Fast Fourier Transform (FFT) of Time Series in Kafka Streams
report
1. A PROJECT REPORT ON
Beat Detection
By
Aditya Gudipudi (930111-2919)
Rohit Raghav (951228-P297)
Kunadha Raju R V Krishnam raju (931113-7971)
Avinash Goud Chekkilla (931218-2430)
3. Chapter 1
Abstract
In signal analysis, beat detection is detecting the beats in a musical tone using
computerized software. Feeling the beat of a song comes naturally to humans. Indeed
it is only a feeling one gets when listening to a melody. There can be several methods
in beat detection, but entirely dependent on speed hand accuracy. The project is a
demonstration of variation of leds in accordance to the beats in a music note with
the help of a digital signal processor.
We will study the variation of beats on a streaming source and calculating its beats
per minute.
The Project aims to know how the codes of these various algorithms work and be
able to adapt and create necessary changes in the code to meet the necessity.
4. CHAPTER 2
Software Used:
1. Matlab 2014
2. Visual Dsp++ 5.0
MATLAB
We first take an audio signal as an input and then divide the signal into equal block
sizes. Zeros are padded and the energy of the audio signal is computed, then we later
computed the FFT of the signal. We get the figure of the various energy signal. Then
the whole audio signal is converted to BPM (beats per minute), then we plot the
signal for the absolute FFT. After we run the program all the necessary plots are
obtained which show the drops in the beats and show its variation which is used to
detect the beat.
Matlab code and graphs obtained are shown below:
Code:
clc;
close all;
[x,fs] = audioread('C:Usersrohit kDesktoptoms.wav');
m=buffer(x,100);
[M,N]=size(m);
p=zeros(N,1);
penergy=zeros(N,1);
alpha=0.95;
for i=2:N
penergy(i)=sum(m(:,i).^2)/M;
p(i)=alpha*(p(i-1))+(1-alpha)*penergy(i);
end
figure(1),
x1=(abs(fft(penergy)));
s=0:2852-1;
7. Figure 4
Visual Dsp ++ 5.0
The Matlab code is then represented in C logic. It is then compiled using Visual Dsp
5 ++, an
Environment highly compatible for compiling the C logic into the Dsp processer.
The processor used in our project is ADSP 21262. It is a SHARC processor which
is used in speech, sound and various signal processing applications. C code used for
implementation of our project is shown below.
8. Chapter 3
Hardware Setup:
The various hardware components used in the project:
Sennheiser HD515 Head Phones
Computer System
ADSP 21262 Sharc Processor
Usb Data Cable
ADZS-HPUSB-ICE Emulator
Power Adaptor
Circuit Setup:
The Power Input is given to the Emulator and another power adaptor is used to power
the Sharc Processor. An Usb Data Cable is connected to the emulator which acts as
an interface between the Visual Dsp software and the processor hardware. A cable
from the HPUSB-Ice emulator is connected to the processor. An Audio input is given
to the processor and the output is taken from the processor device and given to the
speaker output to test the processor working condition. If you get the same output as
given by input simultaneously, the dsp is in working condition.
9. CHAPTER 4
CONCLUSION
We have observed the variation of led in accordance with the variation of beats to
the supplied input Music Audio signal found by calculating the fast Fourier
transform of the power spectral density of the given signal.