gstaudioresample.c 18.9 KB
Newer Older
1
/* GStreamer
2
 * Copyright (C) 1999 Erik Walthinsen <omega@cse.ogi.edu>
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
 * Copyright (C) 2003,2004 David A. Schleef <ds@schleef.org>
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Library General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Library General Public License for more details.
 *
 * You should have received a copy of the GNU Library General Public
 * License along with this library; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 02111-1307, USA.
 */
/* Element-Checklist-Version: 5 */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
25

26 27 28 29 30 31
#include <string.h>
#include <math.h>

/*#define DEBUG_ENABLED */
#include "gstaudioresample.h"
#include <gst/audio/audio.h>
32
#include <gst/base/gstbasetransform.h>
33

34
GST_DEBUG_CATEGORY (audioresample_debug);
35 36 37 38 39 40 41 42 43
#define GST_CAT_DEFAULT audioresample_debug

/* elementfactory information */
static GstElementDetails gst_audioresample_details =
GST_ELEMENT_DETAILS ("Audio scaler",
    "Filter/Converter/Audio",
    "Resample audio",
    "David Schleef <ds@schleef.org>");

44
/* GstAudioresample signals and args */
45 46 47 48 49 50
enum
{
  /* FILL ME */
  LAST_SIGNAL
};

51 52
#define DEFAULT_FILTERLEN	16

53 54 55 56 57 58 59
enum
{
  ARG_0,
  ARG_FILTERLEN
};

#define SUPPORTED_CAPS \
60
GST_STATIC_CAPS ( \
61 62 63 64 65 66
    "audio/x-raw-int, " \
      "rate = (int) [ 1, MAX ], " \
      "channels = (int) [ 1, MAX ], " \
      "endianness = (int) BYTE_ORDER, " \
      "width = (int) 16, " \
      "depth = (int) 16, " \
67 68
      "signed = (boolean) true " \
)
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91

#if 0
  /* disabled because it segfaults */
"audio/x-raw-float, "
    "rate = (int) [ 1, MAX ], "
    "channels = (int) [ 1, MAX ], "
    "endianness = (int) BYTE_ORDER, " "width = (int) 32")
#endif
     static GstStaticPadTemplate gst_audioresample_sink_template =
         GST_STATIC_PAD_TEMPLATE ("sink",
         GST_PAD_SINK, GST_PAD_ALWAYS, SUPPORTED_CAPS);

     static GstStaticPadTemplate gst_audioresample_src_template =
         GST_STATIC_PAD_TEMPLATE ("src",
         GST_PAD_SRC, GST_PAD_ALWAYS, SUPPORTED_CAPS);

     static void gst_audioresample_dispose (GObject * object);

     static void gst_audioresample_set_property (GObject * object,
         guint prop_id, const GValue * value, GParamSpec * pspec);
     static void gst_audioresample_get_property (GObject * object,
         guint prop_id, GValue * value, GParamSpec * pspec);

92 93 94 95 96 97 98 99 100 101
/* vmethods */
     gboolean audioresample_get_unit_size (GstBaseTransform * base,
         GstCaps * caps, guint * size);
     GstCaps *audioresample_transform_caps (GstBaseTransform * base,
         GstPadDirection direction, GstCaps * caps);
     gboolean audioresample_transform_size (GstBaseTransform * trans,
         GstPadDirection direction, GstCaps * incaps, guint insize,
         GstCaps * outcaps, guint * outsize);
     gboolean audioresample_set_caps (GstBaseTransform * base, GstCaps * incaps,
         GstCaps * outcaps);
102 103
     static GstFlowReturn audioresample_pushthrough (GstAudioresample *
         audioresample);
104 105
     static GstFlowReturn audioresample_transform (GstBaseTransform * base,
         GstBuffer * inbuf, GstBuffer * outbuf);
106 107
     static gboolean audioresample_event (GstBaseTransform * base,
         GstEvent * event);
108 109 110

/*static guint gst_audioresample_signals[LAST_SIGNAL] = { 0 }; */

111 112
#define DEBUG_INIT(bla) \
  GST_DEBUG_CATEGORY_INIT (audioresample_debug, "audioresample", 0, "audio resampling element");
113

114 115
GST_BOILERPLATE_FULL (GstAudioresample, gst_audioresample, GstBaseTransform,
    GST_TYPE_BASE_TRANSFORM, DEBUG_INIT);
116

117 118 119
     static void gst_audioresample_base_init (gpointer g_class)
     {
       GstElementClass *gstelement_class = GST_ELEMENT_CLASS (g_class);
120

121 122 123 124
       gst_element_class_add_pad_template (gstelement_class,
           gst_static_pad_template_get (&gst_audioresample_src_template));
       gst_element_class_add_pad_template (gstelement_class,
           gst_static_pad_template_get (&gst_audioresample_sink_template));
125

126 127 128 129 130
       gst_element_class_set_details (gstelement_class,
           &gst_audioresample_details);
     }

static void gst_audioresample_class_init (GstAudioresampleClass * klass)
131 132 133 134 135 136 137 138 139 140 141
{
  GObjectClass *gobject_class;

  gobject_class = (GObjectClass *) klass;

  gobject_class->set_property = gst_audioresample_set_property;
  gobject_class->get_property = gst_audioresample_get_property;
  gobject_class->dispose = gst_audioresample_dispose;

  g_object_class_install_property (G_OBJECT_CLASS (klass), ARG_FILTERLEN,
      g_param_spec_int ("filter_length", "filter_length", "filter_length",
142 143
          0, G_MAXINT, DEFAULT_FILTERLEN,
          G_PARAM_READWRITE | G_PARAM_CONSTRUCT));
144

145 146 147 148 149 150 151 152 153 154
  GST_BASE_TRANSFORM_CLASS (klass)->transform_size =
      GST_DEBUG_FUNCPTR (audioresample_transform_size);
  GST_BASE_TRANSFORM_CLASS (klass)->get_unit_size =
      GST_DEBUG_FUNCPTR (audioresample_get_unit_size);
  GST_BASE_TRANSFORM_CLASS (klass)->transform_caps =
      GST_DEBUG_FUNCPTR (audioresample_transform_caps);
  GST_BASE_TRANSFORM_CLASS (klass)->set_caps =
      GST_DEBUG_FUNCPTR (audioresample_set_caps);
  GST_BASE_TRANSFORM_CLASS (klass)->transform =
      GST_DEBUG_FUNCPTR (audioresample_transform);
155 156
  GST_BASE_TRANSFORM_CLASS (klass)->event =
      GST_DEBUG_FUNCPTR (audioresample_event);
157 158

  GST_BASE_TRANSFORM_CLASS (klass)->passthrough_on_same_caps = TRUE;
159 160
}

161 162
static void
    gst_audioresample_init (GstAudioresample * audioresample,
163
    GstAudioresampleClass * klass)
164
{
165
  ResampleState *r;
166 167 168 169 170 171 172
  GstBaseTransform *trans;

  trans = GST_BASE_TRANSFORM (audioresample);

  /* buffer alloc passthrough is too impossible. FIXME, it
   * is trivial in the passtrough case. */
  gst_pad_set_bufferalloc_function (trans->sinkpad, NULL);
173

174 175
  r = resample_new ();
  audioresample->resample = r;
176 177 178
  audioresample->ts_offset = -1;
  audioresample->offset = -1;
  audioresample->next_ts = -1;
179

180
  resample_set_filter_length (r, DEFAULT_FILTERLEN);
181 182 183 184 185 186
  resample_set_format (r, RESAMPLE_FORMAT_S16);
}

static void gst_audioresample_dispose (GObject * object)
{
  GstAudioresample *audioresample = GST_AUDIORESAMPLE (object);
187

188 189 190
  if (audioresample->resample) {
    resample_free (audioresample->resample);
    audioresample->resample = NULL;
191
  }
192 193

  G_OBJECT_CLASS (parent_class)->dispose (object);
194 195
}

196 197 198 199 200 201 202
/* vmethods */
gboolean
    audioresample_get_unit_size (GstBaseTransform * base, GstCaps * caps,
    guint * size) {
  gint width, channels;
  GstStructure *structure;
  gboolean ret;
203

204
  g_return_val_if_fail (size, FALSE);
205

206 207 208 209 210
  /* this works for both float and int */
  structure = gst_caps_get_structure (caps, 0);
  ret = gst_structure_get_int (structure, "width", &width);
  ret &= gst_structure_get_int (structure, "channels", &channels);
  g_return_val_if_fail (ret, FALSE);
211

212
  *size = width * channels / 8;
213

214
  return TRUE;
215 216
}

217 218
GstCaps *audioresample_transform_caps (GstBaseTransform * base,
    GstPadDirection direction, GstCaps * caps)
219
{
220
  GstCaps *res;
221 222
  GstStructure *structure;

223 224 225 226 227
  /* transform caps gives one single caps so we can just replace
   * the rate property with our range. */
  res = gst_caps_copy (caps);
  structure = gst_caps_get_structure (res, 0);
  gst_structure_set (structure, "rate", GST_TYPE_INT_RANGE, 1, G_MAXINT, NULL);
228

229
  return res;
230 231
}

232 233 234
static gboolean
    resample_set_state_from_caps (ResampleState * state, GstCaps * incaps,
    GstCaps * outcaps, gint * channels, gint * inrate, gint * outrate)
235 236 237
{
  GstStructure *structure;
  gboolean ret;
238 239
  gint myinrate, myoutrate;
  int mychannels;
240

241 242
  GST_DEBUG ("incaps %" GST_PTR_FORMAT ", outcaps %"
      GST_PTR_FORMAT, incaps, outcaps);
243

244
  structure = gst_caps_get_structure (incaps, 0);
245

246 247 248 249 250 251
  /* FIXME: once it does float, set the correct format */
#if 0
  if (g_str_equal (gst_structure_get_name (structure), "audio/x-raw-float")) {
    r->format = GST_RESAMPLE_FLOAT;
  } else {
    r->format = GST_RESAMPLE_S16;
252
  }
253
#endif
254

255 256 257
  ret = gst_structure_get_int (structure, "rate", &myinrate);
  ret &= gst_structure_get_int (structure, "channels", &mychannels);
  g_return_val_if_fail (ret, FALSE);
258

259 260 261
  structure = gst_caps_get_structure (outcaps, 0);
  ret = gst_structure_get_int (structure, "rate", &myoutrate);
  g_return_val_if_fail (ret, FALSE);
262

263 264 265 266 267 268
  if (channels)
    *channels = mychannels;
  if (inrate)
    *inrate = myinrate;
  if (outrate)
    *outrate = myoutrate;
269

270 271 272
  resample_set_n_channels (state, mychannels);
  resample_set_input_rate (state, myinrate);
  resample_set_output_rate (state, myoutrate);
273

274
  return TRUE;
275 276
}

277 278
gboolean
    audioresample_transform_size (GstBaseTransform * base,
279
    GstPadDirection direction, GstCaps * caps, guint size, GstCaps * othercaps,
280
    guint * othersize) {
281 282 283 284 285 286
  GstAudioresample *audioresample = GST_AUDIORESAMPLE (base);
  ResampleState *state;
  GstCaps *srccaps, *sinkcaps;
  gboolean use_internal = FALSE;        /* whether we use the internal state */
  gboolean ret = TRUE;

287 288
  GST_DEBUG_OBJECT (base, "asked to transform size %d in direction %s",
      size, direction == GST_PAD_SINK ? "SINK" : "SRC");
289 290 291 292 293 294 295
  if (direction == GST_PAD_SINK) {
    sinkcaps = caps;
    srccaps = othercaps;
  } else {
    sinkcaps = othercaps;
    srccaps = caps;
  }
296

297 298 299 300 301 302 303
  /* if the caps are the ones that _set_caps got called with; we can use
   * our own state; otherwise we'll have to create a state */
  if (gst_caps_is_equal (sinkcaps, audioresample->sinkcaps) &&
      gst_caps_is_equal (srccaps, audioresample->srccaps)) {
    use_internal = TRUE;
    state = audioresample->resample;
  } else {
304 305
    GST_DEBUG_OBJECT (audioresample,
        "caps are not the set caps, creating state");
306
    state = resample_new ();
307
    resample_set_filter_length (state, audioresample->filter_length);
308 309
    resample_set_state_from_caps (state, sinkcaps, srccaps, NULL, NULL, NULL);
  }
310

311 312 313 314
  if (direction == GST_PAD_SINK) {
    /* asked to convert size of an incoming buffer */
    *othersize = resample_get_output_size_for_input (state, size);
  } else {
315 316
    /* asked to convert size of an outgoing buffer */
    *othersize = resample_get_input_size_for_output (state, size);
317
  }
318 319
  g_assert (*othersize % state->sample_size == 0);

320 321 322
  /* we make room for one extra sample, given that the resampling filter
   * can output an extra one for non-integral i_rate/o_rate */
  GST_DEBUG_OBJECT (base, "transformed size %d to %d", size, *othersize);
323

324 325 326 327 328
  if (!use_internal) {
    resample_free (state);
  }

  return ret;
329 330
}

331 332 333
gboolean
    audioresample_set_caps (GstBaseTransform * base, GstCaps * incaps,
    GstCaps * outcaps) {
334 335 336 337
  gboolean ret;
  gint inrate, outrate;
  int channels;
  GstAudioresample *audioresample = GST_AUDIORESAMPLE (base);
338

339 340
  GST_DEBUG_OBJECT (base, "incaps %" GST_PTR_FORMAT ", outcaps %"
      GST_PTR_FORMAT, incaps, outcaps);
341

342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362
  ret = resample_set_state_from_caps (audioresample->resample, incaps, outcaps,
      &channels, &inrate, &outrate);

  g_return_val_if_fail (ret, FALSE);

  audioresample->channels = channels;
  GST_DEBUG_OBJECT (audioresample, "set channels to %d", channels);
  audioresample->i_rate = inrate;
  GST_DEBUG_OBJECT (audioresample, "set i_rate to %d", inrate);
  audioresample->o_rate = outrate;
  GST_DEBUG_OBJECT (audioresample, "set o_rate to %d", outrate);

  /* save caps so we can short-circuit in the size_transform if the caps
   * are the same */
  /* FIXME: clean them up in state change ? */
  gst_caps_ref (incaps);
  gst_caps_replace (&audioresample->sinkcaps, incaps);
  gst_caps_ref (outcaps);
  gst_caps_replace (&audioresample->srccaps, outcaps);

  return TRUE;
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
static gboolean audioresample_event (GstBaseTransform * base, GstEvent * event)
{
  GstAudioresample *audioresample;

  audioresample = GST_AUDIORESAMPLE (base);

  switch (GST_EVENT_TYPE (event)) {
    case GST_EVENT_FLUSH_START:
      break;
    case GST_EVENT_FLUSH_STOP:
      resample_input_flush (audioresample->resample);
      audioresample->ts_offset = -1;
      audioresample->next_ts = -1;
      audioresample->offset = -1;
      break;
    case GST_EVENT_NEWSEGMENT:
      resample_input_pushthrough (audioresample->resample);
      audioresample_pushthrough (audioresample);
      audioresample->ts_offset = -1;
      audioresample->next_ts = -1;
      audioresample->offset = -1;
      break;
    case GST_EVENT_EOS:
      resample_input_eos (audioresample->resample);
      audioresample_pushthrough (audioresample);
      break;
    default:
      break;
  }
  parent_class->event (base, event);

  return TRUE;
}

399
static GstFlowReturn
400
    audioresample_do_output (GstAudioresample * audioresample,
401
    GstBuffer * outbuf)
402 403
{
  int outsize;
404
  int outsamples;
405
  ResampleState *r;
406 407 408 409

  r = audioresample->resample;

  outsize = resample_get_output_size (r);
410 411 412 413 414
  GST_DEBUG_OBJECT (audioresample, "audioresample can give me %d bytes",
      outsize);

  /* protect against mem corruption */
  if (outsize > GST_BUFFER_SIZE (outbuf)) {
415 416 417 418
    GST_WARNING_OBJECT (audioresample,
        "overriding audioresample's outsize %d with outbuffer's size %d",
        outsize, GST_BUFFER_SIZE (outbuf));
    outsize = GST_BUFFER_SIZE (outbuf);
419
  }
420 421 422 423 424 425
  /* catch possibly wrong size differences */
  if (GST_BUFFER_SIZE (outbuf) - outsize > r->sample_size) {
    GST_WARNING_OBJECT (audioresample,
        "audioresample's outsize %d too far from outbuffer's size %d",
        outsize, GST_BUFFER_SIZE (outbuf));
  }
426 427

  outsize = resample_get_output_data (r, GST_BUFFER_DATA (outbuf), outsize);
428 429 430 431
  outsamples = outsize / r->sample_size;
  GST_LOG_OBJECT (audioresample, "resample gave me %d bytes or %d samples",
      outsize, outsamples);

432
  GST_BUFFER_OFFSET (outbuf) = audioresample->offset;
433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
  GST_BUFFER_TIMESTAMP (outbuf) = audioresample->next_ts;

  if (audioresample->ts_offset != -1) {
    audioresample->offset += outsamples;
    audioresample->ts_offset += outsamples;
    audioresample->next_ts =
        gst_util_uint64_scale_int (audioresample->ts_offset, GST_SECOND,
        audioresample->o_rate);
    GST_BUFFER_OFFSET_END (outbuf) = audioresample->offset;

    /* we calculate DURATION as the difference between "next" timestamp
     * and current timestamp so we ensure a contiguous stream, instead of
     * having rounding errors. */
    GST_BUFFER_DURATION (outbuf) = audioresample->next_ts -
        GST_BUFFER_TIMESTAMP (outbuf);
  } else {
    /* no valid offset know, we can still sortof calculate the duration though */
    GST_BUFFER_DURATION (outbuf) =
        gst_util_uint64_scale_int (outsamples, GST_SECOND,
        audioresample->o_rate);
  }
454

455 456 457 458 459 460 461 462 463 464 465 466 467
  /* check for possible mem corruption */
  if (outsize > GST_BUFFER_SIZE (outbuf)) {
    /* this is an error that when it happens, would need fixing in the
     * resample library; we told
     * it we wanted only GST_BUFFER_SIZE (outbuf), and it gave us more ! */
    GST_WARNING_OBJECT (audioresample,
        "audioresample, you memory corrupting bastard. "
        "you gave me outsize %d while my buffer was size %d",
        outsize, GST_BUFFER_SIZE (outbuf));
    return GST_FLOW_ERROR;
  }
  /* catch possibly wrong size differences */
  if (GST_BUFFER_SIZE (outbuf) - outsize > r->sample_size) {
468
    GST_WARNING_OBJECT (audioresample,
469
        "audioresample's written outsize %d too far from outbuffer's size %d",
470 471
        outsize, GST_BUFFER_SIZE (outbuf));
  }
472
  GST_BUFFER_SIZE (outbuf) = outsize;
473 474

  return GST_FLOW_OK;
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
static GstFlowReturn
    audioresample_transform (GstBaseTransform * base, GstBuffer * inbuf,
    GstBuffer * outbuf)
{
  GstAudioresample *audioresample;
  ResampleState *r;
  guchar *data;
  gulong size;
  GstClockTime timestamp;

  audioresample = GST_AUDIORESAMPLE (base);
  r = audioresample->resample;

  data = GST_BUFFER_DATA (inbuf);
  size = GST_BUFFER_SIZE (inbuf);
  timestamp = GST_BUFFER_TIMESTAMP (inbuf);

  GST_DEBUG_OBJECT (audioresample, "got buffer of %ld bytes", size);

  if (audioresample->ts_offset == -1) {
    /* if we don't know the initial offset yet, calculate it based on the 
     * input timestamp. */
    if (GST_CLOCK_TIME_IS_VALID (timestamp)) {
      GstClockTime stime;

      /* offset used to calculate the timestamps. We use the sample offset for this
       * to make it more accurate. We want the first buffer to have the same timestamp
       * as the incomming timestamp. */
      audioresample->next_ts = timestamp;
      audioresample->ts_offset =
          gst_util_uint64_scale_int (timestamp, r->o_rate, GST_SECOND);
      /* offset used to set as the buffer offset, this offset is always relative
       * to the stream time, note that timestamp is not... */
      stime = (timestamp - base->segment.start) + base->segment.time;
      audioresample->offset =
          gst_util_uint64_scale_int (stime, r->o_rate, GST_SECOND);
    }
  }

  /* need to memdup, resample takes ownership. */
  resample_add_input_data (r, g_memdup (data, size), size, NULL, NULL);

  return audioresample_do_output (audioresample, outbuf);
}

/* push remaining data in the buffers out */
static GstFlowReturn
    audioresample_pushthrough (GstAudioresample * audioresample)
{
  int outsize;
  ResampleState *r;
  GstBuffer *outbuf;
  GstFlowReturn res = GST_FLOW_OK;
  GstBaseTransform *trans;

  r = audioresample->resample;

  outsize = resample_get_output_size (r);
  if (outsize == 0)
    goto done;

  outbuf = gst_buffer_new_and_alloc (outsize);

  res = audioresample_do_output (audioresample, outbuf);
  if (res != GST_FLOW_OK)
    goto done;

  trans = GST_BASE_TRANSFORM (audioresample);

  res = gst_pad_push (trans->srcpad, outbuf);

done:
  return res;
}


553 554 555 556
static void
    gst_audioresample_set_property (GObject * object, guint prop_id,
    const GValue * value, GParamSpec * pspec)
{
557
  GstAudioresample *audioresample;
558 559 560 561 562 563 564

    g_return_if_fail (GST_IS_AUDIORESAMPLE (object));
    audioresample = GST_AUDIORESAMPLE (object);

  switch (prop_id) {
    case ARG_FILTERLEN:
      audioresample->filter_length = g_value_get_int (value);
565
      GST_DEBUG_OBJECT (GST_ELEMENT (audioresample), "new filter length %d",
566 567 568 569 570 571 572 573 574 575 576 577 578
          audioresample->filter_length);
      resample_set_filter_length (audioresample->resample,
          audioresample->filter_length);
      break;
      default:G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
      break;
  }
}

static void
    gst_audioresample_get_property (GObject * object, guint prop_id,
    GValue * value, GParamSpec * pspec)
{
579
  GstAudioresample *audioresample;
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

  g_return_if_fail (GST_IS_AUDIORESAMPLE (object));
  audioresample = GST_AUDIORESAMPLE (object);

  switch (prop_id) {
    case ARG_FILTERLEN:
      g_value_set_int (value, audioresample->filter_length);
      break;
    default:
      G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
      break;
  }
}


static gboolean plugin_init (GstPlugin * plugin)
{
  resample_init ();

  if (!gst_element_register (plugin, "audioresample", GST_RANK_PRIMARY,
          GST_TYPE_AUDIORESAMPLE)) {
    return FALSE;
  }

  return TRUE;
}

GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
    GST_VERSION_MINOR,
    "audioresample",
610 611
    "Resamples audio", plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME,
    GST_PACKAGE_ORIGIN);