Scribd
Upload
72 views

Itc Review 3 PDF

This document describes an implementation of audio compression using the Discrete Cosine Transform (DCT) technique. It begins with an introduction to DCT and how it can be used for audio compression. It then provides a block diagram and methodology for the algorithm. The methodology decomposes the audio signal into DCT basis vectors, quantizes coefficients, and reconstructs the signal using inverse DCT. It also implements Huffman encoding. Results show compression ratios around 2.0 with good quality metrics for tested audio files.

Uploaded by

Gvj Vamsi
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
72 views

Itc Review 3 PDF

This document describes an implementation of audio compression using the Discrete Cosine Transform (DCT) technique. It begins with an introduction to DCT and how it can be used for audio compression. It then provides a block diagram and methodology for the algorithm. The methodology decomposes the audio signal into DCT basis vectors, quantizes coefficients, and reconstructs the signal using inverse DCT. It also implements Huffman encoding. Results show compression ratios around 2.0 with good quality metrics for tested audio files.

Uploaded by

Gvj Vamsi
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

AUDIO COMPRESSION USING DCT

ECE4007 – Information Theory and Coding

Review-3

SLOT: B1+TB1

Team Members:

P.VAISHNAVI 17BEC0066
A.ABID ALI 17BEC0513

Submitted to:

Prof. K.S.Preetha

1|Page
Title of the project:
AUDIO COMPRESSION USING DCT

Abstract:

As technology is growing, there is an increasing need for data storage and


transmission. Compression makes the storage and transmission of data more
effective. There are many forms of audio compression techniques that offer a range of
complexity encoder and decoder, compressed audio quality, and varying amounts of
data compression such as DCT, MPEG, DWT etc. This project aims to implement an
audio signal compression algorithm using the Discrete Cosine Transform (DCT)
technique. Implemented algorithm performance is assessed on the basis of
compression ratio, Signal to Noise Ratio (SNR), and Root Mean Square Error (MSE).
For further, efficient storage and transmission we also implemented Huffman
Encoding.

Introduction:

DISCRETE COSINE TRANSFORM:

DCT is the real part of the Fourier transform. A discrete cosine transform (DCT)
describes a finite sequence of data points in terms of the sum of oscillating cosine
functions at different frequencies.

2|Page
Thanks to the high correlation in neighboring coefficients, discrete Cosine Transform
can be used for audio compression. We can reconstruct a series from very few DCT
coefficients, quite accurately. This DCT property assists in successful data reduction.
Its basic procedure is to take input audio data and convert it from one type of
representation to another, while in our case the signal is an audio sample block. In
this transformation, we transform set of spiral domain points to their equivalents in
the frequency domain.

DCT retains those coefficients which are just enough to represent the signal at a
particular desired quality. The DCT, as it has a strong energy compaction property in
typical applications, appears to concentrate most of the signal information in a few
low-frequency DCT components. It identifies pieces of information that can be
effectively thrown away without seriously reducing the audio's quality.

Tools to be used:

MATLAB software

3|Page
Block Diagram:

4|Page
Methodology:

ALGORITHM:

DCT:
1) select the audio and find the actual signal size.
2) Find amplitude and frequency
3) Set a particular sampling frequency
4) In the procedure, we will decompose the signal into DCT basis vectors. There
will be as many terms in the decomposition as there are samples in the signal.
The expansion coefficients will be assembled in a vector X which will actually
measure how much energy is stored in each of the components. Sort the
coefficients from largest to smallest.
5) Then we will determine the number of DCT coefficients that comprises 99.9%
of the energy in the signal.
6) we will set coefficients to zero that contain the remaining 0.1% of the energy.
7) 7.At last, using inverse discrete cosine transformation (IDCT) we can
reconstruct the signal from the compressed representation of signal
8) 8.Finally we will, analyze the compressed signal through Compression factor,
the quality of the signal by parameters such as Signal to Noise Ratio (PSNR),
and Root Mean Square Error (MSE)

Huffman coding:

1) follow above same instruction up to 6th step.


2) Then we quantize DCT coefficients with 256 levels between their Max and
Min Value.
3) Then we will find the probability distribution of these quantized
coefficients. 4.With the help of probability distribution we generate
“dictionary” for Huffman coding.
4) Then we will perform Huffman encoding, which has much compressed for
storage and transmission.
5) Then we will perform Huffman
encoding. finally apply IDCT to get our
file

5|Page
EXPLANATION:

1) First, we will select the file to extract to our MATLAB simulator


2) Then we sample our audio file using read function at a specific frequency
3) After that we will take the DCT of the generated samples.
4) Then DCT coefficients are rearranged in descending order.
5) We then run a loop and iterate until the sorted vector's magnitude is equal to the
original vector's magnitude of 99 percent. The vector magnitude gives us the
energy that is in the signal.

6) Then we will apply Inverse cosine transform to reconstruct the signal.


7) Finally, by using the audio write function, these samples are again sampled
back to the same sampling frequency thus creating an output compressed
audio.
8) For Huffman coding, we need to quantize the coefficient, then we have found
probability distribution, then dictionary, then encoding-decoding-IDCT to get
compressed file.

Work distribution among the batch members:

Member 1(17BEC066): Collection of various formats of Audio, performing DCT


decomposing the signal into DCT basis vectors. finding the number of DCT
coefficients that represent the signal are just enough to maintain quality of the
audio. reconstruct the signal from the compressed representation using inverse
discrete cosine transform (IDCT).

MEMBER 2(17BEC0513):
Finding the size of compressed signal and also analyze the quality of the signal by
parameters such as Signal to Noise Ratio (SNR), and Root Mean Square Error
(MSE). Performing Huffman encoding which involves quantization of coefficients,
finding probability distribution, dictionary generation for Huffman coding,
Encoding, Decoding, applying IDCT to get compressed file
6|Page
Results and Discussions:

Audiosample_1:

Audiosample_2:

7|Page
Sample PSNR MSE Compression ratio
Audiofile_1 40.9280 2.292 1.9477
Audiofile_2 54.8459 0.46188 2.0011

Thus, we have successfully reconstructed with compression ratio 1.9477, 2.0011 for
Audiofile_1 and Audiofile_2 respectively. We obtained PSN, MSE as 40.9280, 2.292
For Audiofile_1; 54.8459, 0.46188 for Audofile_2 thus providing us a compressed
audio file of good quality without much deviation from the original audio signal.
We have also Implemented Huffman encoding even though there would be
some quantization error, the quality of signal is still reasonable.

References:

1) Sankalp Shukla, Maniram Ahirwar, Ritu Gupta, Sarthak Jain, Dheeraj Singh
Rajput. Audio Compression Algorithm using Discrete Cosine Transform (DCT)
and Lempel-Ziv-Welch(LZW) Encoding Method, 2019 International
Conference on Machine Learning, Big Data, Cloud and Parallel Computing
(Com-IT-Con), India, 14th -16th Feb 2019

2) M. V. Patil, Apoorva Gupta, Ankita Varma, Shikhar Salil.Audio and Speech


Compression Using DCTand DWT Techniques, International Journal of
Innovative Research in Science, Engineering and Technology Vol. 2, Issue 5,
May 2013.

3) Sudha Rawat, Ajeet Kumar Verma Survey paper on image compression


techniques, ,2017

4) Mr. R. R. Karhe Ms. P. B. Shinde Ms. J. N. Fasale Audio Compressionusing


DCT & CS, , 2015

5) 5.A.M.Raid, W.M.Khedr , M. A. El-dosuky and Wesam Ahmed .Jpeg Image


Compression Using Discrete Cosin Transform, , 2014

8|Page

You might also like