Through our analysis we have been exploring different MATLAB functions and DSP tools to help us analyze and extract data from the music files. Many of these functions we have used before, but applying them to our project reveals new uses and functionality that we had not previously explored.
MATLAB FUNCTIONS
- fast fourier transform
- Fast Fourier transform computes discrete Fourier transform(DFT) of a signal
- Helps us look at our audio input in the frequency domain
- instantaneous frequency
- Estimates instantaneous frequency
- Instantaneous Frequency of a nonstationary signal is a time-varying parameter that relates to the average of the frequencies present in the signal as it evolves
- we used the 'tfmoment' method which estimates the instantaneous frequency as the first conditional spectral moment of the time-frequency distribution of the input signal, essentially computing the spectrogram power spectrum and then integrates and averages
- spectrogram
- Computes a spectrogram using short-time Fourier transform
- Spectrogram is a visual representation of the spectrum of frequencies of a signal as it varies with time
- From the visual we can extrapolate what frequency the note is at and at what time samples it corresponds with
- bandpass filter
- Puts a bandpass filter on the signal, letting only a select band of frequencies to pass through
- Filters out noise and other unnecessary frequencies as we only need to consider the range that humans can hear at. Another interesting way we've been using it is to find the lowest frequencies that can accurately represent the beat of the music due to when and how frequently the frequency is repeated.
- novelty curve
- captures changes in music based off from when notes start and stop
- to produce we did horizontal differentiation and vertical accumulation on the spectrogram and filtered for noise past that
- helpful video describing more
- moving average filter
- computed average out of set range around each point in the graph
- useful to smooth out signal and decrease effect of outliers