X-Git-Url: https://git.sesse.net/?p=c64tapwav;a=blobdiff_plain;f=decode.cpp;h=f9e525eeea07b8ebaf351d87d9870be36d1c22cb;hp=ab242d0be5845f7145e0b40ca75c5a646207ca25;hb=611a1748946aa8e9afe7faf7043a10e52654484e;hpb=b0572bf7f70621037e6692c392965f2e85ea2590 diff --git a/decode.cpp b/decode.cpp index ab242d0..f9e525e 100644 --- a/decode.cpp +++ b/decode.cpp @@ -1,3 +1,6 @@ +// Copyright Steinar H. Gunderson +// Licensed under the GPL, v2. (See the file COPYING.) + #include #include #include @@ -11,6 +14,7 @@ #include "interpolate.h" #include "level.h" #include "tap.h" +#include "filter.h" #define BUFSIZE 4096 #define C64_FREQUENCY 985248 @@ -21,6 +25,7 @@ // SPSA options #define NUM_FILTER_COEFF 32 +#define NUM_SPSA_VALS (NUM_FILTER_COEFF + 2) #define NUM_ITER 5000 #define A NUM_ITER/10 // approx #define INITIAL_A 0.005 // A bit of trial and error... @@ -28,38 +33,44 @@ #define GAMMA 0.166 #define ALPHA 1.0 -static float hysteresis_limit = 3000.0 / 32768.0; +static float hysteresis_upper_limit = 0.1; +static float hysteresis_lower_limit = -0.1; static bool do_calibrate = true; static bool output_cycles_plot = false; -static bool use_filter = false; static bool do_crop = false; static float crop_start = 0.0f, crop_end = HUGE_VAL; + +static bool use_fir_filter = false; static float filter_coeff[NUM_FILTER_COEFF] = { 1.0f }; // The rest is filled with 0. +static bool use_rc_filter = false; +static float rc_filter_freq; static bool output_filtered = false; + static bool quiet = false; static bool do_auto_level = false; static bool output_leveled = false; static std::vector train_snap_points; static bool do_train = false; -// between [x,x+1] -double find_zerocrossing(const std::vector &pcm, int x) -{ - if (pcm[x] == 0) { - return x; - } - if (pcm[x + 1] == 0) { - return x + 1; - } +// The frequency to filter on (for do_auto_level), in Hertz. +// Larger values makes the compressor react faster, but if it is too large, +// you'll ruin the waveforms themselves. +static float auto_level_freq = 200.0; - assert(pcm[x + 1] < 0); - assert(pcm[x] > 0); +// The minimum estimated sound level (for do_auto_level) at any given point. +// If you decrease this, you'll be able to amplify really silent signals +// by more, but you'll also increase the level of silent (ie. noise-only) segments, +// possibly caused misdetected pulses in these segments. +static float min_level = 0.05f; +// search for the value between [x,x+1] +double find_crossing(const std::vector &pcm, int x, float limit) +{ double upper = x; double lower = x + 1; while (lower - upper > 1e-3) { double mid = 0.5f * (upper + lower); - if (lanczos_interpolate(pcm, mid) > 0) { + if (lanczos_interpolate(pcm, mid) > limit) { upper = mid; } else { lower = mid; @@ -164,10 +175,14 @@ void output_tap(const std::vector& pulses, double calibration_factor) static struct option long_options[] = { {"auto-level", 0, 0, 'a' }, + {"auto-level-freq", required_argument, 0, 'b' }, + {"output-leveled", 0, 0, 'A' }, + {"min-level", required_argument, 0, 'm' }, {"no-calibrate", 0, 0, 's' }, {"plot-cycles", 0, 0, 'p' }, {"hysteresis-limit", required_argument, 0, 'l' }, {"filter", required_argument, 0, 'f' }, + {"rc-filter", required_argument, 0, 'r' }, {"output-filtered", 0, 0, 'F' }, {"crop", required_argument, 0, 'c' }, {"quiet", 0, 0, 'q' }, @@ -180,11 +195,14 @@ void help() fprintf(stderr, "decode [OPTIONS] AUDIO-FILE > TAP-FILE\n"); fprintf(stderr, "\n"); fprintf(stderr, " -a, --auto-level automatically adjust amplitude levels throughout the file\n"); + fprintf(stderr, " -b, --auto-level-freq minimum frequency in Hertz of corrected level changes (default 200 Hz)\n"); fprintf(stderr, " -A, --output-leveled output leveled waveform to leveled.raw\n"); + fprintf(stderr, " -m, --min-level minimum estimated sound level (0..1) for --auto-level\n"); fprintf(stderr, " -s, --no-calibrate do not try to calibrate on sync pulse length\n"); fprintf(stderr, " -p, --plot-cycles output debugging info to cycles.plot\n"); - fprintf(stderr, " -l, --hysteresis-limit VAL change amplitude threshold for ignoring pulses (0..32768)\n"); + fprintf(stderr, " -l, --hysteresis-limit U[:L] change amplitude threshold for ignoring pulses (-1..1)\n"); fprintf(stderr, " -f, --filter C1:C2:C3:... specify FIR filter (up to %d coefficients)\n", NUM_FILTER_COEFF); + fprintf(stderr, " -r, --rc-filter FREQ send signal through a highpass RC filter with given frequency (in Hertz)\n"); fprintf(stderr, " -F, --output-filtered output filtered waveform to filtered.raw\n"); fprintf(stderr, " -c, --crop START[:END] use only the given part of the file\n"); fprintf(stderr, " -t, --train LEN1:LEN2:... train a filter for detecting any of the given number of cycles\n"); @@ -198,7 +216,7 @@ void parse_options(int argc, char **argv) { for ( ;; ) { int option_index = 0; - int c = getopt_long(argc, argv, "aAspl:f:Fc:t:qh", long_options, &option_index); + int c = getopt_long(argc, argv, "ab:Am:spl:f:r:Fc:t:qh", long_options, &option_index); if (c == -1) break; @@ -207,10 +225,18 @@ void parse_options(int argc, char **argv) do_auto_level = true; break; + case 'b': + auto_level_freq = atof(optarg); + break; + case 'A': output_leveled = true; break; + case 'm': + min_level = atof(optarg); + break; + case 's': do_calibrate = false; break; @@ -219,9 +245,17 @@ void parse_options(int argc, char **argv) output_cycles_plot = true; break; - case 'l': - hysteresis_limit = atof(optarg) / 32768.0; + case 'l': { + const char *hyststr = strtok(optarg, ": "); + hysteresis_upper_limit = atof(hyststr); + hyststr = strtok(NULL, ": "); + if (hyststr == NULL) { + hysteresis_lower_limit = -hysteresis_upper_limit; + } else { + hysteresis_lower_limit = atof(hyststr); + } break; + } case 'f': { const char *coeffstr = strtok(optarg, ": "); @@ -230,10 +264,15 @@ void parse_options(int argc, char **argv) filter_coeff[coeff_index++] = atof(coeffstr); coeffstr = strtok(NULL, ": "); } - use_filter = true; + use_fir_filter = true; break; } + case 'r': + use_rc_filter = true; + rc_filter_freq = atof(optarg); + break; + case 'F': output_filtered = true; break; @@ -290,7 +329,7 @@ std::vector crop(const std::vector& pcm, float crop_start, float c } // TODO: Support AVX here. -std::vector do_filter(const std::vector& pcm, const float* filter) +std::vector do_fir_filter(const std::vector& pcm, const float* filter) { std::vector filtered_pcm; filtered_pcm.reserve(pcm.size()); @@ -311,49 +350,60 @@ std::vector do_filter(const std::vector& pcm, const float* filter) return filtered_pcm; } -std::vector detect_pulses(const std::vector &pcm, int sample_rate) +std::vector do_rc_filter(const std::vector& pcm, float freq, int sample_rate) +{ + // This is only a 6 dB/oct filter, which seemingly works better + // than the Filter class, which is a standard biquad (12 dB/oct). + // The b/c calculations come from libnyquist (atone.c); + // I haven't checked, but I suppose they fall out of the bilinear + // transform of the transfer function H(s) = s/(s + w). + std::vector filtered_pcm; + filtered_pcm.resize(pcm.size()); + const float b = 2.0f - cos(2.0 * M_PI * freq / sample_rate); + const float c = b - sqrt(b * b - 1.0f); + float prev_in = 0.0f; + float prev_out = 0.0f; + for (unsigned i = 0; i < pcm.size(); ++i) { + float in = pcm[i]; + float out = c * (prev_out + in - prev_in); + filtered_pcm[i] = out; + prev_in = in; + prev_out = out; + } + + if (output_filtered) { + FILE *fp = fopen("filtered.raw", "wb"); + fwrite(filtered_pcm.data(), filtered_pcm.size() * sizeof(filtered_pcm[0]), 1, fp); + fclose(fp); + } + + return filtered_pcm; +} + +std::vector detect_pulses(const std::vector &pcm, float hysteresis_upper_limit, float hysteresis_lower_limit, int sample_rate) { std::vector pulses; // Find the flanks. - int last_bit = -1; + enum State { START, ABOVE, BELOW } state = START; double last_downflank = -1; for (unsigned i = 0; i < pcm.size(); ++i) { - int bit = (pcm[i] > 0) ? 1 : 0; - if (bit == 0 && last_bit == 1) { - // Check if we ever go up above before we dip down again. - bool true_pulse = false; - unsigned j; - int min_level_after = 32767; - for (j = i; j < pcm.size(); ++j) { - min_level_after = std::min(min_level_after, pcm[j]); - if (pcm[j] > 0) break; - if (pcm[j] < -hysteresis_limit) { - true_pulse = true; - break; + if (pcm[i] > hysteresis_upper_limit) { + state = ABOVE; + } else if (pcm[i] < hysteresis_lower_limit) { + if (state == ABOVE) { + // down-flank! + double t = find_crossing(pcm, i - 1, hysteresis_lower_limit) * (1.0 / sample_rate) + crop_start; + if (last_downflank > 0) { + pulse p; + p.time = t; + p.len = t - last_downflank; + pulses.push_back(p); } + last_downflank = t; } - - if (!true_pulse) { -#if 0 - fprintf(stderr, "Ignored down-flank at %.6f seconds due to hysteresis (%d < %d).\n", - double(i) / sample_rate, -min_level_after, hysteresis_limit); -#endif - i = j; - continue; - } - - // down-flank! - double t = find_zerocrossing(pcm, i - 1) * (1.0 / sample_rate) + crop_start; - if (last_downflank > 0) { - pulse p; - p.time = t; - p.len = t - last_downflank; - pulses.push_back(p); - } - last_downflank = t; + state = BELOW; } - last_bit = bit; } return pulses; } @@ -368,25 +418,81 @@ void output_cycle_plot(const std::vector &pulses, double calibration_fact fclose(fp); } +std::pair find_closest_point(double x, const std::vector &points) +{ + int best_point = 0; + double best_dist = (x - points[0]) * (x - points[0]); + for (unsigned j = 1; j < train_snap_points.size(); ++j) { + double dist = (x - points[j]) * (x - points[j]); + if (dist < best_dist) { + best_point = j; + best_dist = dist; + } + } + return std::make_pair(best_point, best_dist); +} + float eval_badness(const std::vector& pulses, double calibration_factor) { double sum_badness = 0.0; for (unsigned i = 0; i < pulses.size(); ++i) { double cycles = pulses[i].len * calibration_factor * C64_FREQUENCY; if (cycles > 2000.0) cycles = 2000.0; // Don't make pauses arbitrarily bad. - double badness = (cycles - train_snap_points[0]) * (cycles - train_snap_points[0]); - for (unsigned j = 1; j < train_snap_points.size(); ++j) { - badness = std::min(badness, (cycles - train_snap_points[j]) * (cycles - train_snap_points[j])); - } - sum_badness += badness; + std::pair selected_point_and_sq_dist = find_closest_point(cycles, train_snap_points); + sum_badness += selected_point_and_sq_dist.second; } return sqrt(sum_badness / (pulses.size() - 1)); } -void spsa_train(std::vector &pcm, int sample_rate) +void find_kmeans(const std::vector &pulses, double calibration_factor, const std::vector &initial_centers) +{ + std::vector last_centers = initial_centers; + std::vector sums; + std::vector num; + sums.resize(initial_centers.size()); + num.resize(initial_centers.size()); + for ( ;; ) { + for (unsigned i = 0; i < initial_centers.size(); ++i) { + sums[i] = 0.0f; + num[i] = 0; + } + for (unsigned i = 0; i < pulses.size(); ++i) { + double cycles = pulses[i].len * calibration_factor * C64_FREQUENCY; + // Ignore heavy outliers, which are almost always long pauses. + if (cycles > 2000.0) { + continue; + } + std::pair selected_point_and_sq_dist = find_closest_point(cycles, last_centers); + int p = selected_point_and_sq_dist.first; + sums[p] += cycles; + ++num[p]; + } + bool any_moved = false; + for (unsigned i = 0; i < initial_centers.size(); ++i) { + if (num[i] == 0) { + fprintf(stderr, "K-means broke down, can't output new reference training points\n"); + return; + } + float new_center = sums[i] / num[i]; + if (fabs(new_center - last_centers[i]) > 1e-3) { + any_moved = true; + } + last_centers[i] = new_center; + } + if (!any_moved) { + break; + } + } + fprintf(stderr, "New reference training points:"); + for (unsigned i = 0; i < last_centers.size(); ++i) { + fprintf(stderr, " %.3f", last_centers[i]); + } + fprintf(stderr, "\n"); +} + +void spsa_train(const std::vector &pcm, int sample_rate) { - // Train! - float filter[NUM_FILTER_COEFF] = { 1.0f }; // The rest is filled with 0. + float vals[NUM_SPSA_VALS] = { hysteresis_upper_limit, hysteresis_lower_limit, 1.0f }; // The rest is filled with 0. float start_c = INITIAL_C; double best_badness = HUGE_VAL; @@ -396,45 +502,47 @@ void spsa_train(std::vector &pcm, int sample_rate) float c = start_c * pow(n, -GAMMA); // find a random perturbation - float p[NUM_FILTER_COEFF]; - float filter1[NUM_FILTER_COEFF], filter2[NUM_FILTER_COEFF]; - for (int i = 0; i < NUM_FILTER_COEFF; ++i) { + float p[NUM_SPSA_VALS]; + float vals1[NUM_SPSA_VALS], vals2[NUM_SPSA_VALS]; + for (int i = 0; i < NUM_SPSA_VALS; ++i) { p[i] = (rand() % 2) ? 1.0 : -1.0; - filter1[i] = std::max(std::min(filter[i] - c * p[i], 1.0f), -1.0f); - filter2[i] = std::max(std::min(filter[i] + c * p[i], 1.0f), -1.0f); + vals1[i] = std::max(std::min(vals[i] - c * p[i], 1.0f), -1.0f); + vals2[i] = std::max(std::min(vals[i] + c * p[i], 1.0f), -1.0f); } - std::vector pulses1 = detect_pulses(do_filter(pcm, filter1), sample_rate); - std::vector pulses2 = detect_pulses(do_filter(pcm, filter2), sample_rate); + std::vector pulses1 = detect_pulses(do_fir_filter(pcm, vals1 + 2), vals1[0], vals1[1], sample_rate); + std::vector pulses2 = detect_pulses(do_fir_filter(pcm, vals2 + 2), vals2[0], vals2[1], sample_rate); float badness1 = eval_badness(pulses1, 1.0); float badness2 = eval_badness(pulses2, 1.0); // Find the gradient estimator - float g[NUM_FILTER_COEFF]; - for (int i = 0; i < NUM_FILTER_COEFF; ++i) { + float g[NUM_SPSA_VALS]; + for (int i = 0; i < NUM_SPSA_VALS; ++i) { g[i] = (badness2 - badness1) / (2.0 * c * p[i]); - filter[i] -= a * g[i]; - filter[i] = std::max(std::min(filter[i], 1.0f), -1.0f); + vals[i] -= a * g[i]; + vals[i] = std::max(std::min(vals[i], 1.0f), -1.0f); } if (badness2 < badness1) { std::swap(badness1, badness2); - std::swap(filter1, filter2); + std::swap(vals1, vals2); std::swap(pulses1, pulses2); } if (badness1 < best_badness) { - printf("\nNew best filter (badness=%f):", badness1); + fprintf(stderr, "\nNew best filter (badness=%f):", badness1); for (int i = 0; i < NUM_FILTER_COEFF; ++i) { - printf(" %.5f", filter1[i]); + fprintf(stderr, " %.5f", vals1[i + 2]); } + fprintf(stderr, ", hysteresis limits = %f %f\n", vals1[0], vals1[1]); best_badness = badness1; - printf("\n"); + + find_kmeans(pulses1, 1.0, train_snap_points); if (output_cycles_plot) { output_cycle_plot(pulses1, 1.0); } } - printf("%d ", n); - fflush(stdout); + fprintf(stderr, "%d ", n); + fflush(stderr); } } @@ -453,12 +561,16 @@ int main(int argc, char **argv) pcm = crop(pcm, crop_start, crop_end, sample_rate); } - if (use_filter) { - pcm = do_filter(pcm, filter_coeff); + if (use_fir_filter) { + pcm = do_fir_filter(pcm, filter_coeff); + } + + if (use_rc_filter) { + pcm = do_rc_filter(pcm, rc_filter_freq, sample_rate); } if (do_auto_level) { - pcm = level_samples(pcm, sample_rate); + pcm = level_samples(pcm, min_level, auto_level_freq, sample_rate); if (output_leveled) { FILE *fp = fopen("leveled.raw", "wb"); fwrite(pcm.data(), pcm.size() * sizeof(pcm[0]), 1, fp); @@ -483,7 +595,7 @@ int main(int argc, char **argv) exit(0); } - std::vector pulses = detect_pulses(pcm, sample_rate); + std::vector pulses = detect_pulses(pcm, hysteresis_upper_limit, hysteresis_lower_limit, sample_rate); double calibration_factor = 1.0; if (do_calibrate) {