1 /* GStreamer AAC encoder plugin
2 * Copyright (C) 2011 Kan Hu <kan.hu@linaro.org>
4 * This library is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Library General Public
6 * License as published by the Free Software Foundation; either
7 * version 2 of the License, or (at your option) any later version.
9 * This library is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Library General Public License for more details.
14 * You should have received a copy of the GNU Library General Public
15 * License along with this library; if not, write to the
16 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
17 * Boston, MA 02111-1307, USA.
21 * SECTION:element-voaacenc
23 * AAC audio encoder based on vo-aacenc library
24 * <ulink url="http://sourceforge.net/projects/opencore-amr/files/vo-aacenc/">vo-aacenc library source file</ulink>.
27 * <title>Example launch line</title>
29 * gst-launch filesrc location=abc.wav ! wavparse ! audioresample ! audioconvert ! voaacenc ! filesink location=abc.aac
40 #include <gst/audio/multichannel.h>
41 #include <gst/pbutils/codec-utils.h>
43 #include "gstvoaacenc.h"
45 #define VOAAC_ENC_DEFAULT_BITRATE (128000)
46 #define VOAAC_ENC_DEFAULT_OUTPUTFORMAT (0) /* RAW */
47 #define VOAAC_ENC_MPEGVERSION (4)
48 #define VOAAC_ENC_CODECDATA_LEN (2)
49 #define VOAAC_ENC_BITS_PER_SAMPLE (16)
57 #define SAMPLE_RATES " 8000, " \
70 static GstStaticPadTemplate sink_template = GST_STATIC_PAD_TEMPLATE ("sink",
73 GST_STATIC_CAPS ("audio/x-raw-int, "
76 "signed = (boolean) TRUE, "
77 "endianness = (int) BYTE_ORDER, "
78 "rate = (int) { " SAMPLE_RATES " }, " "channels = (int) [1, 2]")
81 static GstStaticPadTemplate src_template = GST_STATIC_PAD_TEMPLATE ("src",
84 GST_STATIC_CAPS ("audio/mpeg, "
85 "mpegversion = (int) 4, "
86 "rate = (int) { " SAMPLE_RATES " }, "
87 "channels = (int) [1, 2], "
88 "stream-format = (string) { adts, raw }, " "base-profile = (string) lc")
91 GST_DEBUG_CATEGORY_STATIC (gst_voaacenc_debug);
92 #define GST_CAT_DEFAULT gst_voaacenc_debug
94 static gboolean voaacenc_core_init (GstVoAacEnc * voaacenc);
95 static gboolean voaacenc_core_set_parameter (GstVoAacEnc * voaacenc);
96 static void voaacenc_core_uninit (GstVoAacEnc * voaacenc);
98 static gboolean gst_voaacenc_start (GstAudioEncoder * enc);
99 static gboolean gst_voaacenc_stop (GstAudioEncoder * enc);
100 static gboolean gst_voaacenc_set_format (GstAudioEncoder * enc,
101 GstAudioInfo * info);
102 static GstFlowReturn gst_voaacenc_handle_frame (GstAudioEncoder * enc,
104 static GstCaps *gst_voaacenc_getcaps (GstAudioEncoder * enc);
106 GST_BOILERPLATE (GstVoAacEnc, gst_voaacenc, GstAudioEncoder,
107 GST_TYPE_AUDIO_ENCODER);
110 gst_voaacenc_set_property (GObject * object, guint prop_id,
111 const GValue * value, GParamSpec * pspec)
113 GstVoAacEnc *self = GST_VOAACENC (object);
117 self->bitrate = g_value_get_int (value);
120 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
127 gst_voaacenc_get_property (GObject * object, guint prop_id,
128 GValue * value, GParamSpec * pspec)
130 GstVoAacEnc *self = GST_VOAACENC (object);
134 g_value_set_int (value, self->bitrate);
137 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
144 gst_voaacenc_base_init (gpointer klass)
146 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
148 gst_element_class_add_pad_template (element_class,
149 gst_static_pad_template_get (&sink_template));
150 gst_element_class_add_pad_template (element_class,
151 gst_static_pad_template_get (&src_template));
153 gst_element_class_set_details_simple (element_class, "AAC audio encoder",
154 "Codec/Encoder/Audio", "AAC audio encoder", "Kan Hu <kan.hu@linaro.org>");
158 gst_voaacenc_class_init (GstVoAacEncClass * klass)
160 GObjectClass *object_class = G_OBJECT_CLASS (klass);
161 GstAudioEncoderClass *base_class = GST_AUDIO_ENCODER_CLASS (klass);
163 object_class->set_property = GST_DEBUG_FUNCPTR (gst_voaacenc_set_property);
164 object_class->get_property = GST_DEBUG_FUNCPTR (gst_voaacenc_get_property);
166 base_class->start = GST_DEBUG_FUNCPTR (gst_voaacenc_start);
167 base_class->stop = GST_DEBUG_FUNCPTR (gst_voaacenc_stop);
168 base_class->set_format = GST_DEBUG_FUNCPTR (gst_voaacenc_set_format);
169 base_class->handle_frame = GST_DEBUG_FUNCPTR (gst_voaacenc_handle_frame);
170 base_class->getcaps = GST_DEBUG_FUNCPTR (gst_voaacenc_getcaps);
172 g_object_class_install_property (object_class, PROP_BITRATE,
173 g_param_spec_int ("bitrate",
175 "Target Audio Bitrate",
176 0, G_MAXINT, VOAAC_ENC_DEFAULT_BITRATE,
177 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
179 GST_DEBUG_CATEGORY_INIT (gst_voaacenc_debug, "voaacenc", 0, "voaac encoder");
183 gst_voaacenc_init (GstVoAacEnc * voaacenc, GstVoAacEncClass * klass)
185 voaacenc->bitrate = VOAAC_ENC_DEFAULT_BITRATE;
186 voaacenc->output_format = VOAAC_ENC_DEFAULT_OUTPUTFORMAT;
189 voaacenc->handle = NULL;
193 gst_voaacenc_start (GstAudioEncoder * enc)
195 GstVoAacEnc *voaacenc = GST_VOAACENC (enc);
197 GST_DEBUG_OBJECT (enc, "start");
199 if (voaacenc_core_init (voaacenc) == FALSE)
203 voaacenc->channels = 0;
209 gst_voaacenc_stop (GstAudioEncoder * enc)
211 GstVoAacEnc *voaacenc = GST_VOAACENC (enc);
213 GST_DEBUG_OBJECT (enc, "stop");
214 voaacenc_core_uninit (voaacenc);
220 gst_voaacenc_generate_sink_caps (gpointer data)
222 #define VOAAC_ENC_MAX_CHANNELS 6
223 /* describe the channels position */
224 static const GstAudioChannelPosition
225 gst_voaacenc_channel_position[][VOAAC_ENC_MAX_CHANNELS] = {
227 GST_AUDIO_CHANNEL_POSITION_FRONT_MONO},
228 { /* 2 ch: front left + front right (front stereo) */
229 GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
230 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT},
231 { /* 3 ch: front center + front stereo */
232 GST_AUDIO_CHANNEL_POSITION_FRONT_CENTER,
233 GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
234 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT},
235 { /* 4 ch: front center + front stereo + back center */
236 GST_AUDIO_CHANNEL_POSITION_FRONT_CENTER,
237 GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
238 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT,
239 GST_AUDIO_CHANNEL_POSITION_REAR_CENTER},
240 { /* 5 ch: front center + front stereo + back stereo */
241 GST_AUDIO_CHANNEL_POSITION_FRONT_CENTER,
242 GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
243 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT,
244 GST_AUDIO_CHANNEL_POSITION_REAR_LEFT,
245 GST_AUDIO_CHANNEL_POSITION_REAR_RIGHT},
246 { /* 6ch: front center + front stereo + back stereo + LFE */
247 GST_AUDIO_CHANNEL_POSITION_FRONT_CENTER,
248 GST_AUDIO_CHANNEL_POSITION_FRONT_LEFT,
249 GST_AUDIO_CHANNEL_POSITION_FRONT_RIGHT,
250 GST_AUDIO_CHANNEL_POSITION_REAR_LEFT,
251 GST_AUDIO_CHANNEL_POSITION_REAR_RIGHT,
252 GST_AUDIO_CHANNEL_POSITION_LFE}
254 GstCaps *caps = gst_caps_new_empty ();
256 static const int rates[] = {
257 8000, 11025, 12000, 16000, 22050, 24000,
258 32000, 44100, 48000, 64000, 88200, 96000
260 GValue rates_arr = { 0, };
263 g_value_init (&rates_arr, GST_TYPE_LIST);
264 g_value_init (&tmp, G_TYPE_INT);
265 for (i = 0; i < G_N_ELEMENTS (rates); i++) {
266 g_value_set_int (&tmp, rates[i]);
267 gst_value_list_append_value (&rates_arr, &tmp);
269 g_value_unset (&tmp);
271 for (i = 0; i < 2 /*VOAAC_ENC_MAX_CHANNELS */ ; i++) {
272 GValue chanpos = { 0 };
274 GstStructure *structure;
276 g_value_init (&chanpos, GST_TYPE_ARRAY);
277 g_value_init (&pos, GST_TYPE_AUDIO_CHANNEL_POSITION);
279 for (c = 0; c <= i; c++) {
280 g_value_set_enum (&pos, gst_voaacenc_channel_position[i][c]);
281 gst_value_array_append_value (&chanpos, &pos);
284 g_value_unset (&pos);
286 structure = gst_structure_new ("audio/x-raw-int",
287 "width", G_TYPE_INT, 16,
288 "depth", G_TYPE_INT, 16,
289 "signed", G_TYPE_BOOLEAN, TRUE,
290 "endianness", G_TYPE_INT, G_BYTE_ORDER,
291 "channels", G_TYPE_INT, i + 1, NULL);
293 gst_structure_set_value (structure, "rate", &rates_arr);
294 gst_structure_set_value (structure, "channel-positions", &chanpos);
295 g_value_unset (&chanpos);
297 gst_caps_append_structure (caps, structure);
300 g_value_unset (&rates_arr);
302 GST_DEBUG ("generated sink caps: %" GST_PTR_FORMAT, caps);
307 gst_voaacenc_get_sink_caps (void)
309 static GOnce g_once = G_ONCE_INIT;
312 g_once (&g_once, gst_voaacenc_generate_sink_caps, NULL);
313 caps = g_once.retval;
319 gst_voaacenc_getcaps (GstAudioEncoder * benc)
321 return gst_audio_encoder_proxy_getcaps (benc, gst_voaacenc_get_sink_caps ());
324 /* check downstream caps to configure format */
326 gst_voaacenc_negotiate (GstVoAacEnc * voaacenc)
330 caps = gst_pad_get_allowed_caps (GST_AUDIO_ENCODER_SRC_PAD (voaacenc));
332 GST_DEBUG_OBJECT (voaacenc, "allowed caps: %" GST_PTR_FORMAT, caps);
334 if (caps && gst_caps_get_size (caps) > 0) {
335 GstStructure *s = gst_caps_get_structure (caps, 0);
336 const gchar *str = NULL;
338 if ((str = gst_structure_get_string (s, "stream-format"))) {
339 if (strcmp (str, "adts") == 0) {
340 GST_DEBUG_OBJECT (voaacenc, "use ADTS format for output");
341 voaacenc->output_format = 1;
342 } else if (strcmp (str, "raw") == 0) {
343 GST_DEBUG_OBJECT (voaacenc, "use RAW format for output");
344 voaacenc->output_format = 0;
346 GST_DEBUG_OBJECT (voaacenc, "unknown stream-format: %s", str);
347 voaacenc->output_format = VOAAC_ENC_DEFAULT_OUTPUTFORMAT;
353 gst_caps_unref (caps);
357 gst_voaacenc_get_rate_index (gint rate)
359 static const gint rate_table[] = {
360 96000, 88200, 64000, 48000, 44100, 32000,
361 24000, 22050, 16000, 12000, 11025, 8000
364 for (i = 0; i < G_N_ELEMENTS (rate_table); ++i) {
365 if (rate == rate_table[i]) {
373 gst_voaacenc_create_source_pad_caps (GstVoAacEnc * voaacenc)
375 GstCaps *caps = NULL;
376 GstBuffer *codec_data;
378 guint8 data[VOAAC_ENC_CODECDATA_LEN];
380 if ((index = gst_voaacenc_get_rate_index (voaacenc->rate)) >= 0) {
381 /* LC profile only */
382 data[0] = ((0x02 << 3) | (index >> 1));
383 data[1] = ((index & 0x01) << 7) | (voaacenc->channels << 3);
385 caps = gst_caps_new_simple ("audio/mpeg",
386 "mpegversion", G_TYPE_INT, VOAAC_ENC_MPEGVERSION,
387 "channels", G_TYPE_INT, voaacenc->channels,
388 "rate", G_TYPE_INT, voaacenc->rate,
389 "stream-format", G_TYPE_STRING,
390 (voaacenc->output_format ? "adts" : "raw")
393 gst_codec_utils_aac_caps_set_level_and_profile (caps, data, sizeof (data));
395 if (!voaacenc->output_format) {
396 codec_data = gst_buffer_new_and_alloc (VOAAC_ENC_CODECDATA_LEN);
398 memcpy (GST_BUFFER_DATA (codec_data), data, sizeof (data));
400 gst_caps_set_simple (caps, "codec_data", GST_TYPE_BUFFER, codec_data,
403 gst_buffer_unref (codec_data);
411 gst_voaacenc_set_format (GstAudioEncoder * benc, GstAudioInfo * info)
413 gboolean ret = FALSE;
414 GstVoAacEnc *voaacenc;
417 voaacenc = GST_VOAACENC (benc);
419 /* get channel count */
420 voaacenc->channels = GST_AUDIO_INFO_CHANNELS (info);
421 voaacenc->rate = GST_AUDIO_INFO_RATE (info);
423 /* precalc buffer size as it's constant now */
424 voaacenc->inbuf_size = voaacenc->channels * 2 * 1024;
426 gst_voaacenc_negotiate (voaacenc);
428 /* create reverse caps */
429 src_caps = gst_voaacenc_create_source_pad_caps (voaacenc);
432 gst_pad_set_caps (GST_AUDIO_ENCODER_SRC_PAD (voaacenc), src_caps);
433 gst_caps_unref (src_caps);
434 ret = voaacenc_core_set_parameter (voaacenc);
437 /* report needs to base class */
438 gst_audio_encoder_set_frame_samples_min (benc, 1024);
439 gst_audio_encoder_set_frame_samples_max (benc, 1024);
440 gst_audio_encoder_set_frame_max (benc, 1);
446 gst_voaacenc_handle_frame (GstAudioEncoder * benc, GstBuffer * buf)
448 GstVoAacEnc *voaacenc;
449 GstFlowReturn ret = GST_FLOW_OK;
451 VO_AUDIO_OUTPUTINFO output_info = { {0} };
452 VO_CODECBUFFER input = { 0 };
453 VO_CODECBUFFER output = { 0 };
455 voaacenc = GST_VOAACENC (benc);
457 g_return_val_if_fail (voaacenc->handle, GST_FLOW_NOT_NEGOTIATED);
459 if (voaacenc->rate == 0 || voaacenc->channels == 0)
462 /* we don't deal with squeezing remnants, so simply discard those */
463 if (G_UNLIKELY (buf == NULL)) {
464 GST_DEBUG_OBJECT (benc, "no data");
468 if (G_UNLIKELY (GST_BUFFER_SIZE (buf) < voaacenc->inbuf_size)) {
469 GST_DEBUG_OBJECT (voaacenc, "discarding trailing data %d",
470 buf ? GST_BUFFER_SIZE (buf) : 0);
471 ret = gst_audio_encoder_finish_frame (benc, NULL, -1);
477 gst_pad_alloc_buffer_and_set_caps (GST_AUDIO_ENCODER_SRC_PAD
478 (voaacenc), 0, voaacenc->inbuf_size,
479 GST_PAD_CAPS (GST_AUDIO_ENCODER_SRC_PAD (voaacenc)),
480 &out)) != GST_FLOW_OK) {
484 output.Buffer = GST_BUFFER_DATA (out);
485 output.Length = voaacenc->inbuf_size;
487 g_assert (GST_BUFFER_SIZE (buf) == voaacenc->inbuf_size);
488 input.Buffer = GST_BUFFER_DATA (buf);
489 input.Length = voaacenc->inbuf_size;
490 voaacenc->codec_api.SetInputData (voaacenc->handle, &input);
493 if (voaacenc->codec_api.GetOutputData (voaacenc->handle, &output,
494 &output_info) != VO_ERR_NONE) {
495 gst_buffer_unref (out);
499 GST_LOG_OBJECT (voaacenc, "encoded to %d bytes", output.Length);
500 GST_BUFFER_SIZE (out) = output.Length;
502 GST_LOG_OBJECT (voaacenc, "Pushing out buffer time: %" GST_TIME_FORMAT
503 " duration: %" GST_TIME_FORMAT,
504 GST_TIME_ARGS (GST_BUFFER_TIMESTAMP (out)),
505 GST_TIME_ARGS (GST_BUFFER_DURATION (out)));
507 ret = gst_audio_encoder_finish_frame (benc, out, 1024);
515 GST_ELEMENT_ERROR (voaacenc, STREAM, TYPE_NOT_FOUND,
516 (NULL), ("unknown type"));
517 ret = GST_FLOW_NOT_NEGOTIATED;
522 GST_ELEMENT_ERROR (voaacenc, STREAM, ENCODE, (NULL), ("encode failed"));
523 ret = GST_FLOW_ERROR;
529 voaacenc_core_mem_alloc (VO_S32 uID, VO_MEM_INFO * pMemInfo)
532 return VO_ERR_INVALID_ARG;
534 pMemInfo->VBuffer = g_malloc (pMemInfo->Size);
539 voaacenc_core_mem_free (VO_S32 uID, VO_PTR pMem)
546 voaacenc_core_mem_set (VO_S32 uID, VO_PTR pBuff, VO_U8 uValue, VO_U32 uSize)
548 memset (pBuff, uValue, uSize);
553 voaacenc_core_mem_copy (VO_S32 uID, VO_PTR pDest, VO_PTR pSource, VO_U32 uSize)
555 memcpy (pDest, pSource, uSize);
560 voaacenc_core_mem_check (VO_S32 uID, VO_PTR pBuffer, VO_U32 uSize)
566 voaacenc_core_init (GstVoAacEnc * voaacenc)
568 VO_CODEC_INIT_USERDATA user_data = { 0 };
569 voGetAACEncAPI (&voaacenc->codec_api);
571 voaacenc->mem_operator.Alloc = voaacenc_core_mem_alloc;
572 voaacenc->mem_operator.Copy = voaacenc_core_mem_copy;
573 voaacenc->mem_operator.Free = voaacenc_core_mem_free;
574 voaacenc->mem_operator.Set = voaacenc_core_mem_set;
575 voaacenc->mem_operator.Check = voaacenc_core_mem_check;
576 user_data.memflag = VO_IMF_USERMEMOPERATOR;
577 user_data.memData = &voaacenc->mem_operator;
578 voaacenc->codec_api.Init (&voaacenc->handle, VO_AUDIO_CodingAAC, &user_data);
580 if (voaacenc->handle == NULL) {
588 voaacenc_core_set_parameter (GstVoAacEnc * voaacenc)
590 AACENC_PARAM params = { 0 };
593 params.sampleRate = voaacenc->rate;
594 params.bitRate = voaacenc->bitrate;
595 params.nChannels = voaacenc->channels;
596 if (voaacenc->output_format) {
603 voaacenc->codec_api.SetParam (voaacenc->handle, VO_PID_AAC_ENCPARAM,
605 if (ret != VO_ERR_NONE) {
606 GST_ERROR_OBJECT (voaacenc, "Failed to set encoder parameters");
613 voaacenc_core_uninit (GstVoAacEnc * voaacenc)
615 if (voaacenc->handle) {
616 voaacenc->codec_api.Uninit (voaacenc->handle);
617 voaacenc->handle = NULL;