summaryrefslogtreecommitdiffstats
path: root/apps/dsp.c
blob: 8064a881a344a8ae2906b32e3d41556a84fa8692 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
/***************************************************************************
 *             __________               __   ___.
 *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
 *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
 *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
 *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
 *                     \/            \/     \/    \/            \/
 * $Id$
 *
 * Copyright (C) 2005 Miika Pekkarinen
 *
 * All files in this archive are subject to the GNU General Public License.
 * See the file COPYING in the source tree root for full license agreement.
 *
 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
 * KIND, either express or implied.
 *
 ****************************************************************************/
#include <inttypes.h>
#include <string.h>
#include "dsp.h"
#include "kernel.h"
#include "playback.h"
#include "system.h"
#include "settings.h"
#include "replaygain.h"
#include "debug.h"

/* The "dither" code to convert the 24-bit samples produced by libmad was
 * taken from the coolplayer project - coolplayer.sourceforge.net
 */

/* 16-bit samples are scaled based on these constants. The shift should be
 * no more than 15.
 */
#define WORD_SHIFT          12
#define WORD_FRACBITS       27

#define NATIVE_DEPTH        16
#define SAMPLE_BUF_SIZE     256
#define RESAMPLE_BUF_SIZE   (256 * 4)   /* Enough for 11,025 Hz -> 44,100 Hz*/
#define DEFAULT_REPLAYGAIN  0x01000000

#if defined(CPU_COLDFIRE) && !defined(SIMULATOR)

#define INIT() asm volatile ("move.l #0xb0, %macsr") /* frac, round, clip */
/* Multiply two S.31 fractional integers and return the sign bit and the
 * 31 most significant bits of the result.
 */
#define FRACMUL(x, y) \
({ \
    long t; \
    asm volatile ("mac.l    %[a], %[b], %%acc0\n\t" \
                  "movclr.l %%acc0, %[t]\n\t" \
                  : [t] "=r" (t) : [a] "r" (x), [b] "r" (y)); \
    t; \
})

/* Multiply one S.31-bit and one S8.23 fractional integer and return the
 * sign bit and the 31 most significant bits of the result.
 */
#define FRACMUL_8(x, y) \
({ \
    long t; \
    long u; \
    asm volatile ("mac.l    %[a], %[b], %%acc0\n\t" \
                  "move.l   %%accext01, %[u]\n\t" \
                  "movclr.l %%acc0, %[t]\n\t" \
                  : [t] "=r" (t), [u] "=r" (u) : [a] "r" (x), [b] "r" (y)); \
    (t << 8) | (u & 0xff); \
})

/* Multiply one S.31-bit and one S8.23 fractional integer and return the
 * sign bit and the 31 most significant bits of the result. Load next value
 * to multiply with into x from s (and increase s); x must contain the
 * initial value.
 */
#define FRACMUL_8_LOOP(x, y, s) \
({ \
    long t; \
    long u; \
    asm volatile ("mac.l    %[a], %[b], (%[c])+, %[a], %%acc0\n\t" \
                  "move.l   %%accext01, %[u]\n\t" \
                  "movclr.l %%acc0, %[t]\n\t" \
                  : [a] "+r" (x), [c] "+a" (s), [t] "=r" (t), [u] "=r" (u) \
                  : [b] "r" (y)); \
    (t << 8) | (u & 0xff); \
})

#else

#define INIT()
#define FRACMUL(x, y) (long) (((((long long) (x)) * ((long long) (y))) >> 31))
#define FRACMUL_8(x, y) (long) (((((long long) (x)) * ((long long) (y))) >> 23))
#define FRACMUL_8_LOOP(x, y, s) \
({ \
    long t = x; \
    x = *(s)++; \
    (long) (((((long long) (t)) * ((long long) (y))) >> 23)); \
})

#endif

struct dsp_config
{
    long frequency;
    long clip_min;
    long clip_max;
    long track_gain;
    long album_gain;
    long track_peak;
    long album_peak;
    long replaygain;    /* Note that this is in S8.23 format. */
    int sample_depth;
    int sample_bytes;
    int stereo_mode;
    int frac_bits;
    bool dither_enabled;
    bool new_gain;
};

struct resample_data
{
    long last_sample;
    long phase;
    long delta;
};

struct dither_data
{
    long error[3];
    long random;
};

static struct dsp_config dsp_conf[2] IDATA_ATTR;
static struct dither_data dither_data[2] IDATA_ATTR;
static struct resample_data resample_data[2][2] IDATA_ATTR;

extern int current_codec;
struct dsp_config *dsp;

/* The internal format is 32-bit samples, non-interleaved, stereo. This
 * format is similar to the raw output from several codecs, so the amount
 * of copying needed is minimized for that case.
 */

static long sample_buf[SAMPLE_BUF_SIZE] IDATA_ATTR;
static long resample_buf[RESAMPLE_BUF_SIZE] IDATA_ATTR;


/* Convert at most count samples to the internal format, if needed. Returns
 * number of samples ready for further processing. Updates src to point
 * past the samples "consumed" and dst is set to point to the samples to
 * consume. Note that for mono, dst[0] equals dst[1], as there is no point
 * in processing the same data twice.
 */
static int convert_to_internal(char* src[], int count, long* dst[])
{
    count = MIN(SAMPLE_BUF_SIZE / 2, count);

    if ((dsp->sample_depth <= NATIVE_DEPTH)
        || (dsp->stereo_mode == STEREO_INTERLEAVED))
    {
        dst[0] = &sample_buf[0];
        dst[1] = (dsp->stereo_mode == STEREO_MONO)
            ? dst[0] : &sample_buf[SAMPLE_BUF_SIZE / 2];
    }
    else
    {
        dst[0] = (long*) src[0];
        dst[1] = (long*) ((dsp->stereo_mode == STEREO_MONO) ? src[0] : src[1]);
    }

    if (dsp->sample_depth <= NATIVE_DEPTH)
    {
        short* s0 = (short*) src[0];
        long* d0 = dst[0];
        long* d1 = dst[1];
        int scale = WORD_SHIFT;
        int i;

        if (dsp->stereo_mode == STEREO_INTERLEAVED)
        {
            for (i = 0; i < count; i++)
            {
                *d0++ = *s0++ << scale;
                *d1++ = *s0++ << scale;
            }
        }
        else if (dsp->stereo_mode == STEREO_NONINTERLEAVED)
        {
            short* s1 = (short*) src[1];

            for (i = 0; i < count; i++)
            {
                *d0++ = *s0++ << scale;
                *d1++ = *s1++ << scale;
            }
        }
        else
        {
            for (i = 0; i < count; i++)
            {
                *d0++ = *s0++ << scale;
            }
        }
    }
    else if (dsp->stereo_mode == STEREO_INTERLEAVED)
    {
        long* s0 = (long*) src[0];
        long* d0 = dst[0];
        long* d1 = dst[1];
        int i;

        for (i = 0; i < count; i++)
        {
            *d0++ = *s0++;
            *d1++ = *s0++;
        }
    }

    if (dsp->stereo_mode == STEREO_NONINTERLEAVED)
    {
        src[0] += count * dsp->sample_bytes;
        src[1] += count * dsp->sample_bytes;
    }
    else if (dsp->stereo_mode == STEREO_INTERLEAVED)
    {
        src[0] += count * dsp->sample_bytes * 2;
    }
    else
    {
        src[0] += count * dsp->sample_bytes;
    }

    return count;
}

/* Linear resampling that introduces a one sample delay, because of our
 * inability to look into the future at the end of a frame.
 */

static long downsample(long *dst, long *src, int count,
    struct resample_data *r)
{
    long phase = r->phase;
    long delta = r->delta;
    long last_sample = r->last_sample;
    int pos = phase >> 16;
    int i = 1;

    /* Do we need last sample of previous frame for interpolation? */
    if (pos > 0)
    {
        last_sample = src[pos - 1];
    }

    *dst++ = last_sample + FRACMUL((phase & 0xffff) << 15,
        src[pos] - last_sample);
    phase += delta;

    while ((pos = phase >> 16) < count)
    {
        *dst++ = src[pos - 1] + FRACMUL((phase & 0xffff) << 15,
            src[pos] - src[pos - 1]);
        phase += delta;
        i++;
    }

    /* Wrap phase accumulator back to start of next frame. */
    r->phase = phase - (count << 16);
    r->delta = delta;
    r->last_sample = src[count - 1];
    return i;
}

static long upsample(long *dst, long *src, int count, struct resample_data *r)
{
    long phase = r->phase;
    long delta = r->delta;
    long last_sample = r->last_sample;
    int i = 0;
    int pos;

    while ((pos = phase >> 16) == 0)
    {
        *dst++ = last_sample + FRACMUL((phase & 0xffff) << 15,
            src[pos] - last_sample);
        phase += delta;
        i++;
    }

    while ((pos = phase >> 16) < count)
    {
        *dst++ = src[pos - 1] + FRACMUL((phase & 0xffff) << 15,
            src[pos] - src[pos - 1]);
        phase += delta;
        i++;
    }

    /* Wrap phase accumulator back to start of next frame. */
    r->phase = phase - (count << 16);
    r->delta = delta;
    r->last_sample = src[count - 1];
    return i;
}

/* Resample count stereo samples. Updates the src array, if resampling is
 * done, to refer to the resampled data. Returns number of stereo samples
 * for further processing.
 */
static inline int resample(long* src[], int count)
{
    long new_count;

    if (dsp->frequency != NATIVE_FREQUENCY)
    {
        long* d0 = &resample_buf[0];
        /* Only process the second channel if needed. */
        long* d1 = (src[0] == src[1]) ? d0
            : &resample_buf[RESAMPLE_BUF_SIZE / 2];

        if (dsp->frequency < NATIVE_FREQUENCY)
        {
            new_count = upsample(d0, src[0], count,
                            &resample_data[current_codec][0]);

            if (d0 != d1)
            {
                upsample(d1, src[1], count,
                    &resample_data[current_codec][1]);
            }
        }
        else
        {
            new_count = downsample(d0, src[0], count,
                            &resample_data[current_codec][0]);

            if (d0 != d1)
            {
                downsample(d1, src[1], count,
                    &resample_data[current_codec][1]);
            }
        }

        src[0] = d0;
        src[1] = d1;
    }
    else
    {
        new_count = count;
    }

    return new_count;
}

static inline long clip_sample(long sample, long min, long max)
{
    if (sample > max)
    {
        sample = max;
    }
    else if (sample < min)
    {
        sample = min;
    }

    return sample;
}

/* The "dither" code to convert the 24-bit samples produced by libmad was
 * taken from the coolplayer project - coolplayer.sourceforge.net
 */

static long dither_sample(long sample, long bias, long mask,
    struct dither_data* dither)
{
    long output;
    long random;
    long min;
    long max;

    /* Noise shape and bias */

    sample += dither->error[0] - dither->error[1] + dither->error[2];
    dither->error[2] = dither->error[1];
    dither->error[1] = dither->error[0] / 2;

    output = sample + bias;

    /* Dither */

    random = dither->random * 0x0019660dL + 0x3c6ef35fL;
    sample += (random & mask) - (dither->random & mask);
    dither->random = random;

    /* Clip and quantize */

    min = dsp->clip_min;
    max = dsp->clip_max;
    sample = clip_sample(sample, min, max);
    output = clip_sample(output, min, max) & ~mask;

    /* Error feedback */

    dither->error[0] = sample - output;

    return output;
}

/* Apply a constant gain to the samples (e.g., for ReplayGain). May update
 * the src array if gain was applied.
 * Note that this must be called before the resampler.
 */
static void apply_gain(long* src[], int count)
{
    if (dsp->replaygain)
    {
        long* s0 = src[0];
        long* s1 = src[1];
        long* d0 = &sample_buf[0];
        long* d1 = (s0 == s1) ? d0 : &sample_buf[SAMPLE_BUF_SIZE / 2];
        long gain = dsp->replaygain;
        long s;
        long i;

        src[0] = d0;
        src[1] = d1;
        s = *s0++;

        for (i = 0; i < count; i++)
        {
            *d0++ = FRACMUL_8_LOOP(s, gain, s0);
        }

        if (src [0] != src [1])
        {
            s = *s1++;

            for (i = 0; i < count; i++)
            {
                *d1++ = FRACMUL_8_LOOP(s, gain, s1);
            }
        }
    }
}

static void write_samples(short* dst, long* src[], int count)
{
    long* s0 = src[0];
    long* s1 = src[1];
    int scale = dsp->frac_bits + 1 - NATIVE_DEPTH;

    if (dsp->dither_enabled)
    {
        long bias = (1L << (dsp->frac_bits - NATIVE_DEPTH));
        long mask = (1L << scale) - 1;

        while (count-- > 0)
        {
            *dst++ = (short) (dither_sample(*s0++, bias, mask, &dither_data[0])
                >> scale);
            *dst++ = (short) (dither_sample(*s1++, bias, mask, &dither_data[1])
                >> scale);
        }
    }
    else
    {
        long min = dsp->clip_min;
        long max = dsp->clip_max;

        while (count-- > 0)
        {
            *dst++ = (short) (clip_sample(*s0++, min, max) >> scale);
            *dst++ = (short) (clip_sample(*s1++, min, max) >> scale);
        }
    }
}

/* Process and convert src audio to dst based on the DSP configuration,
 * reading size bytes of audio data. dst is assumed to be large enough; use
 * dst_get_dest_size() to get the required size. src is an array of
 * pointers; for mono and interleaved stereo, it contains one pointer to the
 * start of the audio data; for non-interleaved stereo, it contains two
 * pointers, one for each audio channel. Returns number of bytes written to
 * dest.
 */
long dsp_process(char* dst, char* src[], long size)
{
    long* tmp[2];
    long written = 0;
    long factor;
    int samples;

    dsp = &dsp_conf[current_codec];

    factor = (dsp->stereo_mode != STEREO_MONO) ? 2 : 1;
    size /= dsp->sample_bytes * factor;
    INIT();
    dsp_set_replaygain(false);

    while (size > 0)
    {
        samples = convert_to_internal(src, size, tmp);
        size -= samples;
        apply_gain(tmp, samples);
        samples = resample(tmp, samples);
        write_samples((short*) dst, tmp, samples);
        written += samples;
        dst += samples * sizeof(short) * 2;
        yield();
    }

    return written * sizeof(short) * 2;
}

/* Given size bytes of input data, calculate the maximum number of bytes of
 * output data that would be generated (the calculation is not entirely
 * exact and rounds upwards to be on the safe side; during resampling,
 * the number of samples generated depends on the current state of the
 * resampler).
 */
/* dsp_input_size MUST be called afterwards */
long dsp_output_size(long size)
{
    dsp = &dsp_conf[current_codec];
    
    if (dsp->sample_depth > NATIVE_DEPTH)
    {
        size /= 2;
    }

    if (dsp->frequency != NATIVE_FREQUENCY)
    {
        size = (long) ((((unsigned long) size * NATIVE_FREQUENCY)
            + (dsp->frequency - 1)) / dsp->frequency);
    }

    /* round to the next multiple of 2 (these are shorts) */
    size = (size + 1) & ~1;

    if (dsp->stereo_mode == STEREO_MONO)
    {
        size *= 2;
    }

    /* now we have the size in bytes for two resampled channels,
     * and the size in (short) must not exceed RESAMPLE_BUF_SIZE to
     * avoid resample buffer overflow. One must call dsp_input_size()
     * to get the correct input buffer size. */
    if (size > RESAMPLE_BUF_SIZE*2)
        size = RESAMPLE_BUF_SIZE*2;

    return size;
}

/* Given size bytes of output buffer, calculate number of bytes of input
 * data that would be consumed in order to fill the output buffer.
 */
long dsp_input_size(long size)
{
    dsp = &dsp_conf[current_codec];
    
    /* convert to number of output stereo samples. */
    size /= 2;

    /* Mono means we need half input samples to fill the output buffer */
    if (dsp->stereo_mode == STEREO_MONO)
        size /= 2;

    /* size is now the number of resampled input samples. Convert to
       original input samples. */
    if (dsp->frequency != NATIVE_FREQUENCY)
    {
        /* Use the real resampling delta =
         *  (unsigned long) dsp->frequency * 65536 / NATIVE_FREQUENCY, and
         * round towards zero to avoid buffer overflows. */
        size = ((unsigned long)size *
            resample_data[current_codec][0].delta) >> 16;
    }

    /* Convert back to bytes. */
    if (dsp->sample_depth > NATIVE_DEPTH)
        size *= 4;
    else
        size *= 2;

    return size;
}

int dsp_stereo_mode(void)
{
    dsp = &dsp_conf[current_codec];
    
    return dsp->stereo_mode;
}

bool dsp_configure(int setting, void *value)
{
    dsp = &dsp_conf[current_codec];

    switch (setting)
    {
    case DSP_SET_FREQUENCY:
        memset(&resample_data[current_codec][0], 0,
            sizeof(struct resample_data) * 2);
        /* Fall through!!! */
    case DSP_SWITCH_FREQUENCY:
        dsp->frequency = ((int) value == 0) ? NATIVE_FREQUENCY : (int) value;
        resample_data[current_codec][0].delta =
            resample_data[current_codec][1].delta =
            (unsigned long) dsp->frequency * 65536 / NATIVE_FREQUENCY;
        break;

    case DSP_SET_CLIP_MIN:
        dsp->clip_min = (long) value;
        break;

    case DSP_SET_CLIP_MAX:
        dsp->clip_max = (long) value;
        break;

    case DSP_SET_SAMPLE_DEPTH:
        dsp->sample_depth = (long) value;

        if (dsp->sample_depth <= NATIVE_DEPTH)
        {
            dsp->frac_bits = WORD_FRACBITS;
            dsp->sample_bytes = sizeof(short);
            dsp->clip_max =  ((1 << WORD_FRACBITS) - 1);
            dsp->clip_min = -((1 << WORD_FRACBITS));
        }
        else
        {
            dsp->frac_bits = (long) value;
            dsp->sample_bytes = sizeof(long);
        }

        break;

    case DSP_SET_STEREO_MODE:
        dsp->stereo_mode = (long) value;
        break;

    case DSP_RESET:
        dsp->dither_enabled = false;
        dsp->stereo_mode = STEREO_NONINTERLEAVED;
        dsp->clip_max =  ((1 << WORD_FRACBITS) - 1);
        dsp->clip_min = -((1 << WORD_FRACBITS));
        dsp->track_gain = 0;
        dsp->album_gain = 0;
        dsp->track_peak = 0;
        dsp->album_peak = 0;
        dsp->frequency = NATIVE_FREQUENCY;
        dsp->sample_depth = NATIVE_DEPTH;
        dsp->frac_bits = WORD_FRACBITS;
        dsp->new_gain = true;
        break;

    case DSP_DITHER:
        memset(dither_data, 0, sizeof(dither_data));
        dsp->dither_enabled = (bool) value;
        break;

    case DSP_SET_TRACK_GAIN:
        dsp->track_gain = (long) value;
        dsp->new_gain = true;
        break;

    case DSP_SET_ALBUM_GAIN:
        dsp->album_gain = (long) value;
        dsp->new_gain = true;
        break;

    case DSP_SET_TRACK_PEAK:
        dsp->track_peak = (long) value;
        dsp->new_gain = true;
        break;

    case DSP_SET_ALBUM_PEAK:
        dsp->album_peak = (long) value;
        dsp->new_gain = true;
        break;

    default:
        return 0;
    }

    return 1;
}

void dsp_set_replaygain(bool always)
{
    dsp = &dsp_conf[current_codec];
    
    if (always || dsp->new_gain)
    {
        long gain = 0;

        dsp->new_gain = false;

        if (global_settings.replaygain || global_settings.replaygain_noclip)
        {
            long peak;

            if (global_settings.replaygain)
            {
                gain = (global_settings.replaygain_track || !dsp->album_gain)
                    ? dsp->track_gain : dsp->album_gain;

                if (global_settings.replaygain_preamp)
                {
                    long preamp = get_replaygain_int(
                        global_settings.replaygain_preamp * 10);

                    gain = (long) ((((int64_t) gain * preamp)) >> 24);
                }
            }

            peak = (global_settings.replaygain_track || !dsp->album_peak)
                ? dsp->track_peak : dsp->album_peak;

            if (gain == 0)
            {
                /* So that noclip can work even with no gain information. */
                gain = DEFAULT_REPLAYGAIN;
            }

            if (global_settings.replaygain_noclip && (peak != 0)
                && ((((int64_t) gain * peak) >> 24) >= DEFAULT_REPLAYGAIN))
            {
                gain = (((int64_t) DEFAULT_REPLAYGAIN << 24) / peak);
            }

            if (gain == DEFAULT_REPLAYGAIN)
            {
                /* Nothing to do, disable processing. */
                gain = 0;

            }
        }

        /* Store in S8.23 format to simplify calculations. */
        dsp->replaygain = gain >> 1;
    }
}