diff options
| -rw-r--r-- | ChangeLog | 14 | ||||
| -rw-r--r-- | configure.in | 2 | ||||
| -rw-r--r-- | examples/puredata/xtract~.c | 6 | ||||
| -rw-r--r-- | src/delta.c | 8 | ||||
| -rw-r--r-- | src/init.c | 2 | ||||
| -rw-r--r-- | src/libxtract.c | 8 | ||||
| -rw-r--r-- | src/scalar.c | 199 | ||||
| -rw-r--r-- | src/vector.c | 94 | ||||
| -rw-r--r-- | xtract/libxtract.h | 7 | ||||
| -rw-r--r-- | xtract/xtract_delta.h | 8 | ||||
| -rw-r--r-- | xtract/xtract_scalar.h | 84 | ||||
| -rw-r--r-- | xtract/xtract_vector.h | 20 |
12 files changed, 291 insertions, 161 deletions
@@ -1,3 +1,17 @@ +2006-12-15 Jamie Bullock <jamie@postlude.co.uk> + * version 0.3.4 + * Changed xtract_lowest_match(), to xtract_lowest(). It just returns the + lowest non-zero component in the input array, between two bounds + * qualified int *data, int N, and void *argv as const, thus insuring + that they won't be modified by the called function. + * added facilities for some functions to operate on copies of input + data instead of modifying it. This will need to be optimised at some + point maybe. + * Added arguments to spectral flatness to allow for control over input + band and pre-scaling + * Fixed confusing xtract_init_mfcc() parameter order + * Added new xtractor xtract_failsafe_f0 + 2006-12-12 Jamie Bullock <jamie@postlude.co.uk> * version 0.3.3 * Fixed errors in skewnes, kurtosis, irregularity_k, irregularity_j, diff --git a/configure.in b/configure.in index 616faad..d999d8d 100644 --- a/configure.in +++ b/configure.in @@ -4,7 +4,7 @@ m4_define(libxtract_major_version, 0) # Increment for feature additions and enhancements m4_define(libxtract_minor_version, 3) # Increment for fixes -m4_define(libxtract_fix_version, 2) +m4_define(libxtract_fix_version, 4) m4_define(libxtract_version, libxtract_major_version.libxtract_minor_version.libxtract_fix_version) diff --git a/examples/puredata/xtract~.c b/examples/puredata/xtract~.c index 77945f8..04d1079 100644 --- a/examples/puredata/xtract~.c +++ b/examples/puredata/xtract~.c @@ -141,7 +141,7 @@ static void *xtract_new(t_symbol *me, t_int argc, t_atom *argv) { else if(tmp == gensym("slope")) x->feature = SLOPE; else if(tmp == gensym("f0")) x->feature = F0; else if(tmp == gensym("hps"))x->feature = HPS; - else if(tmp == gensym("lowest_match"))x->feature = LOWEST_MATCH; + else if(tmp == gensym("lowest"))x->feature = LOWEST; else if(tmp == gensym("dct")) x->feature = DCT; else if(tmp == gensym("magnitude_spectrum")) x->feature = MAGNITUDE_SPECTRUM; @@ -168,7 +168,7 @@ static void *xtract_new(t_symbol *me, t_int argc, t_atom *argv) { case AVERAGE_DEVIATION: case ROLLOFF: case INHARMONICITY: - case LOWEST_MATCH: + case ODD_EVEN_RATIO: case F0: case TONALITY: floatargs = 1; @@ -176,6 +176,7 @@ static void *xtract_new(t_symbol *me, t_int argc, t_atom *argv) { case SKEWNESS: case KURTOSIS: case PEAKS: + case LOWEST: case HARMONICS: floatargs = 2; break; @@ -194,7 +195,6 @@ static void *xtract_new(t_symbol *me, t_int argc, t_atom *argv) { case NOISINESS: case RMS_AMPLITUDE: case POWER: - case ODD_EVEN_RATIO: case SHARPNESS: case SLOPE: case HPS: diff --git a/src/delta.c b/src/delta.c index b6259d6..8766927 100644 --- a/src/delta.c +++ b/src/delta.c @@ -22,25 +22,25 @@ #include "xtract/libxtract.h" -int xtract_flux(float *data, int N, void *argv , float *result){ +int xtract_flux(const float *data, const int N, const void *argv , float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_attack_time(float *data, int N, void *argv , float *result){ +int xtract_attack_time(const float *data, const int N, const void *argv , float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_decay_time(float *data, int N, void *argv, float *result){ +int xtract_decay_time(const float *data, const int N, const void *argv, float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_delta_feature(float *data, int N, void *argv, float *result){ +int xtract_delta_feature(const float *data, const int N, const void *argv, float *result){ return FEATURE_NOT_IMPLEMENTED; @@ -24,7 +24,7 @@ #include <math.h> #include <stdlib.h> -int xtract_init_mfcc(int N, float nyquist, int style, float freq_max, float freq_min, int freq_bands, float **fft_tables){ +int xtract_init_mfcc(int N, float nyquist, int style, float freq_min, float freq_max, int freq_bands, float **fft_tables){ int n, i, k, *fft_peak, M, next_peak; float norm, mel_freq_max, mel_freq_min, norm_fact, height, inc, val, diff --git a/src/libxtract.c b/src/libxtract.c index a749d96..11faeee 100644 --- a/src/libxtract.c +++ b/src/libxtract.c @@ -22,7 +22,7 @@ #include "xtract/libxtract.h" #define XTRACT -int(*xtract[])(float *, int, void *, float *) = { +int(*xtract[])(const float *, const int, const void *, float *) = { /* xtract_scalar.h */ xtract_mean, xtract_variance, @@ -51,9 +51,10 @@ int(*xtract[])(float *, int, void *, float *) = { xtract_odd_even_ratio, xtract_sharpness, xtract_slope, - xtract_lowest_match, + xtract_lowest, xtract_hps, xtract_f0, + xtract_failsafe_f0, /* xtract_delta.h */ xtract_flux, xtract_attack_time, @@ -100,9 +101,10 @@ char *xtract_help_strings[] = { "xtract_odd_even_ratio", "xtract_sharpness", "xtract_slope", - "xtract_lowest_match", + "xtract_lowest", "xtract_hps", "xtract_f0", + "xtract_failsafe_f0", "xtract_flux", "xtract_attack_time", "xtract_decay_time", diff --git a/src/scalar.c b/src/scalar.c index 476b843..55e7a00 100644 --- a/src/scalar.c +++ b/src/scalar.c @@ -24,8 +24,9 @@ #include "xtract/libxtract.h" #include "math.h" #include <stdlib.h> +#include <string.h> -int xtract_mean(float *data, int N, void *argv, float *result){ +int xtract_mean(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -37,26 +38,26 @@ int xtract_mean(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_variance(float *data, int N, void *argv, float *result){ +int xtract_variance(const float *data, const int N, const void *argv, float *result){ int n = N; while(n--) - *result += data[n] - *(float *)argv; + *result += pow(data[n] - *(float *)argv, 2); - *result = SQ(*result) / (N - 1); + *result = *result / (N - 1); return SUCCESS; } -int xtract_standard_deviation(float *data, int N, void *argv, float *result){ +int xtract_standard_deviation(const float *data, const int N, const void *argv, float *result){ *result = sqrt(*(float *)argv); return SUCCESS; } -int xtract_average_deviation(float *data, int N, void *argv, float *result){ +int xtract_average_deviation(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -68,7 +69,7 @@ int xtract_average_deviation(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_skewness(float *data, int N, void *argv, float *result){ +int xtract_skewness(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -84,7 +85,7 @@ int xtract_skewness(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_kurtosis(float *data, int N, void *argv, float *result){ +int xtract_kurtosis(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -102,11 +103,12 @@ int xtract_kurtosis(float *data, int N, void *argv, float *result){ } -int xtract_centroid(float *data, int N, void *argv, float *result){ +int xtract_centroid(const float *data, const int N, const void *argv, float *result){ int n = (N >> 1); - float *freqs, *amps, FA = 0.f, A = 0.f; + const float *freqs, *amps; + float FA = 0.f, A = 0.f; freqs = data; amps = data + n; @@ -121,7 +123,7 @@ int xtract_centroid(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_irregularity_k(float *data, int N, void *argv, float *result){ +int xtract_irregularity_k(const float *data, const int N, const void *argv, float *result){ int n, M = N - 1; @@ -132,7 +134,7 @@ int xtract_irregularity_k(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_irregularity_j(float *data, int N, void *argv, float *result){ +int xtract_irregularity_j(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -148,7 +150,7 @@ int xtract_irregularity_j(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_tristimulus_1(float *data, int N, void *argv, float *result){ +int xtract_tristimulus_1(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -169,7 +171,7 @@ int xtract_tristimulus_1(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_tristimulus_2(float *data, int N, void *argv, float *result){ +int xtract_tristimulus_2(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -194,7 +196,7 @@ int xtract_tristimulus_2(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_tristimulus_3(float *data, int N, void *argv, float *result){ +int xtract_tristimulus_3(const float *data, const int N, const void *argv, float *result){ int n = N, count = 0; @@ -216,23 +218,30 @@ int xtract_tristimulus_3(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_smoothness(float *data, int N, void *argv, float *result){ +int xtract_smoothness(const float *data, const int N, const void *argv, float *result){ int n = N; - if (data[0] <= 0) data[0] = 1; - if (data[1] <= 0) data[1] = 1; + float *input; + + input = (float *)malloc(N * sizeof(float)); + input = memcpy(input, data, N * sizeof(float)); + + if (input[0] <= 0) input[0] = 1; + if (input[1] <= 0) input[1] = 1; for(n = 2; n < N; n++){ - if(data[n] <= 0) data[n] = 1; - *result += abs(20 * log(data[n-1]) - (20 * log(data[n-2]) + - 20 * log(data[n-1]) + 20 * log(data[n])) / 3); + if(input[n] <= 0) input[n] = 1; + *result += abs(20 * log(input[n-1]) - (20 * log(input[n-2]) + + 20 * log(input[n-1]) + 20 * log(input[n])) / 3); } + + free(input); return SUCCESS; } -int xtract_spread(float *data, int N, void *argv, float *result){ +int xtract_spread(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -249,7 +258,7 @@ int xtract_spread(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_zcr(float *data, int N, void *argv, float *result){ +int xtract_zcr(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -261,7 +270,7 @@ int xtract_zcr(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_rolloff(float *data, int N, void *argv, float *result){ +int xtract_rolloff(const float *data, const int N, const void *argv, float *result){ int n = N; float pivot, temp; @@ -280,7 +289,7 @@ int xtract_rolloff(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_loudness(float *data, int N, void *argv, float *result){ +int xtract_loudness(const float *data, const int N, const void *argv, float *result){ int n = BARK_BANDS; @@ -293,16 +302,25 @@ int xtract_loudness(float *data, int N, void *argv, float *result){ } -int xtract_flatness(float *data, int N, void *argv, float *result){ +int xtract_flatness(const float *data, const int N, const void *argv, float *result){ int n; - float num, den, temp; + float num, den, temp, *tmp, prescale; + int lower, upper; + + tmp = (float *)argv; + lower = (int)tmp[0]; + upper = (int)tmp[1]; + prescale = (float)tmp[2]; + upper = (upper > N ? N : upper); + lower = (lower < 0.f ? 0.f : lower); + den = temp = num = 0.f; - for(n = 0; n < N; n++){ - if((temp = data[n])){ + for(n = lower; n < upper; n++){ + if((temp = data[n] * prescale)){ if(!num) num = den = temp; else{ @@ -320,7 +338,7 @@ int xtract_flatness(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_tonality(float *data, int N, void *argv, float *result){ +int xtract_tonality(const float *data, const int N, const void *argv, float *result){ float sfmdb, sfm; @@ -333,19 +351,19 @@ int xtract_tonality(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_crest(float *data, int N, void *argv, float *result){ +int xtract_crest(const float *data, const int N, const void *argv, float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_noisiness(float *data, int N, void *argv, float *result){ +int xtract_noisiness(const float *data, const int N, const void *argv, float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_rms_amplitude(float *data, int N, void *argv, float *result){ +int xtract_rms_amplitude(const float *data, const int N, const void *argv, float *result){ int n = N; @@ -356,11 +374,11 @@ int xtract_rms_amplitude(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_inharmonicity(float *data, int N, void *argv, float *result){ +int xtract_inharmonicity(const float *data, const int N, const void *argv, float *result){ int n = N >> 1; - float num = 0.f, den = 0.f, - fund, *freqs, *amps; + float num = 0.f, den = 0.f, fund; + const float *freqs, *amps; fund = *(float *)argv; freqs = data; @@ -377,23 +395,29 @@ int xtract_inharmonicity(float *data, int N, void *argv, float *result){ } -int xtract_power(float *data, int N, void *argv, float *result){ +int xtract_power(const float *data, const int N, const void *argv, float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_odd_even_ratio(float *data, int N, void *argv, float *result){ +int xtract_odd_even_ratio(const float *data, const int N, const void *argv, float *result){ - int n = N, j, k; + int M = (N >> 1), n; - float num = 0.f, den = 0.f; + float num = 0.f, den = 0.f, temp, f0; - while(n--){ - j = n * 2; - k = j - 1; - num += data[k]; - den += data[j]; + f0 = *(float *)argv; + + for(n = 0; n < M; n++){ + if((temp = data[n])){ + if(((int)(rintf(temp / f0)) % 2) != 0){ + num += data[M + n]; + } + else{ + den += data[M + n]; + } + } } *result = num / den; @@ -401,34 +425,39 @@ int xtract_odd_even_ratio(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_sharpness(float *data, int N, void *argv, float *result){ +int xtract_sharpness(const float *data, const int N, const void *argv, float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_slope(float *data, int N, void *argv, float *result){ +int xtract_slope(const float *data, const int N, const void *argv, float *result){ return FEATURE_NOT_IMPLEMENTED; } -int xtract_lowest_match(float *data, int N, void *argv, float *result){ +int xtract_lowest(const float *data, const int N, const void *argv, float *result){ - float lowest_match = SR_LIMIT; + float lower, upper, lowest; int n = N; + lower = *(float *)argv; + upper = *((float *)argv+1); + + lowest = upper; + while(n--) { - if(data[n] > 0) - lowest_match = MIN(lowest_match, data[n]); + if(data[n] > lower) + *result = MIN(lowest, data[n]); } - *result = (lowest_match == SR_LIMIT ? 0 : lowest_match); - + *result = (*result == upper ? -0 : *result); + return SUCCESS; } -int xtract_hps(float *data, int N, void *argv, float *result){ +int xtract_hps(const float *data, const int N, const void *argv, float *result){ int n = N, M, m, l, peak_index, position1_lwr; float *coeffs2, *coeffs3, *product, L, @@ -490,12 +519,18 @@ int xtract_hps(float *data, int N, void *argv, float *result){ } -int xtract_f0(float *data, int N, void *argv, float *result){ +int xtract_f0(const float *data, const int N, const void *argv, float *result){ int M, sr, tau, n; - float f0, err_tau_1, err_tau_x, array_max, threshold_peak, threshold_centre; + size_t bytes; + float f0, err_tau_1, err_tau_x, array_max, + threshold_peak, threshold_centre, + *input; sr = *(float *)argv; + + input = (float *)malloc(bytes = N * sizeof(float)); + input = memcpy(input, data, bytes); /* threshold_peak = *((float *)argv+1); threshold_centre = *((float *)argv+2); printf("peak: %.2f\tcentre: %.2f\n", threshold_peak, threshold_centre);*/ @@ -510,44 +545,74 @@ int xtract_f0(float *data, int N, void *argv, float *result){ /* Find the array max */ for(n = 0; n < N; n++){ - if (data[n] > array_max) - array_max = data[n]; + if (input[n] > array_max) + array_max = input[n]; } threshold_peak *= array_max; /* peak clip */ for(n = 0; n < N; n++){ - if(data[n] > threshold_peak) - data[n] = threshold_peak; - else if(data[n] < -threshold_peak) - data[n] = -threshold_peak; + if(input[n] > threshold_peak) + input[n] = threshold_peak; + else if(input[n] < -threshold_peak) + input[n] = -threshold_peak; } threshold_centre *= array_max; /* Centre clip */ for(n = 0; n < N; n++){ - if (data[n] < threshold_centre) - data[n] = 0; + if (input[n] < threshold_centre) + input[n] = 0; else - data[n] -= threshold_centre; + input[n] -= threshold_centre; } /* Estimate fundamental freq */ for (n = 1; n < M; n++) - err_tau_1 = err_tau_1 + fabs(data[n] - data[n+1]); + err_tau_1 = err_tau_1 + fabs(input[n] - input[n+1]); /* FIX: this doesn't pose too much load if it returns 'early', but if it can't find f0, load can be significant for larger block sizes M^2 iterations! */ for (tau = 2; tau < M; tau++){ err_tau_x = 0; for (n = 1; n < M; n++){ - err_tau_x = err_tau_x + fabs(data[n] - data[n+tau]); + err_tau_x = err_tau_x + fabs(input[n] - input[n+tau]); } if (err_tau_x < err_tau_1) { f0 = sr / (tau + (err_tau_x / err_tau_1)); *result = f0; + free(input); return SUCCESS; } } + *result = -0; + free(input); return NO_RESULT; } + +int xtract_failsafe_f0(const float *data, const int N, const void *argv, float *result){ + + float *magnitudes = NULL, argf[2], *peaks = NULL, return_code; + + return_code = xtract_f0(data, N, argv, result); + + if(return_code == NO_RESULT){ + + magnitudes = (float *)malloc(N * sizeof(float)); + peaks = (float *)malloc(N * sizeof(float)); + xtract_magnitude_spectrum(data, N, NULL, magnitudes); + argf[0] = 10.f; + argf[1] = *(float *)argv; + xtract_peaks(magnitudes, N, argf, peaks); + argf[0] = 0.f; + argf[1] = N >> 1; + xtract_lowest(peaks, argf[1], argf, result); + + free(magnitudes); + free(peaks); + } + + return SUCCESS; + +} + diff --git a/src/vector.c b/src/vector.c index 275b2b7..b0e42d5 100644 --- a/src/vector.c +++ b/src/vector.c @@ -23,20 +23,25 @@ #include "xtract/libxtract.h" #include <math.h> +#include <string.h> +#include <stdlib.h> #ifdef XTRACT_FFT #include <fftw3.h> -int xtract_magnitude_spectrum(float *data, int N, void *argv, float *result){ +int xtract_magnitude_spectrum(const float *data, const int N, const void *argv, float *result){ - float *temp; + float *temp, *input; + size_t bytes; int n , M = N >> 1; fftwf_plan plan; temp = (float *)fftwf_malloc(N * sizeof(float)); + input = (float *)malloc(bytes = N * sizeof(float)); + input = memcpy(input, data, bytes); - plan = fftwf_plan_r2r_1d(N, data, temp, FFTW_R2HC, FFTW_ESTIMATE); + plan = fftwf_plan_r2r_1d(N, input, temp, FFTW_R2HC, FFTW_ESTIMATE); fftwf_execute(plan); @@ -50,18 +55,23 @@ int xtract_magnitude_spectrum(float *data, int N, void *argv, float *result){ fftwf_destroy_plan(plan); fftwf_free(temp); + free(input); return SUCCESS; } -int xtract_autocorrelation_fft(float *data, int N, void *argv, float *result){ +int xtract_autocorrelation_fft(const float *data, const int N, const void *argv, float *result){ - float *temp; + float *temp, *input; + size_t bytes; int n; fftwf_plan plan; temp = (float *)fftwf_malloc(N * sizeof(float)); - plan = fftwf_plan_r2r_1d(N, data, temp, FFTW_HC2R, FFTW_ESTIMATE); + input = (float *)malloc(bytes = N * sizeof(float)); + input = memcpy(input, data, bytes); + + plan = fftwf_plan_r2r_1d(N, input, temp, FFTW_HC2R, FFTW_ESTIMATE); fftwf_execute(plan); @@ -70,20 +80,26 @@ int xtract_autocorrelation_fft(float *data, int N, void *argv, float *result){ fftwf_destroy_plan(plan); fftwf_free(temp); + free(input); return SUCCESS; } -int xtract_mfcc(float *data, int N, void *argv, float *result){ +int xtract_mfcc(const float *data, const int N, const void *argv, float *result){ xtract_mel_filter *f; + float *input; + size_t bytes; int n, filter; f = (xtract_mel_filter *)argv; + input = (float *)malloc(bytes = N * sizeof(float)); + input = memcpy(input, data, bytes); + for(filter = 0; filter < f->n_filters; filter++){ for(n = 0; n < N; n++){ - result[filter] += data[n] * f->filters[filter][n]; + result[filter] += input[n] * f->filters[filter][n]; } if(result[filter] < LOG_LIMIT) result[filter] = LOG_LIMIT; result[filter] = log(result[filter]); @@ -93,43 +109,51 @@ int xtract_mfcc(float *data, int N, void *argv, float *result){ xtract_dct(result, f->n_filters, NULL, result); + free(input); + return SUCCESS; } -int xtract_dct(float *data, int N, void *argv, float *result){ +int xtract_dct(const float *data, const int N, const void *argv, float *result){ fftwf_plan plan; + float *input; + size_t bytes; + input = (float *)malloc(bytes = N * sizeof(float)); + input = memcpy(input, data, bytes); + plan = - fftwf_plan_r2r_1d(N, data, result, FFTW_REDFT00, FFTW_ESTIMATE); + fftwf_plan_r2r_1d(N, input, result, FFTW_REDFT00, FFTW_ESTIMATE); fftwf_execute(plan); fftwf_destroy_plan(plan); + free(input); return SUCCESS; } #else -int xtract_magnitude_spectrum(float *data, int N, void *argv, float *result){ +int xtract_magnitude_spectrum(const float *data, const int N, const void *argv, float *result){ NEEDS_FFTW; } -int xtract_autocorrelation_fft(float *data, int N, void *argv, float *result){ +int xtract_autocorrelation_fft(const float *data, const int N, const void *argv, float *result){ NEEDS_FFTW; } -int xtract_mfcc(float *data, int N, void *argv, float *result){ +int xtract_mfcc(const float *data, const int N, const void *argv, float *result){ NEEDS_FFTW; } -int xtract_dct(float *data, int N, void *argv, float *result){ +int xtract_dct(const float *data, const int N, const void *argv, float *result){ NEEDS_FFTW; @@ -137,7 +161,7 @@ int xtract_dct(float *data, int N, void *argv, float *result){ #endif -int xtract_autocorrelation(float *data, int N, void *argv, float *result){ +int xtract_autocorrelation(const float *data, const int N, const void *argv, float *result){ /* Naive time domain implementation */ @@ -156,7 +180,7 @@ int xtract_autocorrelation(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_amdf(float *data, int N, void *argv, float *result){ +int xtract_amdf(const float *data, const int N, const void *argv, float *result){ int n = N, i; @@ -175,7 +199,7 @@ int xtract_amdf(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_asdf(float *data, int N, void *argv, float *result){ +int xtract_asdf(const float *data, const int N, const void *argv, float *result){ int n = N, i; @@ -193,7 +217,7 @@ int xtract_asdf(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_bark_coefficients(float *data, int N, void *argv, float *result){ +int xtract_bark_coefficients(const float *data, const int N, const void *argv, float *result){ int *limits, band, n; @@ -207,9 +231,12 @@ int xtract_bark_coefficients(float *data, int N, void *argv, float *result){ return SUCCESS; } -int xtract_peaks(float *data, int N, void *argv, float *result){ +int xtract_peaks(const float *data, const int N, const void *argv, float *result){ - float thresh, max, y, y2, y3, p, width, sr; + float thresh, max, y, y2, + y3, p, width, sr, + *input = NULL; + size_t bytes; int n = N, M, return_code = SUCCESS; if(argv != NULL){ @@ -222,6 +249,13 @@ int xtract_peaks(float *data, int N, void *argv, float *result){ sr = 44100; } + input = (float *)malloc(bytes = N * sizeof(float)); + + if(input != NULL) + input = memcpy(input, data, bytes); + else + return MALLOC_FAILED; + M = N >> 1; width = sr / N; @@ -238,12 +272,12 @@ int xtract_peaks(float *data, int N, void *argv, float *result){ } while(n--){ - max = MAX(max, data[n]); + max = MAX(max, input[n]); /* ensure we never take log10(0) */ - /*data[n] = (data[n] < LOG_LIMIT ? LOG_LIMIT : data[n]);*/ - if ((data[n] * 100000) <= 1) + /*input[n] = (input[n] < LOG_LIMIT ? LOG_LIMIT : input[n]);*/ + if ((input[n] * 100000) <= 1) /* We get a more stable peak this way */ - data[n] = 1; + input[n] = 1; } @@ -253,9 +287,9 @@ int xtract_peaks(float *data, int N, void *argv, float *result){ result[M] = 0; for(n = 1; n < M; n++){ - if(data[n] >= thresh){ - if(data[n] > data[n - 1] && data[n] > data[n + 1]){ - result[n] = width * (n + (p = .5 * (y = 20 * log10(data[n-1]) - (y3 = 20 * log10(data[n+1]))) / (20 * log10(data[n - 1]) - 2 * (y2 = 20 * log10(data[n])) + 20 * log10(data[n + 1])))); + if(input[n] >= thresh){ + if(input[n] > input[n - 1] && input[n] > input[n + 1]){ + result[n] = width * (n + (p = .5 * (y = 20 * log10(input[n-1]) - (y3 = 20 * log10(input[n+1]))) / (20 * log10(input[n - 1]) - 2 * (y2 = 20 * log10(input[n])) + 20 * log10(input[n + 1])))); result[M + n] = y2 - .25 * (y - y3) * p; } else{ @@ -269,14 +303,16 @@ int xtract_peaks(float *data, int N, void *argv, float *result){ } } + free(input); return (return_code ? return_code : SUCCESS); } -int xtract_harmonics(float *data, int N, void *argv, float *result){ +int xtract_harmonics(const float *data, const int N, const void *argv, float *result){ int n = (N >> 1), M = n; - float *freqs, *amps, f0, thresh, ratio, nearest, distance; + const float *freqs, *amps; + float f0, thresh, ratio, nearest, distance; freqs = data; amps = data + n; diff --git a/xtract/libxtract.h b/xtract/libxtract.h index 489cddf..7670c83 100644 --- a/xtract/libxtract.h +++ b/xtract/libxtract.h @@ -53,7 +53,7 @@ extern "C" { * @{ */ -#define XTRACT_FEATURES 44 +#define XTRACT_FEATURES 45 #define LOG_LIMIT 10e-10 #define VERY_BIG_NUMBER 2e10 @@ -89,9 +89,10 @@ enum features_ { ODD_EVEN_RATIO, SHARPNESS, SLOPE, - LOWEST_MATCH, + LOWEST, HPS, F0, + FAILSAFE_F0, FLUX, ATTACK_TIME, DECAY_TIME, @@ -178,7 +179,7 @@ printf("Mean = %.2f\n", mean); * */ #ifdef XTRACT -extern int(*xtract[XTRACT_FEATURES])(float *data, int N, void *argv, float *result); +extern int(*xtract[XTRACT_FEATURES])(const float *data, const int N, const void *argv, float *result); /** \brief An array of pointers to function help strings * diff --git a/xtract/xtract_delta.h b/xtract/xtract_delta.h index bf6fa15..461ced6 100644 --- a/xtract/xtract_delta.h +++ b/xtract/xtract_delta.h @@ -30,20 +30,20 @@ extern "C" { #include "xtract_types.h" /* \brief Extract spectral flux as defined by Gaël Richard (2006)*/ -int xtract_flux(float *data, int N, void *argv , float *result); +int xtract_flux(const float *data, const int N, const void *argv , float *result); /*xtract_frame_tracker *xf */ /** \brief Extract attack Time */ -int xtract_attack_time(float *data, int N, void *argv , float *result); +int xtract_attack_time(const float *data, const int N, const void *argv , float *result); /* xtract_amp_tracker *xa */ /** Extract temporal decrease */ -int xtract_decay_time(float *data, int N, void *argv, float *result); +int xtract_decay_time(const float *data, const int N, const void *argv, float *result); /* xtract_amp_tracker *xa */ /** \brief A generic function to calculate the delta of a feature over a given period (in frames) */ -int xtract_delta_feature(float *data, int N, void *argv, float *result); +int xtract_delta_feature(const float *data, const int N, const void *argv, float *result); /*xtract_frame_tracker *xf */ /*float frames*/ diff --git a/xtract/xtract_scalar.h b/xtract/xtract_scalar.h index 8beceec..c4618fa 100644 --- a/xtract/xtract_scalar.h +++ b/xtract/xtract_scalar.h @@ -41,7 +41,7 @@ extern "C" { * \param *argv: a pointer to NULL * \param *result: the mean of N values from the array pointed to by *data */ -int xtract_mean(float *data, int N, void *argv, float *result); +int xtract_mean(const float *data, const int N, const void *argv, float *result); /** \brief Extract the variance of an input vector * @@ -50,7 +50,7 @@ int xtract_mean(float *data, int N, void *argv, float *result); * \param *argv: a pointer to a float representing the mean of the input vector * \param *result: the variance of N values from the array pointed to by *data */ -int xtract_variance(float *data, int N, void *argv, float *result); +int xtract_variance(const float *data, const int N, const void *argv, float *result); /** \brief Extract the deviation of an input vector * @@ -59,7 +59,7 @@ int xtract_variance(float *data, int N, void *argv, float *result); * \param *argv: a pointer to a float representing the variance of the input vector * \param *result: the deviation of N values from the array pointed to by *data */ -int xtract_standard_deviation(float *data, int N, void *argv, float *result); +int xtract_standard_deviation(const float *data, const int N, const void *argv, float *result); /** \brief Extract the average deviation of an input vector * @@ -68,7 +68,7 @@ int xtract_standard_deviation(float *data, int N, void *argv, float *result); * \param *argv: a pointer to a float representing the mean of the input vector * \param *result: the average deviation of N values from the array pointed to by *data */ -int xtract_average_deviation(float *data, int N, void *argv, float *result); +int xtract_average_deviation(const float *data, const int N, const void *argv, float *result); /** \brief Extract the skewness of an input vector * @@ -77,7 +77,7 @@ int xtract_average_deviation(float *data, int N, void *argv, float *result); * \param *argv: a pointer to an array of floats representing the mean and standard deviation of the input vector * \param *result: the skewness of N values from the array pointed to by *data */ -int xtract_skewness(float *data, int N, void *argv, float *result); +int xtract_skewness(const float *data, const int N, const void *argv, float *result); /** \brief Extract the kurtosis of an input vector * @@ -86,7 +86,7 @@ int xtract_skewness(float *data, int N, void *argv, float *result); * \param *argv: a pointer to an array of values representing the mean and standard deviation of the input vector * \param *result: the kurtosis of N values from the array pointed to by *data */ -int xtract_kurtosis(float *data, int N, void *argv, float *result); +int xtract_kurtosis(const float *data, const int N, const void *argv, float *result); /** \brief Extract the centroid of an input vector * @@ -95,7 +95,7 @@ int xtract_kurtosis(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the centroid of the values pointed to by *data */ -int xtract_centroid(float *data, int N, void *argv, float *result); +int xtract_centroid(const float *data, const int N, const void *argv, float *result); /** \brief Calculate the Irregularity of an input vector using a method described by Krimphoff (1994) * @@ -104,7 +104,7 @@ int xtract_centroid(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the irregularity of N values from the array pointed to by *data */ -int xtract_irregularity_k(float *data, int N, void *argv, float *result); +int xtract_irregularity_k(const float *data, const int N, const void *argv, float *result); /** \brief Calculate the Irregularity of an input vector using a method described by Jensen (1999) * @@ -113,7 +113,7 @@ int xtract_irregularity_k(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the irregularity of N values from the array pointed to by *data */ -int xtract_irregularity_j(float *data, int N, void *argv, float *result); +int xtract_irregularity_j(const float *data, const int N, const void *argv, float *result); /** \brief Calculate the Tristimulus of an input vector using a method described by Pollard and Jansson (1982) * @@ -125,18 +125,18 @@ int xtract_irregularity_j(float *data, int N, void *argv, float *result); * These three functions provide the first, second and third order tristimulus formulae * */ -int xtract_tristimulus_1(float *data, int N, void *argv, float *result); -int xtract_tristimulus_2(float *data, int N, void *argv, float *result); -int xtract_tristimulus_3(float *data, int N, void *argv, float *result); +int xtract_tristimulus_1(const float *data, const int N, const void *argv, float *result); +int xtract_tristimulus_2(const float *data, const int N, const void *argv, float *result); +int xtract_tristimulus_3(const float *data, const int N, const void *argv, float *result); /** \brief Extract the smoothness of an input vector using a method described by McAdams (1999) * * \param *data: a pointer to the first element in an array of floats representing the magnitude spectrum of an audio vector * \param N: the number of elements to be considered - * \param *argv: a pointer to NULL + * \param *argv: a pointer to the first element of an array of integers containing the lower bound, upper bound, and pre-scaling factor, whereby array data in the range lower < n < upper will be pre-scaled by p before processing. * \param *result: the smoothness of N values from the array pointed to by *data */ -int xtract_smoothness(float *data, int N, void *argv, float *result); +int xtract_smoothness(const float *data, const int N, const void *argv, float *result); /** \brief Extract the spectral spread of an input vector using a method described by Casagrande(2005) * @@ -145,7 +145,7 @@ int xtract_smoothness(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the spectral spread of N values from the array pointed to by *data */ -int xtract_spread(float *data, int N, void *argv, float *result); +int xtract_spread(const float *data, const int N, const void *argv, float *result); /* Zero crossing rate */ @@ -156,7 +156,7 @@ int xtract_spread(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the zero crossing rate of N values from the array pointed to by *data */ -int xtract_zcr(float *data, int N, void *argv, float *result); +int xtract_zcr(const float *data, const int N, const void *argv, float *result); /** \brief Extract the spectral rolloff of an input vector using a method described by Bee Suan Ong (2005) * @@ -165,7 +165,7 @@ int xtract_zcr(float *data, int N, void *argv, float *result); * \param *argv: a pointer to an array containing a floating point value representing the threshold for rolloff, i.e. the percentile at which the rolloff is determined, expressed in the range 0-1.0, and a float representing the sample rate in Hz * \param *result: the spectral rolloff in Hz of N values from the array pointed to by *data. This is the point in the spectrum below which argv[0] of the energy is distributed. */ -int xtract_rolloff(float *data, int N, void *argv, float *result); +int xtract_rolloff(const float *data, const int N, const void *argv, float *result); /* Loudness */ /* A set of BARK_BANDS bark coefficients must be passed in, the loudness is calculated approximately according to Moore, Glasberg et al, 1997 */ @@ -177,7 +177,7 @@ int xtract_rolloff(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the loudness of N values from the array pointed to by *data */ -int xtract_loudness(float *data, int N, void *argv, float *result); +int xtract_loudness(const float *data, const int N, const void *argv, float *result); /** \brief Extract the spectral flatness measure of an input vector using a method described by Tristan Jehan (2005) * @@ -186,7 +186,7 @@ int xtract_loudness(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the spectral flatness of N values from the array pointed to by *data */ -int xtract_flatness(float *data, int N, void *argv, float *result); +int xtract_flatness(const float *data, const int N, const void *argv, float *result); /** \brief Extract the tonality factor of an input vector using a method described by Tristan Jehan (2005) @@ -196,7 +196,7 @@ int xtract_flatness(float *data, int N, void *argv, float *result); * \param *argv: a pointer to the spectral flatness measure of an audio vector (e.g. the output from xtract_flatness) * \param *result: the tonality factor of N values from the array pointed to by *data */ -int xtract_tonality(float *data, int N, void *argv, float *result); +int xtract_tonality(const float *data, const int N, const void *argv, float *result); /** \brief Extract the noisiness of an input vector using a method described by Tae Hong Park (2000) * @@ -205,7 +205,7 @@ int xtract_tonality(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the noisiness of N values from the array pointed to by *data */ -int xtract_noisiness(float *data, int N, void *argv, float *result); +int xtract_noisiness(const float *data, const int N, const void *argv, float *result); /** \brief Extract the RMS amplitude of an input vector using a method described by Tae Hong Park (2000) * @@ -214,7 +214,7 @@ int xtract_noisiness(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the RMS amplitude of N values from the array pointed to by *data */ -int xtract_rms_amplitude(float *data, int N, void *argv, float *result); +int xtract_rms_amplitude(const float *data, const int N, const void *argv, float *result); /** \brief Extract the Inharmonicity of an input vector * @@ -223,7 +223,7 @@ int xtract_rms_amplitude(float *data, int N, void *argv, float *result); * \param *argv: a pointer to a float representing the fundamental frequency of the input vector. * \param *result: the inharmonicity of N values from the array pointed to by *data */ -int xtract_inharmonicity(float *data, int N, void *argv, float *result); +int xtract_inharmonicity(const float *data, const int N, const void *argv, float *result); /** \brief Extract the spectral crest of an input vector using a method described by Peeters (2003) * @@ -232,7 +232,7 @@ int xtract_inharmonicity(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the spectral crest of N values from the array pointed to by *data */ -int xtract_crest(float *data, int N, void *argv, float *result); +int xtract_crest(const float *data, const int N, const void *argv, float *result); /** \brief Extract the Spectral Power of an input vector using a method described by Bee Suan Ong (2005) * @@ -241,7 +241,7 @@ int xtract_crest(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the spectral power of N values from the array pointed to by *data */ -int xtract_power(float *data, int N, void *argv, float *result); +int xtract_power(const float *data, const int N, const void *argv, float *result); /* Odd to even harmonic ratio */ /** \brief Extract the Odd to even harmonic ratio of an input vector @@ -251,7 +251,7 @@ int xtract_power(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the odd/even harmonic ratio of N values from the array pointed to by *data */ -int xtract_odd_even_ratio(float *data, int N, void *argv, float *result); +int xtract_odd_even_ratio(const float *data, const int N, const void *argv, float *result); /** \brief Extract the Sharpness of an input vector * @@ -260,7 +260,7 @@ int xtract_odd_even_ratio(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the Sharpness of N values from the array pointed to by *data */ -int xtract_sharpness(float *data, int N, void *argv, float *result); +int xtract_sharpness(const float *data, const int N, const void *argv, float *result); /** \brief Extract the Slope of an input vector * @@ -269,17 +269,17 @@ int xtract_sharpness(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the Slope of N values from the array pointed to by *data */ -int xtract_slope(float *data, int N, void *argv, float *result); +int xtract_slope(const float *data, const int N, const void *argv, float *result); -/** \brief Extract the value of the first partial in an input vector that closely matches a certain 'guess' +/** \brief Extract the value of the lowest value in an input vector that between two bounds * - * \param *data: a pointer to the first element in an array of floats that represents the frequencies of the spectral peaks of an audio vector + * \param *data: a pointer to the first element in an array of floats * \param N: the number of elements to be considered - * \param *argv: a pointer to a float value representing the guess - * \param *result: the F0 of N values from the array pointed to by *data + * \param *argv: a pointer to an array containing a lower and upper bounds for search, where lower < n < upper. + * \param *result: a pointer to a value representing the lowest non-zero component in *data. If no match is found then -0 is returned. * - * This method takes a guess which can come from taking the ZCR of an autocorrelation function, and then finds the spectral peak that most closely matches the gess */ -int xtract_lowest_match(float *data, int N, void *argv, float *result); + */ +int xtract_lowest(const float *data, const int N, const void *argv, float *result); /** \brief Extract the Pitch of an input vector using Harmonic Product Spectrum (HPS) analysis * @@ -290,7 +290,7 @@ int xtract_lowest_match(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the pitch of N values from the array pointed to by *data */ -int xtract_hps(float *data, int N, void *argv, float *result); +int xtract_hps(const float *data, const int N, const void *argv, float *result); /** \brief Extract the fundamental frequency of an input vector * @@ -302,7 +302,19 @@ int xtract_hps(float *data, int N, void *argv, float *result); * This algorithm is based on the AMDF, with peak and centre clipping. It would benefit from further improvements to improve noise robustness and overall efficiency * */ -int xtract_f0(float *data, int N, void *argv, float *result); +int xtract_f0(const float *data, const int N, const void *argv, float *result); + +/** \brief Extract the fundamental frequency of an input vector + * + * \param *data: a pointer to the first element in an array of floats representing an audio vector + * \param N: the number of elements to be considered + * \param *argv: a pointer to a float representing the audio sample rate + * \param *result: the pitch of N values from the array pointed to by *data + * + * This function wraps xtract_f0, but provides the frequency of the lowest partial in the peak spectrum if f0 can't be found. + * + */ +int xtract_failsafe_f0(const float *data, const int N, const void *argv, float *result); /** @} */ diff --git a/xtract/xtract_vector.h b/xtract/xtract_vector.h index a8b0550..215c782 100644 --- a/xtract/xtract_vector.h +++ b/xtract/xtract_vector.h @@ -41,7 +41,7 @@ extern "C" { * \param *argv: a pointer to NULL * \param *result: the magnitude spectrum of N values from the array pointed to by *data */ -int xtract_magnitude_spectrum(float *data, int N, void *argv, float *result); +int xtract_magnitude_spectrum(const float *data, const int N, const void *argv, float *result); /** \brief Extract autocorrelation from time domain signal using FFT based method * @@ -50,7 +50,7 @@ int xtract_magnitude_spectrum(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the autocorrelation of N values from the array pointed to by *data */ -int xtract_autocorrelation_fft(float *data, int N, void *argv, float *result); +int xtract_autocorrelation_fft(const float *data, const int N, const void *argv, float *result); /** \brief Extract Mel Frequency Cepstral Coefficients based on a method described by Rabiner * @@ -61,7 +61,7 @@ int xtract_autocorrelation_fft(float *data, int N, void *argv, float *result); * * The data structure pointed to by *argv must be obtained by first calling xtract_init_mfcc */ -int xtract_mfcc(float *data, int N, void *argv, float *result); +int xtract_mfcc(const float *data, const int N, const void *argv, float *result); /** \brief Extract the Discrete Cosine transform of a time domain signal * \param *data: a pointer to the first element in an array of floats representing an audio vector @@ -69,7 +69,7 @@ int xtract_mfcc(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: a pointer to an array containing resultant dct coefficients */ -int xtract_dct(float *data, int N, void *argv, float *result); +int xtract_dct(const float *data, const int N, const void *argv, float *result); /** \brief Extract autocorrelation from time domain signal using time-domain autocorrelation technique * @@ -78,7 +78,7 @@ int xtract_dct(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the autocorrelation of N values from the array pointed to by *data */ -int xtract_autocorrelation(float *data, int N, void *argv, float *result); +int xtract_autocorrelation(const float *data, const int N, const void *argv, float *result); /** \brief Extract Average Magnitude Difference Function from time domain signal * @@ -87,7 +87,7 @@ int xtract_autocorrelation(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the AMDF of N values from the array pointed to by *data */ -int xtract_amdf(float *data, int N, void *argv, float *result); +int xtract_amdf(const float *data, const int N, const void *argv, float *result); /** \brief Extract Average Squared Difference Function from time domain signal * @@ -96,7 +96,7 @@ int xtract_amdf(float *data, int N, void *argv, float *result); * \param *argv: a pointer to NULL * \param *result: the ASDF of N values from the array pointed to by *data */ -int xtract_asdf(float *data, int N, void *argv, float *result); +int xtract_asdf(const float *data, const int N, const void *argv, float *result); /** \brief Extract Bark band coefficients based on a method * \param *data: a pointer to the first element in an array of floats representing the magnitude spectrum of an audio vector @@ -107,7 +107,7 @@ int xtract_asdf(float *data, int N, void *argv, float *result); * The limits array pointed to by *argv must be obtained by first calling xtract_init_bark * */ -int xtract_bark_coefficients(float *data, int N, void *argv, float *result); +int xtract_bark_coefficients(const float *data, const int N, const void *argv, float *result); /** \brief Extract the frequency and amplitude of spectral peaks from a magnitude spectrum * \param *data: a pointer to the first element in an array of floats representing the magnitude spectrum of an audio vector @@ -115,7 +115,7 @@ int xtract_bark_coefficients(float *data, int N, void *argv, float *result); * \param *argv: a pointer to an array containing peak threshold as percentage below max peak, and sample rate * \param *result: a pointer to an array of size N, containing N/2 freqs and N/2 amplitudes, amplitudes are on a decibel scale with dbFS = 0 */ -int xtract_peaks(float *data, int N, void *argv, float *result); +int xtract_peaks(const float *data, const int N, const void *argv, float *result); /** \brief Extract the harmonic spectrum of from a of a peak spectrum * \param *data: a pointer to the first element in an array of floats representing the peak spectrum of an audio vector (e.g. *result from xtract_peaks). It is expected that the first half of the array pointed to by *data will contain frequencies for each peak considered, and the the second half will contain the respective amplitudes @@ -123,7 +123,7 @@ int xtract_peaks(float *data, int N, void *argv, float *result); * \param *argv: a pointer to an array containing the fundamental (f0) of the spectrum, and a threshold (t) where 0<=t<=1.0, and t determines the distance from the nearest harmonic number within which a partial can be considered harmonic. * \param *result: a pointer to an array of size N containing N/2 freqs and N/2 amplitudes, amplitudes are on a decibel scale with dbFS = 0 */ -int xtract_harmonics(float *data, int N, void *argv, float *result); +int xtract_harmonics(const float *data, const int N, const void *argv, float *result); /** @} */ |