Implementation DFT and overlap Method(overlapadd & overlap-save)

• This project just deals with DFT and Block Convoltuion using Two Overlap Methods in DSP
• Implement using C++11
• FFT uses the kiss_fft library
• Use Low-pass Filter
• it was tested on the Window 10, Visual Studio 15

Project Guide

How to use

---

Simple Test

1. Download Project

git clone https://github.com/fbdp1202/DSP_Overlap-Save_Overlap_Add.git

2. Open Visual Studio Project "DSP_Overlap-Save_Overlap_Add.sln"

3. Just Build and Run

---

Result

1. Read and Write wav file

2. WAV spectrogram

• Spectrogram of Original WAV

• Spectrogram of Filtered WAV

---

Optional mode

// main.cpp
std::string fileName = "sa1.wav";

const int N = 1024;
const int M = 512;

Block_conv* tests[4];
std::string strOverlapAdd = "overlap_add";
std::string strOverlapSave = "overlap_save";
std::string strFFT = "fft";
std::string stdDFT = "dft";

tests[0] = new Block_conv(fileName, strOverlapAdd, strFFT, N, M);
tests[1] = new Block_conv(fileName, strOverlapAdd, stdDFT, N, M);
tests[2] = new Block_conv(fileName, strOverlapSave, strFFT, N, M);
tests[3] = new Block_conv(fileName, strOverlapSave, stdDFT, N, M);

• fileName : Choose Wav File (only Single Channel and 16bit Sample Length)

• N : number of DFT

• M : length of Low-Filter

• Func@Block_conv( std::string input wav file Name,

  ​ std::string method // Choose method "overlap_add" or "overlap_save"

  ​ std::string ) mode // Choose mode "fft" or "dft"

Overall System Block Diagram

1. Read wav file header

2. Store wav data in x[n] (block size)

3. DTFT of h(n) and x[n]

4. Multiply Y(z)=𝐻(𝑧)×𝑋(𝑧)

5. IDTFT of 𝑌(𝑧)

6. Do overlap-save or overlap-add

7. Repeat 2~6 until all wav data is read

8. Save wav

Implementation of DFT

Overlap Add

Input

---

Output

---

Overlap Save

Input

Output

---

LICENSE

BSD-3-Clause MIT Unlicense

Reference

Wav header format

How to read Wav file

kiss fft

overlap-save concept

overlap-add concept

DFT Matrix