A FFTW3 Audio library for Visualization use
Rather barebones at the moment, not clean at all. Needs many improvements, see Goals further below.
#include "listener.h"
#include <unistd.h>
#include <stdio.h>
int main(void){
// first, create listener
int buf_size = 1;
int n = 8192; // more complex numbers, essentially more processing for audio, and more memory.
int sample_rate = 44100;
double decay_rate = 0.992; // choose what you prefer, more => slower but smoother drop (in returned level), less => faster but more rigid drop (in returned level), 0 for no effect.
int frequencies[] = {50, 500}; // frequency range to listen in between
double sensitivity = 1.0; // bigger => (more) artificially multiplied returned level to essentially boost the output in any case. can be 0.
int num_ranges = sizeof(frequencies) / sizeof(frequencies[0]); // size of ranges in frequencies
Listener *bassListener = listener_create(buf_size, n, sample_rate, decay_rate, sensitivity, num_ranges, frequencies, "Sink #57");
printf("Listener initialized: buf_size=%d, sample_rate=%d, n=%d, decay_rate=%f, sensitivity=%f\n",
bassListener->buf_size, bassListener->sample_rate, bassListener->n, bassListener->decay_rate, bassListener->sensitivity);
if (!bassListener)
return 1;
if (start_listening(bassListener) != 0)
return 1;
printf("Ready!");
for (int i = 0; i < 10000; i += 100)
{
FreqInfo bass_info = get_freq_info(bassListener, 0);
printf("%f\n", bass_info.freq_level);
usleep(100000);
}
}and compile it like so
gcc -o example example.c listener.c -lm -lfftw3 -lpulseThe code as it is at the moment, is in a real bad state. It needs great refinement, optimization, and to be safer. I am hoping to turn this into a real library, and the current description is a little misleading. Current goals include:
[ ] Different compilation (currently you have to copy the file to the working directory or add it to the include PATH, and compile it like so .. -o example example.c listener.c ..).
[ ] More, better examples (example.c is different from the code provided above, and uses ncurses. The code in that file results in the program hanging upon receiving SIGINT/CTRL+C).
[ ] Lower latency if possible, though, the buffer is customizable as it is at the moment.
[ ] More options, more choices other than just PulseAudio.
[ ] Perhaps something else in the future.
This started initially as a concept I wanted to commit to making a real thing solely because I was bored. It took many hours and procrastination to get it working, especially threading, but I finally got it to a fine, working state and decided to release it to the public in hopes of learning of what can be further done to improve it. Any contribution is greatly appreciated.
As of the commit date involving this message here, I am still learning C and I am not good with optimization and safe code in general, so the code will look horrible.