3 * Copyright (c) 2001, 2002 Fabrice Bellard
6 * Copyright (c) 2012 Paul B Mahol
8 * WAV muxer RF64 support
9 * Copyright (c) 2013 Daniel Verkamp <daniel@drv.nu>
11 * EBU Tech 3285 - Supplement 3 - Peak Envelope Chunk encoder
12 * Copyright (c) 2014 Georg Lippitsch <georg.lippitsch@gmx.at>
14 * This file is part of FFmpeg.
16 * FFmpeg is free software; you can redistribute it and/or
17 * modify it under the terms of the GNU Lesser General Public
18 * License as published by the Free Software Foundation; either
19 * version 2.1 of the License, or (at your option) any later version.
21 * FFmpeg is distributed in the hope that it will be useful,
22 * but WITHOUT ANY WARRANTY; without even the implied warranty of
23 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
24 * Lesser General Public License for more details.
26 * You should have received a copy of the GNU Lesser General Public
27 * License along with FFmpeg; if not, write to the Free Software
28 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
34 #include "libavutil/avstring.h"
35 #include "libavutil/dict.h"
36 #include "libavutil/common.h"
37 #include "libavutil/intreadwrite.h"
38 #include "libavutil/mathematics.h"
39 #include "libavutil/opt.h"
40 #include "libavutil/time.h"
44 #include "avio_internal.h"
48 #define RF64_AUTO (-1)
52 #define PEAK_BUFFER_SIZE 1024
61 PEAK_FORMAT_UINT8 = 1,
65 typedef struct WAVMuxContext {
72 int16_t *peak_maxpos, *peak_maxneg;
73 uint32_t peak_num_frames;
74 uint32_t peak_outbuf_size;
75 uint32_t peak_outbuf_bytes;
76 uint32_t peak_pos_pop;
84 PeakFormat peak_format;
91 static inline void bwf_write_bext_string(AVFormatContext *s, const char *key, int maxlen)
93 AVDictionaryEntry *tag;
96 if (tag = av_dict_get(s->metadata, key, NULL, 0)) {
97 len = strlen(tag->value);
98 len = FFMIN(len, maxlen);
99 avio_write(s->pb, tag->value, len);
102 ffio_fill(s->pb, 0, maxlen - len);
105 static void bwf_write_bext_chunk(AVFormatContext *s)
107 AVDictionaryEntry *tmp_tag;
108 uint64_t time_reference = 0;
109 int64_t bext = ff_start_tag(s->pb, "bext");
111 bwf_write_bext_string(s, "description", 256);
112 bwf_write_bext_string(s, "originator", 32);
113 bwf_write_bext_string(s, "originator_reference", 32);
114 bwf_write_bext_string(s, "origination_date", 10);
115 bwf_write_bext_string(s, "origination_time", 8);
117 if (tmp_tag = av_dict_get(s->metadata, "time_reference", NULL, 0))
118 time_reference = strtoll(tmp_tag->value, NULL, 10);
119 avio_wl64(s->pb, time_reference);
120 avio_wl16(s->pb, 1); // set version to 1
122 if (tmp_tag = av_dict_get(s->metadata, "umid", NULL, 0)) {
123 unsigned char umidpart_str[17] = {0};
126 int len = strlen(tmp_tag->value+2);
128 for (i = 0; i < len/16; i++) {
129 memcpy(umidpart_str, tmp_tag->value + 2 + (i*16), 16);
130 umidpart = strtoll(umidpart_str, NULL, 16);
131 avio_wb64(s->pb, umidpart);
133 ffio_fill(s->pb, 0, 64 - i*8);
135 ffio_fill(s->pb, 0, 64); // zero UMID
137 ffio_fill(s->pb, 0, 190); // Reserved
139 if (tmp_tag = av_dict_get(s->metadata, "coding_history", NULL, 0))
140 avio_put_str(s->pb, tmp_tag->value);
142 ff_end_tag(s->pb, bext);
145 static av_cold void peak_free_buffers(AVFormatContext *s)
147 WAVMuxContext *wav = s->priv_data;
149 av_freep(&wav->peak_maxpos);
150 av_freep(&wav->peak_maxneg);
151 av_freep(&wav->peak_output);
154 static av_cold int peak_init_writer(AVFormatContext *s)
156 WAVMuxContext *wav = s->priv_data;
157 AVCodecContext *enc = s->streams[0]->codec;
159 if (enc->codec_id != AV_CODEC_ID_PCM_S8 &&
160 enc->codec_id != AV_CODEC_ID_PCM_S16LE &&
161 enc->codec_id != AV_CODEC_ID_PCM_U8 &&
162 enc->codec_id != AV_CODEC_ID_PCM_U16LE) {
163 av_log(s, AV_LOG_ERROR, "%s codec not supported for Peak Chunk\n",
164 s->streams[0]->codec->codec ? s->streams[0]->codec->codec->name : "NONE");
168 wav->peak_bps = av_get_bits_per_sample(enc->codec_id) / 8;
170 if (wav->peak_bps == 1 && wav->peak_format == PEAK_FORMAT_UINT16) {
171 av_log(s, AV_LOG_ERROR,
172 "Writing 16 bit peak for 8 bit audio does not make sense\n");
176 wav->peak_maxpos = av_mallocz(enc->channels * sizeof(*wav->peak_maxpos));
177 if (!wav->peak_maxpos)
179 wav->peak_maxneg = av_mallocz(enc->channels * sizeof(*wav->peak_maxneg));
180 if (!wav->peak_maxneg)
183 wav->peak_output = av_malloc(PEAK_BUFFER_SIZE);
184 if (!wav->peak_output)
187 wav->peak_outbuf_size = PEAK_BUFFER_SIZE;
192 av_log(s, AV_LOG_ERROR, "Out of memory\n");
193 peak_free_buffers(s);
194 return AVERROR(ENOMEM);
197 static void peak_write_frame(AVFormatContext *s)
199 WAVMuxContext *wav = s->priv_data;
200 AVCodecContext *enc = s->streams[0]->codec;
204 if (!wav->peak_output)
207 for (c = 0; c < enc->channels; c++) {
208 wav->peak_maxneg[c] = -wav->peak_maxneg[c];
210 if (wav->peak_bps == 2 && wav->peak_format == PEAK_FORMAT_UINT8) {
211 wav->peak_maxpos[c] = wav->peak_maxpos[c] / 256;
212 wav->peak_maxneg[c] = wav->peak_maxneg[c] / 256;
215 if (wav->peak_ppv == 1)
216 wav->peak_maxpos[c] =
217 FFMAX(wav->peak_maxpos[c], wav->peak_maxneg[c]);
219 peak_of_peaks = FFMAX3(wav->peak_maxpos[c], wav->peak_maxneg[c],
221 if (peak_of_peaks > wav->peak_pop)
222 wav->peak_pos_pop = wav->peak_num_frames;
223 wav->peak_pop = peak_of_peaks;
225 if (wav->peak_outbuf_size - wav->peak_outbuf_bytes <
226 wav->peak_format * wav->peak_ppv) {
227 wav->peak_outbuf_size += PEAK_BUFFER_SIZE;
228 wav->peak_output = av_realloc(wav->peak_output,
229 wav->peak_outbuf_size);
230 if (!wav->peak_output) {
231 av_log(s, AV_LOG_ERROR, "No memory for peak data\n");
236 if (wav->peak_format == PEAK_FORMAT_UINT8) {
237 wav->peak_output[wav->peak_outbuf_bytes++] =
239 if (wav->peak_ppv == 2) {
240 wav->peak_output[wav->peak_outbuf_bytes++] =
244 AV_WL16(wav->peak_output + wav->peak_outbuf_bytes,
245 wav->peak_maxpos[c]);
246 wav->peak_outbuf_bytes += 2;
247 if (wav->peak_ppv == 2) {
248 AV_WL16(wav->peak_output + wav->peak_outbuf_bytes,
249 wav->peak_maxneg[c]);
250 wav->peak_outbuf_bytes += 2;
253 wav->peak_maxpos[c] = 0;
254 wav->peak_maxneg[c] = 0;
256 wav->peak_num_frames++;
259 static void peak_write_chunk(AVFormatContext *s)
261 WAVMuxContext *wav = s->priv_data;
262 AVIOContext *pb = s->pb;
263 AVCodecContext *enc = s->streams[0]->codec;
264 int64_t peak = ff_start_tag(s->pb, "levl");
269 /* Peak frame of incomplete block at end */
270 if (wav->peak_block_pos)
273 memset(timestamp, 0, sizeof(timestamp));
274 if (!(s->flags & AVFMT_FLAG_BITEXACT)) {
275 av_log(s, AV_LOG_INFO, "Writing local time and date to Peak Envelope Chunk\n");
277 now_secs = now0 / 1000000;
278 strftime(timestamp, sizeof(timestamp), "%Y:%m:%d:%H:%M:%S:", localtime(&now_secs));
279 av_strlcatf(timestamp, sizeof(timestamp), "%03d", (int)((now0 / 1000) % 1000));
282 avio_wl32(pb, 1); /* version */
283 avio_wl32(pb, wav->peak_format); /* 8 or 16 bit */
284 avio_wl32(pb, wav->peak_ppv); /* positive and negative */
285 avio_wl32(pb, wav->peak_block_size); /* frames per value */
286 avio_wl32(pb, enc->channels); /* number of channels */
287 avio_wl32(pb, wav->peak_num_frames); /* number of peak frames */
288 avio_wl32(pb, wav->peak_pos_pop); /* audio sample frame index */
289 avio_wl32(pb, 128); /* equal to size of header */
290 avio_write(pb, timestamp, 28); /* ASCII time stamp */
291 ffio_fill(pb, 0, 60);
293 avio_write(pb, wav->peak_output, wav->peak_outbuf_bytes);
295 ff_end_tag(pb, peak);
301 static int wav_write_header(AVFormatContext *s)
303 WAVMuxContext *wav = s->priv_data;
304 AVIOContext *pb = s->pb;
307 if (s->nb_streams != 1) {
308 av_log(s, AV_LOG_ERROR, "WAVE files have exactly one stream\n");
309 return AVERROR(EINVAL);
312 if (wav->rf64 == RF64_ALWAYS) {
313 ffio_wfourcc(pb, "RF64");
314 avio_wl32(pb, -1); /* RF64 chunk size: use size in ds64 */
316 ffio_wfourcc(pb, "RIFF");
317 avio_wl32(pb, -1); /* file length */
320 ffio_wfourcc(pb, "WAVE");
322 if (wav->rf64 != RF64_NEVER) {
323 /* write empty ds64 chunk or JUNK chunk to reserve space for ds64 */
324 ffio_wfourcc(pb, wav->rf64 == RF64_ALWAYS ? "ds64" : "JUNK");
325 avio_wl32(pb, 28); /* chunk size */
326 wav->ds64 = avio_tell(pb);
327 ffio_fill(pb, 0, 28);
330 if (wav->write_peak != 2) {
332 fmt = ff_start_tag(pb, "fmt ");
333 if (ff_put_wav_header(pb, s->streams[0]->codec, 0) < 0) {
334 const AVCodecDescriptor *desc = avcodec_descriptor_get(s->streams[0]->codec->codec_id);
335 av_log(s, AV_LOG_ERROR, "%s codec not supported in WAVE format\n",
336 desc ? desc->name : "unknown");
337 return AVERROR(ENOSYS);
342 if (s->streams[0]->codec->codec_tag != 0x01 /* hence for all other than PCM */
343 && s->pb->seekable) {
344 wav->fact_pos = ff_start_tag(pb, "fact");
346 ff_end_tag(pb, wav->fact_pos);
350 bwf_write_bext_chunk(s);
352 if (wav->write_peak) {
354 if ((ret = peak_init_writer(s)) < 0)
358 avpriv_set_pts_info(s->streams[0], 64, 1, s->streams[0]->codec->sample_rate);
359 wav->maxpts = wav->last_duration = 0;
360 wav->minpts = INT64_MAX;
362 if (wav->write_peak != 2) {
364 ff_riff_write_info(s);
367 wav->data = ff_start_tag(pb, "data");
375 static int wav_write_packet(AVFormatContext *s, AVPacket *pkt)
377 AVIOContext *pb = s->pb;
378 WAVMuxContext *wav = s->priv_data;
380 if (wav->write_peak != 2)
381 avio_write(pb, pkt->data, pkt->size);
383 if (wav->write_peak) {
386 for (i = 0; i < pkt->size; i += wav->peak_bps) {
387 if (wav->peak_bps == 1) {
388 wav->peak_maxpos[c] = FFMAX(wav->peak_maxpos[c], *(int8_t*)(pkt->data + i));
389 wav->peak_maxneg[c] = FFMIN(wav->peak_maxneg[c], *(int8_t*)(pkt->data + i));
391 wav->peak_maxpos[c] = FFMAX(wav->peak_maxpos[c], (int16_t)AV_RL16(pkt->data + i));
392 wav->peak_maxneg[c] = FFMIN(wav->peak_maxneg[c], (int16_t)AV_RL16(pkt->data + i));
394 if (++c == s->streams[0]->codec->channels) {
396 if (++wav->peak_block_pos == wav->peak_block_size) {
398 wav->peak_block_pos = 0;
404 if(pkt->pts != AV_NOPTS_VALUE) {
405 wav->minpts = FFMIN(wav->minpts, pkt->pts);
406 wav->maxpts = FFMAX(wav->maxpts, pkt->pts);
407 wav->last_duration = pkt->duration;
409 av_log(s, AV_LOG_ERROR, "wav_write_packet: NOPTS\n");
413 static int wav_write_trailer(AVFormatContext *s)
415 AVIOContext *pb = s->pb;
416 WAVMuxContext *wav = s->priv_data;
417 int64_t file_size, data_size;
418 int64_t number_of_samples = 0;
423 if (s->pb->seekable) {
424 if (wav->write_peak != 2) {
425 ff_end_tag(pb, wav->data);
429 if (wav->write_peak && wav->peak_output) {
434 /* update file size */
435 file_size = avio_tell(pb);
436 data_size = file_size - wav->data;
437 if (wav->rf64 == RF64_ALWAYS || (wav->rf64 == RF64_AUTO && file_size - 8 > UINT32_MAX)) {
440 avio_seek(pb, 4, SEEK_SET);
441 avio_wl32(pb, (uint32_t)(file_size - 8));
442 avio_seek(pb, file_size, SEEK_SET);
447 number_of_samples = av_rescale(wav->maxpts - wav->minpts + wav->last_duration,
448 s->streams[0]->codec->sample_rate * (int64_t)s->streams[0]->time_base.num,
449 s->streams[0]->time_base.den);
451 if(s->streams[0]->codec->codec_tag != 0x01) {
452 /* Update num_samps in fact chunk */
453 avio_seek(pb, wav->fact_pos, SEEK_SET);
454 if (rf64 || (wav->rf64 == RF64_AUTO && number_of_samples > UINT32_MAX)) {
458 avio_wl32(pb, number_of_samples);
459 avio_seek(pb, file_size, SEEK_SET);
465 /* overwrite RIFF with RF64 */
466 avio_seek(pb, 0, SEEK_SET);
467 ffio_wfourcc(pb, "RF64");
470 /* write ds64 chunk (overwrite JUNK if rf64 == RF64_AUTO) */
471 avio_seek(pb, wav->ds64 - 8, SEEK_SET);
472 ffio_wfourcc(pb, "ds64");
473 avio_wl32(pb, 28); /* ds64 chunk size */
474 avio_wl64(pb, file_size - 8); /* RF64 chunk size */
475 avio_wl64(pb, data_size); /* data chunk size */
476 avio_wl64(pb, number_of_samples); /* fact chunk number of samples */
477 avio_wl32(pb, 0); /* number of table entries for non-'data' chunks */
479 /* write -1 in data chunk size */
480 avio_seek(pb, wav->data - 4, SEEK_SET);
483 avio_seek(pb, file_size, SEEK_SET);
489 peak_free_buffers(s);
494 #define OFFSET(x) offsetof(WAVMuxContext, x)
495 #define ENC AV_OPT_FLAG_ENCODING_PARAM
496 static const AVOption options[] = {
497 { "write_bext", "Write BEXT chunk.", OFFSET(write_bext), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, ENC },
498 { "write_peak", "Write Peak Envelope chunk.", OFFSET(write_peak), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 2, ENC, "peak" },
499 { "off", "Do not write peak chunk.", 0, AV_OPT_TYPE_CONST, { .i64 = PEAK_OFF }, 0, 0, ENC, "peak" },
500 { "on", "Append peak chunk after wav data.", 0, AV_OPT_TYPE_CONST, { .i64 = PEAK_ON }, 0, 0, ENC, "peak" },
501 { "only", "Write only peak chunk, omit wav data.", 0, AV_OPT_TYPE_CONST, { .i64 = PEAK_ONLY }, 0, 0, ENC, "peak" },
502 { "rf64", "Use RF64 header rather than RIFF for large files.", OFFSET(rf64), AV_OPT_TYPE_INT, { .i64 = RF64_NEVER },-1, 1, ENC, "rf64" },
503 { "auto", "Write RF64 header if file grows large enough.", 0, AV_OPT_TYPE_CONST, { .i64 = RF64_AUTO }, 0, 0, ENC, "rf64" },
504 { "always", "Always write RF64 header regardless of file size.", 0, AV_OPT_TYPE_CONST, { .i64 = RF64_ALWAYS }, 0, 0, ENC, "rf64" },
505 { "never", "Never write RF64 header regardless of file size.", 0, AV_OPT_TYPE_CONST, { .i64 = RF64_NEVER }, 0, 0, ENC, "rf64" },
506 { "peak_block_size", "Number of audio samples used to generate each peak frame.", OFFSET(peak_block_size), AV_OPT_TYPE_INT, { .i64 = 256 }, 0, 65536, ENC },
507 { "peak_format", "The format of the peak envelope data (1: uint8, 2: uint16).", OFFSET(peak_format), AV_OPT_TYPE_INT, { .i64 = PEAK_FORMAT_UINT16 }, PEAK_FORMAT_UINT8, PEAK_FORMAT_UINT16, ENC },
508 { "peak_ppv", "Number of peak points per peak value (1 or 2).", OFFSET(peak_ppv), AV_OPT_TYPE_INT, { .i64 = 2 }, 1, 2, ENC },
512 static const AVClass wav_muxer_class = {
513 .class_name = "WAV muxer",
514 .item_name = av_default_item_name,
516 .version = LIBAVUTIL_VERSION_INT,
519 AVOutputFormat ff_wav_muxer = {
521 .long_name = NULL_IF_CONFIG_SMALL("WAV / WAVE (Waveform Audio)"),
522 .mime_type = "audio/x-wav",
524 .priv_data_size = sizeof(WAVMuxContext),
525 .audio_codec = AV_CODEC_ID_PCM_S16LE,
526 .video_codec = AV_CODEC_ID_NONE,
527 .write_header = wav_write_header,
528 .write_packet = wav_write_packet,
529 .write_trailer = wav_write_trailer,
530 .flags = AVFMT_TS_NONSTRICT,
531 .codec_tag = (const AVCodecTag* const []){ ff_codec_wav_tags, 0 },
532 .priv_class = &wav_muxer_class,
534 #endif /* CONFIG_WAV_MUXER */
539 static void start_guid(AVIOContext *pb, const uint8_t *guid, int64_t *pos)
541 *pos = avio_tell(pb);
543 avio_write(pb, guid, 16);
544 avio_wl64(pb, INT64_MAX);
547 static void end_guid(AVIOContext *pb, int64_t start)
549 int64_t end, pos = avio_tell(pb);
551 end = FFALIGN(pos, 8);
552 ffio_fill(pb, 0, end - pos);
553 avio_seek(pb, start + 16, SEEK_SET);
554 avio_wl64(pb, end - start);
555 avio_seek(pb, end, SEEK_SET);
558 static int w64_write_header(AVFormatContext *s)
560 WAVMuxContext *wav = s->priv_data;
561 AVIOContext *pb = s->pb;
565 avio_write(pb, ff_w64_guid_riff, sizeof(ff_w64_guid_riff));
567 avio_write(pb, ff_w64_guid_wave, sizeof(ff_w64_guid_wave));
568 start_guid(pb, ff_w64_guid_fmt, &start);
569 if ((ret = ff_put_wav_header(pb, s->streams[0]->codec, 0)) < 0) {
570 av_log(s, AV_LOG_ERROR, "%s codec not supported\n",
571 s->streams[0]->codec->codec ? s->streams[0]->codec->codec->name : "NONE");
576 if (s->streams[0]->codec->codec_tag != 0x01 /* hence for all other than PCM */
577 && s->pb->seekable) {
578 start_guid(pb, ff_w64_guid_fact, &wav->fact_pos);
580 end_guid(pb, wav->fact_pos);
583 start_guid(pb, ff_w64_guid_data, &wav->data);
588 static int w64_write_trailer(AVFormatContext *s)
590 AVIOContext *pb = s->pb;
591 WAVMuxContext *wav = s->priv_data;
595 end_guid(pb, wav->data);
597 file_size = avio_tell(pb);
598 avio_seek(pb, 16, SEEK_SET);
599 avio_wl64(pb, file_size);
601 if (s->streams[0]->codec->codec_tag != 0x01) {
602 int64_t number_of_samples;
604 number_of_samples = av_rescale(wav->maxpts - wav->minpts + wav->last_duration,
605 s->streams[0]->codec->sample_rate * (int64_t)s->streams[0]->time_base.num,
606 s->streams[0]->time_base.den);
607 avio_seek(pb, wav->fact_pos + 24, SEEK_SET);
608 avio_wl64(pb, number_of_samples);
611 avio_seek(pb, file_size, SEEK_SET);
618 AVOutputFormat ff_w64_muxer = {
620 .long_name = NULL_IF_CONFIG_SMALL("Sony Wave64"),
622 .priv_data_size = sizeof(WAVMuxContext),
623 .audio_codec = AV_CODEC_ID_PCM_S16LE,
624 .video_codec = AV_CODEC_ID_NONE,
625 .write_header = w64_write_header,
626 .write_packet = wav_write_packet,
627 .write_trailer = w64_write_trailer,
628 .flags = AVFMT_TS_NONSTRICT,
629 .codec_tag = (const AVCodecTag* const []){ ff_codec_wav_tags, 0 },
631 #endif /* CONFIG_W64_MUXER */