Add some heuristics to deal with shorter initial sync periods.
[c64tapwav] / decode.cpp
index 798a872..8ed2f1b 100644 (file)
@@ -1,69 +1,24 @@
 #include <stdio.h>
 #include <string.h>
 #include <math.h>
-#include <unistd.h>
 #include <assert.h>
+#include <limits.h>
 #include <vector>
 #include <algorithm>
 
-#define LANCZOS_RADIUS 30
+#include "audioreader.h"
+#include "interpolate.h"
+#include "tap.h"
+
 #define BUFSIZE 4096
 #define HYSTERESIS_LIMIT 3000
-#define SAMPLE_RATE 44100
 #define C64_FREQUENCY 985248
-#define TAP_RESOLUTION 8
 
 #define SYNC_PULSE_START 1000
-#define SYNC_PULSE_END 15000
-#define SYNC_PULSE_LENGTH 380.0
+#define SYNC_PULSE_END 20000
+#define SYNC_PULSE_LENGTH 378.0
 #define SYNC_TEST_TOLERANCE 1.10
 
-struct tap_header {
-       char identifier[12];
-       char version;
-       char reserved[3];
-       unsigned int data_len;
-};
-
-double sinc(double x)
-{
-       if (fabs(x) < 1e-6) {
-               return 1.0f - fabs(x);
-       } else {
-               return sin(x) / x;
-       }
-}
-
-#if 0
-double weight(double x)
-{
-       if (fabs(x) > LANCZOS_RADIUS) {
-               return 0.0f;
-       }
-       return sinc(M_PI * x) * sinc(M_PI * x / LANCZOS_RADIUS);
-}
-#else
-double weight(double x)
-{
-       if (fabs(x) > 1.0f) {
-               return 0.0f;
-       }
-       return 1.0f - fabs(x);
-}
-#endif
-
-double interpolate(const std::vector<short> &pcm, double i)
-{
-       int lower = std::max<int>(ceil(i - LANCZOS_RADIUS), 0);
-       int upper = std::min<int>(floor(i + LANCZOS_RADIUS), pcm.size() - 1);
-       double sum = 0.0f;
-
-       for (int x = lower; x <= upper; ++x) {
-               sum += pcm[x] * weight(i - x);
-       }
-       return sum;
-}
-       
 // between [x,x+1]
 double find_zerocrossing(const std::vector<short> &pcm, int x)
 {
@@ -74,14 +29,14 @@ double find_zerocrossing(const std::vector<short> &pcm, int x)
                return x + 1;
        }
 
-       assert(pcm[x + 1] > 0);
-       assert(pcm[x] < 0);
+       assert(pcm[x + 1] < 0);
+       assert(pcm[x] > 0);
 
-       double lower = x;
-       double upper = x + 1;
-       while (upper - lower > 1e-6) {
+       double upper = x;
+       double lower = x + 1;
+       while (lower - upper > 1e-3) {
                double mid = 0.5f * (upper + lower);
-               if (interpolate(pcm, mid) > 0) {
+               if (lanczos_interpolate(pcm, mid) > 0) {
                        upper = mid;
                } else {
                        lower = mid;
@@ -95,18 +50,68 @@ struct pulse {
        double time;  // in seconds from start
        double len;   // in seconds
 };
+
+// Calibrate on the first ~25k pulses (skip a few, just to be sure).
+double calibrate(const std::vector<pulse> &pulses) {
+       if (pulses.size() < SYNC_PULSE_END) {
+               fprintf(stderr, "Too few pulses, not calibrating!\n");
+               return 1.0;
+       }
+
+       int sync_pulse_end = -1;
+       double sync_pulse_stddev = -1.0;
+
+       // Compute the standard deviation (to check for uneven speeds).
+       // If it suddenly skyrockets, we assume that sync ended earlier
+       // than we thought (it should be 25000 cycles), and that we should
+       // calibrate on fewer cycles.
+       for (int try_end : { 2000, 4000, 5000, 7500, 10000, 15000, SYNC_PULSE_END }) {
+               double sum2 = 0.0;
+               for (int i = SYNC_PULSE_START; i < try_end; ++i) {
+                       double cycles = pulses[i].len * C64_FREQUENCY;
+                       sum2 += (cycles - SYNC_PULSE_LENGTH) * (cycles - SYNC_PULSE_LENGTH);
+               }
+               double stddev = sqrt(sum2 / (try_end - SYNC_PULSE_START - 1));
+               if (sync_pulse_end != -1 && stddev > 5.0 && stddev / sync_pulse_stddev > 1.3) {
+                       fprintf(stderr, "Stopping at %d sync pulses because standard deviation would be too big (%.2f cycles); shorter-than-usual trailer?\n",
+                               sync_pulse_end, stddev);
+                       break;
+               }
+               sync_pulse_end = try_end;
+               sync_pulse_stddev = stddev;
+       }
+       fprintf(stderr, "Sync pulse length standard deviation: %.2f cycles\n",
+               sync_pulse_stddev);
+
+       double sum = 0.0;
+       for (int i = SYNC_PULSE_START; i < sync_pulse_end; ++i) {
+               sum += pulses[i].len;
+       }
+       double mean_length = C64_FREQUENCY * sum / (sync_pulse_end - SYNC_PULSE_START);
+       double calibration_factor = SYNC_PULSE_LENGTH / mean_length;
+       fprintf(stderr, "Calibrated sync pulse length: %.2f -> %.2f (change %+.2f%%)\n",
+               mean_length, SYNC_PULSE_LENGTH, 100.0 * (calibration_factor - 1.0));
+
+       // Check for pulses outside +/- 10% (sign of misdetection).
+       for (int i = SYNC_PULSE_START; i < sync_pulse_end; ++i) {
+               double cycles = pulses[i].len * calibration_factor * C64_FREQUENCY;
+               if (cycles < SYNC_PULSE_LENGTH / SYNC_TEST_TOLERANCE || cycles > SYNC_PULSE_LENGTH * SYNC_TEST_TOLERANCE) {
+                       fprintf(stderr, "Sync cycle with downflank at %.6f was detected at %.0f cycles; misdetect?\n",
+                               pulses[i].time, cycles);
+               }
+       }
+
+       return calibration_factor;
+}
        
 int main(int argc, char **argv)
 {
+       make_lanczos_weight_table();
        std::vector<short> pcm;
-
-       while (!feof(stdin)) {
-               short buf[BUFSIZE];
-               ssize_t ret = fread(buf, 2, BUFSIZE, stdin);
-               if (ret >= 0) {
-                       pcm.insert(pcm.end(), buf, buf + ret);
-               }
-       }       
+       int sample_rate;
+       if (!read_audio_file(argv[1], &pcm, &sample_rate)) {
+               exit(1);
+       }
 
 #if 0
        for (int i = 0; i < LEN; ++i) {
@@ -124,66 +129,55 @@ int main(int argc, char **argv)
 
        // Find the flanks.
        int last_bit = -1;
-       double last_upflank = -1;
-       int last_max_level = 0;
+       double last_downflank = -1;
        for (unsigned i = 0; i < pcm.size(); ++i) {
                int bit = (pcm[i] > 0) ? 1 : 0;
-               if (bit == 1 && last_bit == 0 && last_max_level > HYSTERESIS_LIMIT) {
-                       // up-flank!
-                       double t = find_zerocrossing(pcm, i - 1) * (1.0 / SAMPLE_RATE);
-                       if (last_upflank > 0) {
+               if (bit == 0 && last_bit == 1) {
+                       // Check if we ever go up above HYSTERESIS_LIMIT 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<int>(min_level_after, pcm[j]);
+                               if (pcm[j] > 0) break;
+                               if (pcm[j] < -HYSTERESIS_LIMIT) {
+                                       true_pulse = true;
+                                       break;
+                               }
+                       }
+
+                       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);
+                       if (last_downflank > 0) {
                                pulse p;
                                p.time = t;
-                               p.len = t - last_upflank;
+                               p.len = t - last_downflank;
                                pulses.push_back(p);
                        }
-                       last_upflank = t;
-                       last_max_level = 0;
+                       last_downflank = t;
                }
-               last_max_level = std::max(last_max_level, abs(pcm[i]));
                last_bit = bit;
        }
 
-       // Calibrate on the first ~25k pulses (skip a few, just to be sure).
-       double calibration_factor = 1.0f;
-       if (pulses.size() < SYNC_PULSE_END) {
-               fprintf(stderr, "Too few pulses, not calibrating!\n");
-       } else {
-               double sum = 0.0;
-               for (int i = SYNC_PULSE_START; i < SYNC_PULSE_END; ++i) {
-                       sum += pulses[i].len;
-               }
-               double mean_length = C64_FREQUENCY * sum / (SYNC_PULSE_END - SYNC_PULSE_START);
-               calibration_factor = SYNC_PULSE_LENGTH / mean_length;
-               fprintf(stderr, "Calibrated sync pulse length: %.2f -> 380.0 (change %+.2f%%)\n",
-                       mean_length, 100.0 * (calibration_factor - 1.0));
-
-               // Check for pulses outside +/- 10% (sign of misdetection).
-               for (int i = SYNC_PULSE_START; i < SYNC_PULSE_END; ++i) {
-                       double cycles = pulses[i].len * calibration_factor * C64_FREQUENCY;
-                       if (cycles < SYNC_PULSE_LENGTH / SYNC_TEST_TOLERANCE || cycles > SYNC_PULSE_LENGTH * SYNC_TEST_TOLERANCE) {
-                               fprintf(stderr, "Sync cycle with upflank at %.6f was detected at %.0f cycles; misdetect?\n",
-                                       pulses[i].time, cycles);
-                       }
-               }
-
-               // Compute the standard deviation (to check for uneven speeds).
-               double sum2 = 0.0;
-               for (int i = SYNC_PULSE_START; i < SYNC_PULSE_END; ++i) {
-                       double cycles = pulses[i].len * calibration_factor * C64_FREQUENCY;
-                       sum2 += (cycles - SYNC_PULSE_LENGTH) * (cycles - SYNC_PULSE_LENGTH);
-               }
-               double stddev = sqrt(sum2 / (SYNC_PULSE_END - SYNC_PULSE_START - 1));
-               fprintf(stderr, "Sync pulse length standard deviation: %.2f cycles\n",
-                       stddev);
-       }
+       double calibration_factor = calibrate(pulses);
 
+       FILE *fp = fopen("cycles.plot", "w");
        std::vector<char> tap_data;
        for (unsigned i = 0; i < pulses.size(); ++i) {
                double cycles = pulses[i].len * calibration_factor * C64_FREQUENCY;
+               fprintf(fp, "%f %f\n", pulses[i].time, cycles);
                int len = lrintf(cycles / TAP_RESOLUTION);
                if (i > SYNC_PULSE_END && (cycles < 100 || cycles > 800)) {
-                       fprintf(stderr, "Cycle with upflank at %.6f was detected at %.0f cycles; misdetect?\n",
+                       fprintf(stderr, "Cycle with downflank at %.6f was detected at %.0f cycles; misdetect?\n",
                                        pulses[i].time, cycles);
                }
                if (len <= 255) {
@@ -196,6 +190,7 @@ int main(int argc, char **argv)
                        tap_data.push_back(overflow_len >> 16);
                }
        }
+       fclose(fp);
 
        tap_header hdr;
        memcpy(hdr.identifier, "C64-TAPE-RAW", 12);
@@ -205,4 +200,20 @@ int main(int argc, char **argv)
 
        fwrite(&hdr, sizeof(hdr), 1, stdout);
        fwrite(tap_data.data(), tap_data.size(), 1, stdout);
+
+       // Output a debug raw file with pulse detection points.
+       fp = fopen("debug.raw", "wb");
+       short one = 32767;
+       short zero = 0;
+       unsigned pulsenum = 0;
+       for (unsigned i = 0; i < pcm.size(); ++i) {
+               unsigned next_pulse = (pulsenum >= pulses.size()) ? INT_MAX : int(pulses[pulsenum].time * sample_rate);
+               if (i >= next_pulse) {
+                       fwrite(&one, sizeof(one), 1, fp);
+                       ++pulsenum;
+               } else {
+                       fwrite(&zero, sizeof(zero), 1, fp);
+               }
+       }
+       fclose(fp);
 }