//-----------------------------------------------------------------------------
// Test program.
// test.exe input.wav outout.wav f0 spec flag
// input.wav : argv[1] Input file
// output.wav : argv[2] Output file
// f0 : argv[3] F0 scaling (a positive number)
// spec : argv[4] Formant shift (a positive number)
//-----------------------------------------------------------------------------
int main(int argc, char *argv[]) {
if (argc != 2 && argc != 3 && argc != 4 && argc != 5) {
printf("error\n");
return -2;
}
// 2016/01/28: Important modification.
// Memory allocation is carried out in advanse.
// This is for compatibility with C language.
int x_length = GetAudioLength(argv[1]);
if (x_length <= 0) {
if (x_length == 0)
printf("error: File not found.\n");
else
printf("error: The file is not .wav format.\n");
return -1;
}
double *x = (double*) malloc(sizeof(double) * (x_length));
// wavread() must be called after GetAudioLength().
int fs, nbit;
wavread(argv[1], &fs, &nbit, x);
DisplayInformation(fs, nbit, x_length);
//---------------------------------------------------------------------------
// Analysis part
//---------------------------------------------------------------------------
// 2016/02/02
// A new struct is introduced to implement safe program.
WorldParameters world_parameters = { 0 };
// You must set fs and frame_period before analysis/synthesis.
world_parameters.fs = fs;
// 5.0 ms is the default value.
// Generally, the inverse of the lowest F0 of speech is the best.
// However, the more elapsed time is required.
world_parameters.frame_period = 5.0;
// F0 estimation
F0Estimation(x, x_length, &world_parameters);
// Spectral envelope estimation
SpectralEnvelopeEstimation(x, x_length, &world_parameters);
// Aperiodicity estimation by D4C
AperiodicityEstimation(x, x_length, &world_parameters);
// Note that F0 must not be changed until all parameters are estimated.
ParameterModification(argc, argv, fs, world_parameters.f0_length,
world_parameters.fft_size, world_parameters.f0,
world_parameters.spectrogram);
//---------------------------------------------------------------------------
// Synthesis part
//---------------------------------------------------------------------------
// The length of the output waveform
int y_length = (int)((world_parameters.f0_length - 1) *
world_parameters.frame_period / 1000.0 * fs) + 1;
double *y = (double*) malloc(sizeof(double) * (y_length));
// Synthesis
WaveformSynthesis(&world_parameters, fs, y_length, y);
// Output
wavwrite(y, y_length, fs, 16, argv[2]);
if(y != NULL) {
free(y);
y = NULL;
}
if(x != NULL) {
free(x);
x = NULL;
}
DestroyMemory(&world_parameters);
printf("complete.\n");
return 0;
}
//-----------------------------------------------------------------------------
// Test program.
// test.exe input.wav outout.wav f0 spec flag
// input.wav : argv[1] Input file
// output.wav : argv[2] Output file
// f0 : argv[3] F0 scaling (a positive number)
// spec : argv[4] Formant shift (a positive number)
//-----------------------------------------------------------------------------
int main(int argc, char *argv[]) {
if (argc != 2 && argc != 3 && argc != 4 && argc != 5) {
printf("error\n");
return -2;
}
// 2016/01/28: Important modification.
// Memory allocation is carried out in advanse.
// This is for compatibility with C language.
int x_length = GetAudioLength(argv[1]);
if (x_length <= 0) {
if (x_length == 0)
printf("error: File not found.\n");
else
printf("error: The file is not .wav format.\n");
return -1;
}
double *x = new double[x_length];
// wavread() must be called after GetAudioLength().
int fs, nbit;
wavread(argv[1], &fs, &nbit, x);
DisplayInformation(fs, nbit, x_length);
//---------------------------------------------------------------------------
// Analysis part
//---------------------------------------------------------------------------
// 2016/02/02
// A new struct is introduced to implement safe program.
WorldParameters world_parameters = { 0 };
// You must set fs and frame_period before analysis/synthesis.
world_parameters.fs = fs;
// 5.0 ms is the default value.
// Generally, the inverse of the lowest F0 of speech is the best.
// However, the more elapsed time is required.
world_parameters.frame_period = 5.0;
// F0 estimation
// DIO
// F0EstimationDio(x, x_length, &world_parameters);
// Harvest
F0EstimationHarvest(x, x_length, &world_parameters);
// Spectral envelope estimation
SpectralEnvelopeEstimation(x, x_length, &world_parameters);
// Aperiodicity estimation by D4C
AperiodicityEstimation(x, x_length, &world_parameters);
// Note that F0 must not be changed until all parameters are estimated.
ParameterModification(argc, argv, fs, world_parameters.f0_length,
world_parameters.fft_size, world_parameters.f0,
world_parameters.spectrogram);
//---------------------------------------------------------------------------
// Synthesis part (2016/04/19)
// There are three samples in speech synthesis
// 1: Conventional synthesis
// 2: Example of real-time synthesis
// 3: Example of real-time synthesis (Ring buffer is efficiently used)
//---------------------------------------------------------------------------
char filename[100];
// The length of the output waveform
int y_length = static_cast<int>((world_parameters.f0_length - 1) *
world_parameters.frame_period / 1000.0 * fs) + 1;
double *y = new double[y_length];
// Synthesis 1 (conventional synthesis)
for (int i = 0; i < y_length; ++i) y[i] = 0.0;
WaveformSynthesis(&world_parameters, fs, y_length, y);
sprintf(filename, "01%s", argv[2]);
wavwrite(y, y_length, fs, 16, filename);
// Synthesis 2 (All frames are added at the same time)
for (int i = 0; i < y_length; ++i) y[i] = 0.0;
WaveformSynthesis2(&world_parameters, fs, y_length, y);
sprintf(filename, "02%s", argv[2]);
wavwrite(y, y_length, fs, 16, filename);
// Synthesis 3 (Ring buffer is efficiently used.)
for (int i = 0; i < y_length; ++i) y[i] = 0.0;
WaveformSynthesis3(&world_parameters, fs, y_length, y);
sprintf(filename, "03%s", argv[2]);
wavwrite(y, y_length, fs, 16, filename);
delete[] y;
delete[] x;
DestroyMemory(&world_parameters);
printf("complete.\n");
return 0;
}
/**
* Main function
*
*/
int main(int argc, char *argv[])
{
if (argc != 5)
{
std::cerr << argv[0] << "<input_wav_file> <output_f0_file> <output_spectrum_file> <output_aperiodicity_file>" << std::endl;
return EXIT_FAILURE;
}
// 2016/01/28: Important modification.
// Memory allocation is carried out in advanse.
// This is for compatibility with C language.
int x_length = GetAudioLength(argv[1]);
if (x_length <= 0) {
if (x_length == 0)
std::cerr << "error: File \"" << argv[1] << "\" not found" << std::endl;
else
std::cerr << "error: File \"" << argv[1] << "\" is not a .wav format" << std::endl;
return EXIT_FAILURE;
}
double *x = new double[x_length];
// wavread() must be called after GetAudioLength().
int fs, nbit;
wavread(argv[1], &fs, &nbit, x);
DisplayInformation(fs, nbit, x_length);
// 2016/02/02
// A new struct is introduced to implement safe program.
WorldParameters world_parameters;
// You must set fs and frame_period before analysis/synthesis.
world_parameters.fs = fs;
// 5.0 ms is the default value.
// Generally, the inverse of the lowest F0 of speech is the best.
// However, the more elapsed time is required.
world_parameters.frame_period = 5.0;
//---------------------------------------------------------------------------
// Analysis part
//---------------------------------------------------------------------------
// F0 estimation
F0Estimation(x, x_length, &world_parameters);
// Spectral envelope estimation
SpectralEnvelopeEstimation(x, x_length, &world_parameters);
// Aperiodicity estimation by D4C
AperiodicityEstimation(x, x_length, &world_parameters);
std::cout << "fft size = " << world_parameters.fft_size << std::endl;
//---------------------------------------------------------------------------
// Saving part
//---------------------------------------------------------------------------
// F0 saving
std::ofstream out_f0(argv[2], std::ios::out | std::ios::binary);
if(!out_f0)
{
std::cerr << "Cannot open file: " << argv[2] << std::endl;
return EXIT_FAILURE;
}
out_f0.write(reinterpret_cast<const char*>(world_parameters.f0),
std::streamsize(world_parameters.f0_length * sizeof(double)));
out_f0.close();
// Spectrogram saving
std::ofstream out_spectrogram(argv[3], std::ios::out | std::ios::binary);
if(!out_spectrogram)
{
std::cerr << "Cannot open file: " << argv[3] << std::endl;
return EXIT_FAILURE;
}
// write the sampling frequency
out_spectrogram.write(reinterpret_cast<const char*>(&world_parameters.fs),
std::streamsize( sizeof(world_parameters.fs) ) );
// write the sampling frequency
out_spectrogram.write(reinterpret_cast<const char*>(&world_parameters.frame_period),
std::streamsize( sizeof(world_parameters.frame_period) ) );
// write the spectrogram data
for (int i=0; i<world_parameters.f0_length; i++)
{
out_spectrogram.write(reinterpret_cast<const char*>(world_parameters.spectrogram[i]),
std::streamsize((world_parameters.fft_size / 2 + 1) * sizeof(double)));
}
out_spectrogram.close();
// Aperiodicity saving
std::ofstream out_aperiodicity(argv[4], std::ios::out | std::ios::binary);
if(!out_aperiodicity)
{
std::cerr << "Cannot open file: " << argv[4] << std::endl;
return EXIT_FAILURE;
}
for (int i=0; i<world_parameters.f0_length; i++)
{
out_aperiodicity.write(reinterpret_cast<const char*>(world_parameters.aperiodicity[i]),
std::streamsize((world_parameters.fft_size / 2 + 1) * sizeof(double)));
}
out_aperiodicity.close();
//---------------------------------------------------------------------------
// Cleaning part
//---------------------------------------------------------------------------
delete[] x;
DestroyMemory(&world_parameters);
std::cout << "complete" << std::endl;
return EXIT_SUCCESS;
}
