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");
173 return AVERROR(EINVAL
);
176 wav
->peak_maxpos
= av_mallocz_array(enc
->channels
, sizeof(*wav
->peak_maxpos
));
177 wav
->peak_maxneg
= av_mallocz_array(enc
->channels
, sizeof(*wav
->peak_maxneg
));
178 wav
->peak_output
= av_malloc(PEAK_BUFFER_SIZE
);
179 if (!wav
->peak_maxpos
|| !wav
->peak_maxneg
|| !wav
->peak_output
)
182 wav
->peak_outbuf_size
= PEAK_BUFFER_SIZE
;
187 av_log(s
, AV_LOG_ERROR
, "Out of memory\n");
188 peak_free_buffers(s
);
189 return AVERROR(ENOMEM
);
192 static void peak_write_frame(AVFormatContext
*s
)
194 WAVMuxContext
*wav
= s
->priv_data
;
195 AVCodecContext
*enc
= s
->streams
[0]->codec
;
199 if (!wav
->peak_output
)
202 for (c
= 0; c
< enc
->channels
; c
++) {
203 wav
->peak_maxneg
[c
] = -wav
->peak_maxneg
[c
];
205 if (wav
->peak_bps
== 2 && wav
->peak_format
== PEAK_FORMAT_UINT8
) {
206 wav
->peak_maxpos
[c
] = wav
->peak_maxpos
[c
] / 256;
207 wav
->peak_maxneg
[c
] = wav
->peak_maxneg
[c
] / 256;
210 if (wav
->peak_ppv
== 1)
211 wav
->peak_maxpos
[c
] =
212 FFMAX(wav
->peak_maxpos
[c
], wav
->peak_maxneg
[c
]);
214 peak_of_peaks
= FFMAX3(wav
->peak_maxpos
[c
], wav
->peak_maxneg
[c
],
216 if (peak_of_peaks
> wav
->peak_pop
)
217 wav
->peak_pos_pop
= wav
->peak_num_frames
;
218 wav
->peak_pop
= peak_of_peaks
;
220 if (wav
->peak_outbuf_size
- wav
->peak_outbuf_bytes
<
221 wav
->peak_format
* wav
->peak_ppv
) {
222 wav
->peak_outbuf_size
+= PEAK_BUFFER_SIZE
;
223 wav
->peak_output
= av_realloc(wav
->peak_output
,
224 wav
->peak_outbuf_size
);
225 if (!wav
->peak_output
) {
226 av_log(s
, AV_LOG_ERROR
, "No memory for peak data\n");
231 if (wav
->peak_format
== PEAK_FORMAT_UINT8
) {
232 wav
->peak_output
[wav
->peak_outbuf_bytes
++] =
234 if (wav
->peak_ppv
== 2) {
235 wav
->peak_output
[wav
->peak_outbuf_bytes
++] =
239 AV_WL16(wav
->peak_output
+ wav
->peak_outbuf_bytes
,
240 wav
->peak_maxpos
[c
]);
241 wav
->peak_outbuf_bytes
+= 2;
242 if (wav
->peak_ppv
== 2) {
243 AV_WL16(wav
->peak_output
+ wav
->peak_outbuf_bytes
,
244 wav
->peak_maxneg
[c
]);
245 wav
->peak_outbuf_bytes
+= 2;
248 wav
->peak_maxpos
[c
] = 0;
249 wav
->peak_maxneg
[c
] = 0;
251 wav
->peak_num_frames
++;
254 static void peak_write_chunk(AVFormatContext
*s
)
256 WAVMuxContext
*wav
= s
->priv_data
;
257 AVIOContext
*pb
= s
->pb
;
258 AVCodecContext
*enc
= s
->streams
[0]->codec
;
259 int64_t peak
= ff_start_tag(s
->pb
, "levl");
264 /* Peak frame of incomplete block at end */
265 if (wav
->peak_block_pos
)
268 memset(timestamp
, 0, sizeof(timestamp
));
269 if (!(s
->flags
& AVFMT_FLAG_BITEXACT
)) {
270 av_log(s
, AV_LOG_INFO
, "Writing local time and date to Peak Envelope Chunk\n");
272 now_secs
= now0
/ 1000000;
273 strftime(timestamp
, sizeof(timestamp
), "%Y:%m:%d:%H:%M:%S:", localtime(&now_secs
));
274 av_strlcatf(timestamp
, sizeof(timestamp
), "%03d", (int)((now0
/ 1000) % 1000));
277 avio_wl32(pb
, 1); /* version */
278 avio_wl32(pb
, wav
->peak_format
); /* 8 or 16 bit */
279 avio_wl32(pb
, wav
->peak_ppv
); /* positive and negative */
280 avio_wl32(pb
, wav
->peak_block_size
); /* frames per value */
281 avio_wl32(pb
, enc
->channels
); /* number of channels */
282 avio_wl32(pb
, wav
->peak_num_frames
); /* number of peak frames */
283 avio_wl32(pb
, wav
->peak_pos_pop
); /* audio sample frame index */
284 avio_wl32(pb
, 128); /* equal to size of header */
285 avio_write(pb
, timestamp
, 28); /* ASCII time stamp */
286 ffio_fill(pb
, 0, 60);
288 avio_write(pb
, wav
->peak_output
, wav
->peak_outbuf_bytes
);
290 ff_end_tag(pb
, peak
);
296 static int wav_write_header(AVFormatContext
*s
)
298 WAVMuxContext
*wav
= s
->priv_data
;
299 AVIOContext
*pb
= s
->pb
;
302 if (s
->nb_streams
!= 1) {
303 av_log(s
, AV_LOG_ERROR
, "WAVE files have exactly one stream\n");
304 return AVERROR(EINVAL
);
307 if (wav
->rf64
== RF64_ALWAYS
) {
308 ffio_wfourcc(pb
, "RF64");
309 avio_wl32(pb
, -1); /* RF64 chunk size: use size in ds64 */
311 ffio_wfourcc(pb
, "RIFF");
312 avio_wl32(pb
, -1); /* file length */
315 ffio_wfourcc(pb
, "WAVE");
317 if (wav
->rf64
!= RF64_NEVER
) {
318 /* write empty ds64 chunk or JUNK chunk to reserve space for ds64 */
319 ffio_wfourcc(pb
, wav
->rf64
== RF64_ALWAYS
? "ds64" : "JUNK");
320 avio_wl32(pb
, 28); /* chunk size */
321 wav
->ds64
= avio_tell(pb
);
322 ffio_fill(pb
, 0, 28);
325 if (wav
->write_peak
!= 2) {
327 fmt
= ff_start_tag(pb
, "fmt ");
328 if (ff_put_wav_header(pb
, s
->streams
[0]->codec
, 0) < 0) {
329 const AVCodecDescriptor
*desc
= avcodec_descriptor_get(s
->streams
[0]->codec
->codec_id
);
330 av_log(s
, AV_LOG_ERROR
, "%s codec not supported in WAVE format\n",
331 desc
? desc
->name
: "unknown");
332 return AVERROR(ENOSYS
);
337 if (s
->streams
[0]->codec
->codec_tag
!= 0x01 /* hence for all other than PCM */
338 && s
->pb
->seekable
) {
339 wav
->fact_pos
= ff_start_tag(pb
, "fact");
341 ff_end_tag(pb
, wav
->fact_pos
);
345 bwf_write_bext_chunk(s
);
347 if (wav
->write_peak
) {
349 if ((ret
= peak_init_writer(s
)) < 0)
353 avpriv_set_pts_info(s
->streams
[0], 64, 1, s
->streams
[0]->codec
->sample_rate
);
354 wav
->maxpts
= wav
->last_duration
= 0;
355 wav
->minpts
= INT64_MAX
;
357 if (wav
->write_peak
!= 2) {
359 ff_riff_write_info(s
);
362 wav
->data
= ff_start_tag(pb
, "data");
370 static int wav_write_packet(AVFormatContext
*s
, AVPacket
*pkt
)
372 AVIOContext
*pb
= s
->pb
;
373 WAVMuxContext
*wav
= s
->priv_data
;
375 if (wav
->write_peak
!= 2)
376 avio_write(pb
, pkt
->data
, pkt
->size
);
378 if (wav
->write_peak
) {
381 for (i
= 0; i
< pkt
->size
; i
+= wav
->peak_bps
) {
382 if (wav
->peak_bps
== 1) {
383 wav
->peak_maxpos
[c
] = FFMAX(wav
->peak_maxpos
[c
], *(int8_t*)(pkt
->data
+ i
));
384 wav
->peak_maxneg
[c
] = FFMIN(wav
->peak_maxneg
[c
], *(int8_t*)(pkt
->data
+ i
));
386 wav
->peak_maxpos
[c
] = FFMAX(wav
->peak_maxpos
[c
], (int16_t)AV_RL16(pkt
->data
+ i
));
387 wav
->peak_maxneg
[c
] = FFMIN(wav
->peak_maxneg
[c
], (int16_t)AV_RL16(pkt
->data
+ i
));
389 if (++c
== s
->streams
[0]->codec
->channels
) {
391 if (++wav
->peak_block_pos
== wav
->peak_block_size
) {
393 wav
->peak_block_pos
= 0;
399 if(pkt
->pts
!= AV_NOPTS_VALUE
) {
400 wav
->minpts
= FFMIN(wav
->minpts
, pkt
->pts
);
401 wav
->maxpts
= FFMAX(wav
->maxpts
, pkt
->pts
);
402 wav
->last_duration
= pkt
->duration
;
404 av_log(s
, AV_LOG_ERROR
, "wav_write_packet: NOPTS\n");
408 static int wav_write_trailer(AVFormatContext
*s
)
410 AVIOContext
*pb
= s
->pb
;
411 WAVMuxContext
*wav
= s
->priv_data
;
412 int64_t file_size
, data_size
;
413 int64_t number_of_samples
= 0;
418 if (s
->pb
->seekable
) {
419 if (wav
->write_peak
!= 2) {
420 ff_end_tag(pb
, wav
->data
);
424 if (wav
->write_peak
&& wav
->peak_output
) {
429 /* update file size */
430 file_size
= avio_tell(pb
);
431 data_size
= file_size
- wav
->data
;
432 if (wav
->rf64
== RF64_ALWAYS
|| (wav
->rf64
== RF64_AUTO
&& file_size
- 8 > UINT32_MAX
)) {
435 avio_seek(pb
, 4, SEEK_SET
);
436 avio_wl32(pb
, (uint32_t)(file_size
- 8));
437 avio_seek(pb
, file_size
, SEEK_SET
);
442 number_of_samples
= av_rescale(wav
->maxpts
- wav
->minpts
+ wav
->last_duration
,
443 s
->streams
[0]->codec
->sample_rate
* (int64_t)s
->streams
[0]->time_base
.num
,
444 s
->streams
[0]->time_base
.den
);
446 if(s
->streams
[0]->codec
->codec_tag
!= 0x01) {
447 /* Update num_samps in fact chunk */
448 avio_seek(pb
, wav
->fact_pos
, SEEK_SET
);
449 if (rf64
|| (wav
->rf64
== RF64_AUTO
&& number_of_samples
> UINT32_MAX
)) {
453 avio_wl32(pb
, number_of_samples
);
454 avio_seek(pb
, file_size
, SEEK_SET
);
460 /* overwrite RIFF with RF64 */
461 avio_seek(pb
, 0, SEEK_SET
);
462 ffio_wfourcc(pb
, "RF64");
465 /* write ds64 chunk (overwrite JUNK if rf64 == RF64_AUTO) */
466 avio_seek(pb
, wav
->ds64
- 8, SEEK_SET
);
467 ffio_wfourcc(pb
, "ds64");
468 avio_wl32(pb
, 28); /* ds64 chunk size */
469 avio_wl64(pb
, file_size
- 8); /* RF64 chunk size */
470 avio_wl64(pb
, data_size
); /* data chunk size */
471 avio_wl64(pb
, number_of_samples
); /* fact chunk number of samples */
472 avio_wl32(pb
, 0); /* number of table entries for non-'data' chunks */
474 /* write -1 in data chunk size */
475 avio_seek(pb
, wav
->data
- 4, SEEK_SET
);
478 avio_seek(pb
, file_size
, SEEK_SET
);
484 peak_free_buffers(s
);
489 #define OFFSET(x) offsetof(WAVMuxContext, x)
490 #define ENC AV_OPT_FLAG_ENCODING_PARAM
491 static const AVOption options
[] = {
492 { "write_bext", "Write BEXT chunk.", OFFSET(write_bext
), AV_OPT_TYPE_INT
, { .i64
= 0 }, 0, 1, ENC
},
493 { "write_peak", "Write Peak Envelope chunk.", OFFSET(write_peak
), AV_OPT_TYPE_INT
, { .i64
= 0 }, 0, 2, ENC
, "peak" },
494 { "off", "Do not write peak chunk.", 0, AV_OPT_TYPE_CONST
, { .i64
= PEAK_OFF
}, 0, 0, ENC
, "peak" },
495 { "on", "Append peak chunk after wav data.", 0, AV_OPT_TYPE_CONST
, { .i64
= PEAK_ON
}, 0, 0, ENC
, "peak" },
496 { "only", "Write only peak chunk, omit wav data.", 0, AV_OPT_TYPE_CONST
, { .i64
= PEAK_ONLY
}, 0, 0, ENC
, "peak" },
497 { "rf64", "Use RF64 header rather than RIFF for large files.", OFFSET(rf64
), AV_OPT_TYPE_INT
, { .i64
= RF64_NEVER
},-1, 1, ENC
, "rf64" },
498 { "auto", "Write RF64 header if file grows large enough.", 0, AV_OPT_TYPE_CONST
, { .i64
= RF64_AUTO
}, 0, 0, ENC
, "rf64" },
499 { "always", "Always write RF64 header regardless of file size.", 0, AV_OPT_TYPE_CONST
, { .i64
= RF64_ALWAYS
}, 0, 0, ENC
, "rf64" },
500 { "never", "Never write RF64 header regardless of file size.", 0, AV_OPT_TYPE_CONST
, { .i64
= RF64_NEVER
}, 0, 0, ENC
, "rf64" },
501 { "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
},
502 { "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
},
503 { "peak_ppv", "Number of peak points per peak value (1 or 2).", OFFSET(peak_ppv
), AV_OPT_TYPE_INT
, { .i64
= 2 }, 1, 2, ENC
},
507 static const AVClass wav_muxer_class
= {
508 .class_name
= "WAV muxer",
509 .item_name
= av_default_item_name
,
511 .version
= LIBAVUTIL_VERSION_INT
,
514 AVOutputFormat ff_wav_muxer
= {
516 .long_name
= NULL_IF_CONFIG_SMALL("WAV / WAVE (Waveform Audio)"),
517 .mime_type
= "audio/x-wav",
519 .priv_data_size
= sizeof(WAVMuxContext
),
520 .audio_codec
= AV_CODEC_ID_PCM_S16LE
,
521 .video_codec
= AV_CODEC_ID_NONE
,
522 .write_header
= wav_write_header
,
523 .write_packet
= wav_write_packet
,
524 .write_trailer
= wav_write_trailer
,
525 .flags
= AVFMT_TS_NONSTRICT
,
526 .codec_tag
= (const AVCodecTag
* const []){ ff_codec_wav_tags
, 0 },
527 .priv_class
= &wav_muxer_class
,
529 #endif /* CONFIG_WAV_MUXER */
534 static void start_guid(AVIOContext
*pb
, const uint8_t *guid
, int64_t *pos
)
536 *pos
= avio_tell(pb
);
538 avio_write(pb
, guid
, 16);
539 avio_wl64(pb
, INT64_MAX
);
542 static void end_guid(AVIOContext
*pb
, int64_t start
)
544 int64_t end
, pos
= avio_tell(pb
);
546 end
= FFALIGN(pos
, 8);
547 ffio_fill(pb
, 0, end
- pos
);
548 avio_seek(pb
, start
+ 16, SEEK_SET
);
549 avio_wl64(pb
, end
- start
);
550 avio_seek(pb
, end
, SEEK_SET
);
553 static int w64_write_header(AVFormatContext
*s
)
555 WAVMuxContext
*wav
= s
->priv_data
;
556 AVIOContext
*pb
= s
->pb
;
560 avio_write(pb
, ff_w64_guid_riff
, sizeof(ff_w64_guid_riff
));
562 avio_write(pb
, ff_w64_guid_wave
, sizeof(ff_w64_guid_wave
));
563 start_guid(pb
, ff_w64_guid_fmt
, &start
);
564 if ((ret
= ff_put_wav_header(pb
, s
->streams
[0]->codec
, 0)) < 0) {
565 av_log(s
, AV_LOG_ERROR
, "%s codec not supported\n",
566 s
->streams
[0]->codec
->codec
? s
->streams
[0]->codec
->codec
->name
: "NONE");
571 if (s
->streams
[0]->codec
->codec_tag
!= 0x01 /* hence for all other than PCM */
572 && s
->pb
->seekable
) {
573 start_guid(pb
, ff_w64_guid_fact
, &wav
->fact_pos
);
575 end_guid(pb
, wav
->fact_pos
);
578 start_guid(pb
, ff_w64_guid_data
, &wav
->data
);
583 static int w64_write_trailer(AVFormatContext
*s
)
585 AVIOContext
*pb
= s
->pb
;
586 WAVMuxContext
*wav
= s
->priv_data
;
590 end_guid(pb
, wav
->data
);
592 file_size
= avio_tell(pb
);
593 avio_seek(pb
, 16, SEEK_SET
);
594 avio_wl64(pb
, file_size
);
596 if (s
->streams
[0]->codec
->codec_tag
!= 0x01) {
597 int64_t number_of_samples
;
599 number_of_samples
= av_rescale(wav
->maxpts
- wav
->minpts
+ wav
->last_duration
,
600 s
->streams
[0]->codec
->sample_rate
* (int64_t)s
->streams
[0]->time_base
.num
,
601 s
->streams
[0]->time_base
.den
);
602 avio_seek(pb
, wav
->fact_pos
+ 24, SEEK_SET
);
603 avio_wl64(pb
, number_of_samples
);
606 avio_seek(pb
, file_size
, SEEK_SET
);
613 AVOutputFormat ff_w64_muxer
= {
615 .long_name
= NULL_IF_CONFIG_SMALL("Sony Wave64"),
617 .priv_data_size
= sizeof(WAVMuxContext
),
618 .audio_codec
= AV_CODEC_ID_PCM_S16LE
,
619 .video_codec
= AV_CODEC_ID_NONE
,
620 .write_header
= w64_write_header
,
621 .write_packet
= wav_write_packet
,
622 .write_trailer
= w64_write_trailer
,
623 .flags
= AVFMT_TS_NONSTRICT
,
624 .codec_tag
= (const AVCodecTag
* const []){ ff_codec_wav_tags
, 0 },
626 #endif /* CONFIG_W64_MUXER */