22,050 = 77175 bytes Fourier Series Creation of periodic functions Series of sines and cosines or exponentials to “build” signals Orthogonal building blocks see Table Sines and Cosines Exponentials We want to use components that have nothing in common = orthogonal Inner product or dot product is zero Trigonometric Fourier Series Exponential Fourier Series Limit of Complex EFS creates Fourier Transform Frequency Domain (Hertz) Transform takes us from the time domain to the frequency domain Fourier Transform Function in Matlab, command fft( ) Notable Frequency Information CD’s = 44,100Hz Sampling Frequency Voice recording = 22,050Hz Sampling Frequency Human Speech < 4,000Hz FM (Frequency Modulation) High carrier frequencies Matlab Tutorial with command lines for you to enter in as you proceed Variables Plots FFT Wave File Tuning Fork Matlab (Matrix Laboratory) Import a wave file Plot in time domain (seconds as x axis) Carry out Fast Fourier Transform Algorithm Plot in frequency domain (Hertz) Analysis Your voice pronouncing vowels, consonants, words Fundamental frequency is first spike in frequency domain All other relevant frequency information is a multiple of that frequency Prove this in Matlab Voice Recording Record your own voice onto computer via microphone Just as completed in the Tutorial, import the file Plot in time domain Take FFT Plot in frequency domain Possible Applications – Projects 1) voice print identification – can you tell who someone is based upon their voice? 2) intelligent hearing aid – filter out noise, focus on voice output so as to improve the quality 3) filters – 4) MP3 vs WAV files