For the given input audio file (.wav) the program will give the real time audio spectrum by performing Fast Fourier Transform (FFT) on the audio samples and plot the graph using OpenGL. This Project is implemented in C++ using OpenGL. Three frameworks Aquila-dsp, Kiss-Fft, and SFML are used to perform audio sampling, fast fourier transform and audio playback respectively. The development and testing of this project is done on Linux (ubuntu) using 4.2.0-16-generic Kernel. Interface for the program is given with the help of keyboard.
The following c++ frameworks are used in this project.
- Simple and Fast Multimedia Library (SFML)
- Aquila-dsp
- Kiss-fft
- OpenGL
The development of the project was done with these requirements. May support lower versions also but not been tested.
- C++ compiler : g++ with c++11 support (-std=gnu++11) (version >= 5.2.1)
- OpenGL (version >=v4.5)
- Aquila-dsp framework (version >=v3.0)
- Kiss-fft Framework (version >=v1.3.0)
- SFML framework >=2.3.2
- Git >=v2.5.0
Use `Left/Right` to move horizontally.
Use `Up/Down` to change the horizontal scale.
Use `Home` button to reset the position and scale.
Press ` F7 ` to toggle interpolation.
Press `F8` to toggle clamping.
Press `F9` to toggle drawing points.
Press `q` to exit the program.
Press `p` to to toggle audio play/pause.
Press `r` to reload audio and play from the beginning.
Press `right arrow` to seek audio forward by 5 seconds and move +ve x axis.
Press `left arrow` to seek audio backward by 5 seconds and move -ve x axis.
Press `down arrow` to decrease scale/1.5.
Press `up arrow` to increase scale*1.5.
Step 1: Begin
Int framePointer = 0, N = 32768.
Step 2: Load input audio file and play it using the audio Library
Step 3 : For i = framePointer to → framePointer + N < total_samples_count
Collect N samples from the audio file
Step 4: Apply suitable window function (e.g. Hann aka Hanning window)
Step 5: Apply Fast Fourier transform (FFT) on the array elements and collect N/2 Complex numbers having real part and imaginary part. (NB: if using typical complex-to-complex FFT then set imaginary parts of input array to zero)
Step 6: Calculate the magnitude of N/2FFT data.
magnitude = sqrt(re * re +img *img)
Step 7: Convert magnitude to dB (log)scale. (optional)
20 * log10(magnitude)
Step 8: Plot N/2 log(magnitude)values.
Step 9: If N >= total_samples_count
Exit.
Else goto Step 3.
Step 10: End.
Prerequisites
CMake: Aquila relies on CMake as a build tool, so make sure you have it installed (version 2.8 or later) Clone repository.
git clone git://github.com/zsiciarz/aquila.git aquila-src
Buid
Where to build:
a-top-level-directory/
aquila-build/
aquila-src/
In directory you have to build do
cmake ..\aquila-src
-DCMAKE_INSTALL_PREFIX=”~/home/mylib”
make
make install
###SFLM: Installing SFML is simple. You need administrative privilege on the machine
sudo apt-get install libsfml-dev
###Kiss FFT There is no need to build kiss-fft. You need to add their files while compiling
For easy compilation use the script ./compile from the project directory
g++ -std=c++11 -c draw.cpp
g++ -std=gnu++11 draw.o \
kiss_fft130/kiss_fft.c \
-L /home/<YourUserName>/mylib/lib/ \
-lAquila -lOoura_fft -lm \
-lglut -lGLEW -lGL -lGLU \
-lfreetype -lsfml-system \
-lsfml-audio ./common/shader_utils.o \
-o draw
To run the project:
./draw <audio-input>
Name : get_timestamp ()
Cspecification: static timestamp_t get_timestamp ()
Description: Used to get the current system time.
Name : getFft ()
Cspecification: void getFft ()
Description: Used to get the fourier transform of the input array elements.
Name : getData()
Cspecification: void getData()
Description: Used to fetch N samplings, convert them to frequency domain, and convert them to log10scale.
Name : init_resources()
Cspecification: int init_resources()
Description: Initializes the resources needed for the OpenGL to draw the graph.
Name : checkEnd()
Cspecification: int checkEnd()
Description: Returns 0 if all the samples are collected. Returns -1 otherwise.
Name : display()
Cspecification: void display()
Description: Loads the array of points into the buffer and draw the graph.
Name : special()
Cspecification: void special()
Description: Accepts a key as input and performs desired operation.
Name : free_resources()
Cspecification: void free_resources()
Description: Clears the resources allocated for the program.
This project is compiled and executed using g++ compiler g++ (version 5.2.1).
This Project is successfully completed and stable version 1.0 have been released.
Audio files of type (.wav) is supported.
Power Spectrum or Frequency Domain spectrum can be generated for the audio.
Basic Music player functions such as Load, Play, Pause, Seek and Replay is supported.
Clamping, interpolation, and showing of points can be toggled.
The spectrum can be scaled in Y axis.
Time synchronization of audio and spectrum is not perfect. Mp3 support is not included.
With proper implementation of window function (Hann Window) Accurate time synchronized spectrum of the audio can be generated. GUI can be provided for loading audio and other keyboard functions. Progress bar of the audio can be displayed. Spectrum can be implemented in 3D Program can be ported to Windows and other operating systems.
##Reference.
[1] http://music.columbia.edu/cmc/MusicAndComputers/
[2] https://stackoverflow.com/
[3] http://dsp.stackexchange.com/
[4] https://en.wikipedia.org/wiki/Frequency_domain
[5] https://en.wikipedia.org/wiki/Window_function