audacia/lib-src/libscorealign/audioreader.cpp

100 lines
3.4 KiB
C++

/* audioreader.cpp -- reads sequence of overlapping windows
*
* 14-Jul-08 RBD
*/
#include <assert.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <math.h>
#include "sautils.h"
#include "string.h"
#include <fstream>
#include "allegro.h"
#include "audioreader.h"
#include "scorealign.h"
using namespace std;
long Audio_reader::read_window(float *data)
// reads the next window of samples
// the first time, fill half the window with zeros and the second half
// with data from the file
// after that, shift the window by hop_size and fill the end of the window
// with hop_size new samples
// the window is actually constructed in temp_data, then copied to data.
// That way, the caller can apply a smoothing function to data and we'll
// still have a copy.
// the function returns false on the next call when detecting that there
// is no more samples,
// data -- the window to be returned
// temp_data -- since we destroy data by windowing, temp_data saves
// overlapping samples so we don't have to read them again
// samples_per_frame -- must be even, note that first window is padded
// half-full with zeros
// hop_samples -- additional samples read each time after the first window
{
int frames_read; // how many frames did we read?
int hop = hop_samples;
if (reading_first_window) {
hop = samples_per_frame / 2; // first time we read more data
// zero end of temp_data, which will shift to beginning
memset(temp_data + hop, 0,
sizeof(float) * (samples_per_frame - hop));
reading_first_window = false;
}
// before reading in new sounds, shift temp_data by hop_size
memmove(temp_data, temp_data + hop,
(samples_per_frame - hop) * sizeof(float));
frames_read = read(temp_data + samples_per_frame - hop, hop);
// zero any leftovers (happens at last frame):
//printf("check fr %i hs %i ws %i ",frames_read,hop_size,window_size);
memset(temp_data + samples_per_frame - hop + frames_read, 0,
sizeof(float) * (hop - frames_read));
assert(samples_per_frame - frames_read >= 0);
// now copy temp_data to data
memcpy(data, temp_data, sizeof(float) * samples_per_frame);
if (frames_read != hop && reading_last_window == false) {
reading_last_window = true;
return true;
} else if (reading_last_window == true) {
return false;
} else {
return true;
}
}
void Audio_reader::calculate_parameters(Scorealign &sa, bool verbose)
{
double sample_rate = get_sample_rate();
long pcm_frames = get_frames();
// we want to make sure samples_per_frame is even, to keep things
// consistent we'll change hopsize_samples the same way
samples_per_frame = (int) (sa.window_size * sample_rate + 0.5);
if (samples_per_frame % 2 == 1)
samples_per_frame = samples_per_frame + 1;
/*=============================================================*/
hop_samples = (int)(sa.frame_period * sample_rate + 0.5);
if (hop_samples % 2 == 1)
hop_samples = hop_samples + 1;
actual_frame_period = (hop_samples / sample_rate);
// this is stored back in a field in sa as well as here in the reader
frame_count= (int) ceil(((float) pcm_frames / hop_samples + 1));
temp_data = ALLOC(float, samples_per_frame);
memset(temp_data, 0, samples_per_frame * sizeof(temp_data[0]));
assert(temp_data);
}