2
0

swresample.h 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579
  1. /*
  2. * Copyright (C) 2011-2013 Michael Niedermayer (michaelni@gmx.at)
  3. *
  4. * This file is part of libswresample
  5. *
  6. * libswresample is free software; you can redistribute it and/or
  7. * modify it under the terms of the GNU Lesser General Public
  8. * License as published by the Free Software Foundation; either
  9. * version 2.1 of the License, or (at your option) any later version.
  10. *
  11. * libswresample is distributed in the hope that it will be useful,
  12. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  14. * Lesser General Public License for more details.
  15. *
  16. * You should have received a copy of the GNU Lesser General Public
  17. * License along with libswresample; if not, write to the Free Software
  18. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19. */
  20. #ifndef SWRESAMPLE_SWRESAMPLE_H
  21. #define SWRESAMPLE_SWRESAMPLE_H
  22. /**
  23. * @file
  24. * @ingroup lswr
  25. * libswresample public header
  26. */
  27. /**
  28. * @defgroup lswr libswresample
  29. * @{
  30. *
  31. * Audio resampling, sample format conversion and mixing library.
  32. *
  33. * Interaction with lswr is done through SwrContext, which is
  34. * allocated with swr_alloc() or swr_alloc_set_opts(). It is opaque, so all parameters
  35. * must be set with the @ref avoptions API.
  36. *
  37. * The first thing you will need to do in order to use lswr is to allocate
  38. * SwrContext. This can be done with swr_alloc() or swr_alloc_set_opts(). If you
  39. * are using the former, you must set options through the @ref avoptions API.
  40. * The latter function provides the same feature, but it allows you to set some
  41. * common options in the same statement.
  42. *
  43. * For example the following code will setup conversion from planar float sample
  44. * format to interleaved signed 16-bit integer, downsampling from 48kHz to
  45. * 44.1kHz and downmixing from 5.1 channels to stereo (using the default mixing
  46. * matrix). This is using the swr_alloc() function.
  47. * @code
  48. * SwrContext *swr = swr_alloc();
  49. * av_opt_set_channel_layout(swr, "in_channel_layout", AV_CH_LAYOUT_5POINT1, 0);
  50. * av_opt_set_channel_layout(swr, "out_channel_layout", AV_CH_LAYOUT_STEREO, 0);
  51. * av_opt_set_int(swr, "in_sample_rate", 48000, 0);
  52. * av_opt_set_int(swr, "out_sample_rate", 44100, 0);
  53. * av_opt_set_sample_fmt(swr, "in_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
  54. * av_opt_set_sample_fmt(swr, "out_sample_fmt", AV_SAMPLE_FMT_S16, 0);
  55. * @endcode
  56. *
  57. * The same job can be done using swr_alloc_set_opts() as well:
  58. * @code
  59. * SwrContext *swr = swr_alloc_set_opts(NULL, // we're allocating a new context
  60. * AV_CH_LAYOUT_STEREO, // out_ch_layout
  61. * AV_SAMPLE_FMT_S16, // out_sample_fmt
  62. * 44100, // out_sample_rate
  63. * AV_CH_LAYOUT_5POINT1, // in_ch_layout
  64. * AV_SAMPLE_FMT_FLTP, // in_sample_fmt
  65. * 48000, // in_sample_rate
  66. * 0, // log_offset
  67. * NULL); // log_ctx
  68. * @endcode
  69. *
  70. * Once all values have been set, it must be initialized with swr_init(). If
  71. * you need to change the conversion parameters, you can change the parameters
  72. * using @ref AVOptions, as described above in the first example; or by using
  73. * swr_alloc_set_opts(), but with the first argument the allocated context.
  74. * You must then call swr_init() again.
  75. *
  76. * The conversion itself is done by repeatedly calling swr_convert().
  77. * Note that the samples may get buffered in swr if you provide insufficient
  78. * output space or if sample rate conversion is done, which requires "future"
  79. * samples. Samples that do not require future input can be retrieved at any
  80. * time by using swr_convert() (in_count can be set to 0).
  81. * At the end of conversion the resampling buffer can be flushed by calling
  82. * swr_convert() with NULL in and 0 in_count.
  83. *
  84. * The samples used in the conversion process can be managed with the libavutil
  85. * @ref lavu_sampmanip "samples manipulation" API, including av_samples_alloc()
  86. * function used in the following example.
  87. *
  88. * The delay between input and output, can at any time be found by using
  89. * swr_get_delay().
  90. *
  91. * The following code demonstrates the conversion loop assuming the parameters
  92. * from above and caller-defined functions get_input() and handle_output():
  93. * @code
  94. * uint8_t **input;
  95. * int in_samples;
  96. *
  97. * while (get_input(&input, &in_samples)) {
  98. * uint8_t *output;
  99. * int out_samples = av_rescale_rnd(swr_get_delay(swr, 48000) +
  100. * in_samples, 44100, 48000, AV_ROUND_UP);
  101. * av_samples_alloc(&output, NULL, 2, out_samples,
  102. * AV_SAMPLE_FMT_S16, 0);
  103. * out_samples = swr_convert(swr, &output, out_samples,
  104. * input, in_samples);
  105. * handle_output(output, out_samples);
  106. * av_freep(&output);
  107. * }
  108. * @endcode
  109. *
  110. * When the conversion is finished, the conversion
  111. * context and everything associated with it must be freed with swr_free().
  112. * A swr_close() function is also available, but it exists mainly for
  113. * compatibility with libavresample, and is not required to be called.
  114. *
  115. * There will be no memory leak if the data is not completely flushed before
  116. * swr_free().
  117. */
  118. #include <stdint.h>
  119. #include "libavutil/channel_layout.h"
  120. #include "libavutil/frame.h"
  121. #include "libavutil/samplefmt.h"
  122. #include "libswresample/version.h"
  123. /**
  124. * @name Option constants
  125. * These constants are used for the @ref avoptions interface for lswr.
  126. * @{
  127. *
  128. */
  129. #define SWR_FLAG_RESAMPLE 1 ///< Force resampling even if equal sample rate
  130. //TODO use int resample ?
  131. //long term TODO can we enable this dynamically?
  132. /** Dithering algorithms */
  133. enum SwrDitherType {
  134. SWR_DITHER_NONE = 0,
  135. SWR_DITHER_RECTANGULAR,
  136. SWR_DITHER_TRIANGULAR,
  137. SWR_DITHER_TRIANGULAR_HIGHPASS,
  138. SWR_DITHER_NS = 64, ///< not part of API/ABI
  139. SWR_DITHER_NS_LIPSHITZ,
  140. SWR_DITHER_NS_F_WEIGHTED,
  141. SWR_DITHER_NS_MODIFIED_E_WEIGHTED,
  142. SWR_DITHER_NS_IMPROVED_E_WEIGHTED,
  143. SWR_DITHER_NS_SHIBATA,
  144. SWR_DITHER_NS_LOW_SHIBATA,
  145. SWR_DITHER_NS_HIGH_SHIBATA,
  146. SWR_DITHER_NB, ///< not part of API/ABI
  147. };
  148. /** Resampling Engines */
  149. enum SwrEngine {
  150. SWR_ENGINE_SWR, /**< SW Resampler */
  151. SWR_ENGINE_SOXR, /**< SoX Resampler */
  152. SWR_ENGINE_NB, ///< not part of API/ABI
  153. };
  154. /** Resampling Filter Types */
  155. enum SwrFilterType {
  156. SWR_FILTER_TYPE_CUBIC, /**< Cubic */
  157. SWR_FILTER_TYPE_BLACKMAN_NUTTALL, /**< Blackman Nuttall windowed sinc */
  158. SWR_FILTER_TYPE_KAISER, /**< Kaiser windowed sinc */
  159. };
  160. /**
  161. * @}
  162. */
  163. /**
  164. * The libswresample context. Unlike libavcodec and libavformat, this structure
  165. * is opaque. This means that if you would like to set options, you must use
  166. * the @ref avoptions API and cannot directly set values to members of the
  167. * structure.
  168. */
  169. typedef struct SwrContext SwrContext;
  170. /**
  171. * Get the AVClass for SwrContext. It can be used in combination with
  172. * AV_OPT_SEARCH_FAKE_OBJ for examining options.
  173. *
  174. * @see av_opt_find().
  175. * @return the AVClass of SwrContext
  176. */
  177. const AVClass *swr_get_class(void);
  178. /**
  179. * @name SwrContext constructor functions
  180. * @{
  181. */
  182. /**
  183. * Allocate SwrContext.
  184. *
  185. * If you use this function you will need to set the parameters (manually or
  186. * with swr_alloc_set_opts()) before calling swr_init().
  187. *
  188. * @see swr_alloc_set_opts(), swr_init(), swr_free()
  189. * @return NULL on error, allocated context otherwise
  190. */
  191. struct SwrContext *swr_alloc(void);
  192. /**
  193. * Initialize context after user parameters have been set.
  194. * @note The context must be configured using the AVOption API.
  195. *
  196. * @see av_opt_set_int()
  197. * @see av_opt_set_dict()
  198. *
  199. * @param[in,out] s Swr context to initialize
  200. * @return AVERROR error code in case of failure.
  201. */
  202. int swr_init(struct SwrContext *s);
  203. /**
  204. * Check whether an swr context has been initialized or not.
  205. *
  206. * @param[in] s Swr context to check
  207. * @see swr_init()
  208. * @return positive if it has been initialized, 0 if not initialized
  209. */
  210. int swr_is_initialized(struct SwrContext *s);
  211. /**
  212. * Allocate SwrContext if needed and set/reset common parameters.
  213. *
  214. * This function does not require s to be allocated with swr_alloc(). On the
  215. * other hand, swr_alloc() can use swr_alloc_set_opts() to set the parameters
  216. * on the allocated context.
  217. *
  218. * @param s existing Swr context if available, or NULL if not
  219. * @param out_ch_layout output channel layout (AV_CH_LAYOUT_*)
  220. * @param out_sample_fmt output sample format (AV_SAMPLE_FMT_*).
  221. * @param out_sample_rate output sample rate (frequency in Hz)
  222. * @param in_ch_layout input channel layout (AV_CH_LAYOUT_*)
  223. * @param in_sample_fmt input sample format (AV_SAMPLE_FMT_*).
  224. * @param in_sample_rate input sample rate (frequency in Hz)
  225. * @param log_offset logging level offset
  226. * @param log_ctx parent logging context, can be NULL
  227. *
  228. * @see swr_init(), swr_free()
  229. * @return NULL on error, allocated context otherwise
  230. */
  231. struct SwrContext *swr_alloc_set_opts(struct SwrContext *s,
  232. int64_t out_ch_layout, enum AVSampleFormat out_sample_fmt, int out_sample_rate,
  233. int64_t in_ch_layout, enum AVSampleFormat in_sample_fmt, int in_sample_rate,
  234. int log_offset, void *log_ctx);
  235. /**
  236. * @}
  237. *
  238. * @name SwrContext destructor functions
  239. * @{
  240. */
  241. /**
  242. * Free the given SwrContext and set the pointer to NULL.
  243. *
  244. * @param[in] s a pointer to a pointer to Swr context
  245. */
  246. void swr_free(struct SwrContext **s);
  247. /**
  248. * Closes the context so that swr_is_initialized() returns 0.
  249. *
  250. * The context can be brought back to life by running swr_init(),
  251. * swr_init() can also be used without swr_close().
  252. * This function is mainly provided for simplifying the usecase
  253. * where one tries to support libavresample and libswresample.
  254. *
  255. * @param[in,out] s Swr context to be closed
  256. */
  257. void swr_close(struct SwrContext *s);
  258. /**
  259. * @}
  260. *
  261. * @name Core conversion functions
  262. * @{
  263. */
  264. /** Convert audio.
  265. *
  266. * in and in_count can be set to 0 to flush the last few samples out at the
  267. * end.
  268. *
  269. * If more input is provided than output space, then the input will be buffered.
  270. * You can avoid this buffering by using swr_get_out_samples() to retrieve an
  271. * upper bound on the required number of output samples for the given number of
  272. * input samples. Conversion will run directly without copying whenever possible.
  273. *
  274. * @param s allocated Swr context, with parameters set
  275. * @param out output buffers, only the first one need be set in case of packed audio
  276. * @param out_count amount of space available for output in samples per channel
  277. * @param in input buffers, only the first one need to be set in case of packed audio
  278. * @param in_count number of input samples available in one channel
  279. *
  280. * @return number of samples output per channel, negative value on error
  281. */
  282. int swr_convert(struct SwrContext *s, uint8_t **out, int out_count,
  283. const uint8_t **in , int in_count);
  284. /**
  285. * Convert the next timestamp from input to output
  286. * timestamps are in 1/(in_sample_rate * out_sample_rate) units.
  287. *
  288. * @note There are 2 slightly differently behaving modes.
  289. * @li When automatic timestamp compensation is not used, (min_compensation >= FLT_MAX)
  290. * in this case timestamps will be passed through with delays compensated
  291. * @li When automatic timestamp compensation is used, (min_compensation < FLT_MAX)
  292. * in this case the output timestamps will match output sample numbers.
  293. * See ffmpeg-resampler(1) for the two modes of compensation.
  294. *
  295. * @param s[in] initialized Swr context
  296. * @param pts[in] timestamp for the next input sample, INT64_MIN if unknown
  297. * @see swr_set_compensation(), swr_drop_output(), and swr_inject_silence() are
  298. * function used internally for timestamp compensation.
  299. * @return the output timestamp for the next output sample
  300. */
  301. int64_t swr_next_pts(struct SwrContext *s, int64_t pts);
  302. /**
  303. * @}
  304. *
  305. * @name Low-level option setting functions
  306. * These functons provide a means to set low-level options that is not possible
  307. * with the AVOption API.
  308. * @{
  309. */
  310. /**
  311. * Activate resampling compensation ("soft" compensation). This function is
  312. * internally called when needed in swr_next_pts().
  313. *
  314. * @param[in,out] s allocated Swr context. If it is not initialized,
  315. * or SWR_FLAG_RESAMPLE is not set, swr_init() is
  316. * called with the flag set.
  317. * @param[in] sample_delta delta in PTS per sample
  318. * @param[in] compensation_distance number of samples to compensate for
  319. * @return >= 0 on success, AVERROR error codes if:
  320. * @li @c s is NULL,
  321. * @li @c compensation_distance is less than 0,
  322. * @li @c compensation_distance is 0 but sample_delta is not,
  323. * @li compensation unsupported by resampler, or
  324. * @li swr_init() fails when called.
  325. */
  326. int swr_set_compensation(struct SwrContext *s, int sample_delta, int compensation_distance);
  327. /**
  328. * Set a customized input channel mapping.
  329. *
  330. * @param[in,out] s allocated Swr context, not yet initialized
  331. * @param[in] channel_map customized input channel mapping (array of channel
  332. * indexes, -1 for a muted channel)
  333. * @return >= 0 on success, or AVERROR error code in case of failure.
  334. */
  335. int swr_set_channel_mapping(struct SwrContext *s, const int *channel_map);
  336. /**
  337. * Generate a channel mixing matrix.
  338. *
  339. * This function is the one used internally by libswresample for building the
  340. * default mixing matrix. It is made public just as a utility function for
  341. * building custom matrices.
  342. *
  343. * @param in_layout input channel layout
  344. * @param out_layout output channel layout
  345. * @param center_mix_level mix level for the center channel
  346. * @param surround_mix_level mix level for the surround channel(s)
  347. * @param lfe_mix_level mix level for the low-frequency effects channel
  348. * @param rematrix_maxval if 1.0, coefficients will be normalized to prevent
  349. * overflow. if INT_MAX, coefficients will not be
  350. * normalized.
  351. * @param[out] matrix mixing coefficients; matrix[i + stride * o] is
  352. * the weight of input channel i in output channel o.
  353. * @param stride distance between adjacent input channels in the
  354. * matrix array
  355. * @param matrix_encoding matrixed stereo downmix mode (e.g. dplii)
  356. * @param log_ctx parent logging context, can be NULL
  357. * @return 0 on success, negative AVERROR code on failure
  358. */
  359. int swr_build_matrix(uint64_t in_layout, uint64_t out_layout,
  360. double center_mix_level, double surround_mix_level,
  361. double lfe_mix_level, double rematrix_maxval,
  362. double rematrix_volume, double *matrix,
  363. int stride, enum AVMatrixEncoding matrix_encoding,
  364. void *log_ctx);
  365. /**
  366. * Set a customized remix matrix.
  367. *
  368. * @param s allocated Swr context, not yet initialized
  369. * @param matrix remix coefficients; matrix[i + stride * o] is
  370. * the weight of input channel i in output channel o
  371. * @param stride offset between lines of the matrix
  372. * @return >= 0 on success, or AVERROR error code in case of failure.
  373. */
  374. int swr_set_matrix(struct SwrContext *s, const double *matrix, int stride);
  375. /**
  376. * @}
  377. *
  378. * @name Sample handling functions
  379. * @{
  380. */
  381. /**
  382. * Drops the specified number of output samples.
  383. *
  384. * This function, along with swr_inject_silence(), is called by swr_next_pts()
  385. * if needed for "hard" compensation.
  386. *
  387. * @param s allocated Swr context
  388. * @param count number of samples to be dropped
  389. *
  390. * @return >= 0 on success, or a negative AVERROR code on failure
  391. */
  392. int swr_drop_output(struct SwrContext *s, int count);
  393. /**
  394. * Injects the specified number of silence samples.
  395. *
  396. * This function, along with swr_drop_output(), is called by swr_next_pts()
  397. * if needed for "hard" compensation.
  398. *
  399. * @param s allocated Swr context
  400. * @param count number of samples to be dropped
  401. *
  402. * @return >= 0 on success, or a negative AVERROR code on failure
  403. */
  404. int swr_inject_silence(struct SwrContext *s, int count);
  405. /**
  406. * Gets the delay the next input sample will experience relative to the next output sample.
  407. *
  408. * Swresample can buffer data if more input has been provided than available
  409. * output space, also converting between sample rates needs a delay.
  410. * This function returns the sum of all such delays.
  411. * The exact delay is not necessarily an integer value in either input or
  412. * output sample rate. Especially when downsampling by a large value, the
  413. * output sample rate may be a poor choice to represent the delay, similarly
  414. * for upsampling and the input sample rate.
  415. *
  416. * @param s swr context
  417. * @param base timebase in which the returned delay will be:
  418. * @li if it's set to 1 the returned delay is in seconds
  419. * @li if it's set to 1000 the returned delay is in milliseconds
  420. * @li if it's set to the input sample rate then the returned
  421. * delay is in input samples
  422. * @li if it's set to the output sample rate then the returned
  423. * delay is in output samples
  424. * @li if it's the least common multiple of in_sample_rate and
  425. * out_sample_rate then an exact rounding-free delay will be
  426. * returned
  427. * @returns the delay in 1 / @c base units.
  428. */
  429. int64_t swr_get_delay(struct SwrContext *s, int64_t base);
  430. /**
  431. * Find an upper bound on the number of samples that the next swr_convert
  432. * call will output, if called with in_samples of input samples. This
  433. * depends on the internal state, and anything changing the internal state
  434. * (like further swr_convert() calls) will may change the number of samples
  435. * swr_get_out_samples() returns for the same number of input samples.
  436. *
  437. * @param in_samples number of input samples.
  438. * @note any call to swr_inject_silence(), swr_convert(), swr_next_pts()
  439. * or swr_set_compensation() invalidates this limit
  440. * @note it is recommended to pass the correct available buffer size
  441. * to all functions like swr_convert() even if swr_get_out_samples()
  442. * indicates that less would be used.
  443. * @returns an upper bound on the number of samples that the next swr_convert
  444. * will output or a negative value to indicate an error
  445. */
  446. int swr_get_out_samples(struct SwrContext *s, int in_samples);
  447. /**
  448. * @}
  449. *
  450. * @name Configuration accessors
  451. * @{
  452. */
  453. /**
  454. * Return the @ref LIBSWRESAMPLE_VERSION_INT constant.
  455. *
  456. * This is useful to check if the build-time libswresample has the same version
  457. * as the run-time one.
  458. *
  459. * @returns the unsigned int-typed version
  460. */
  461. unsigned swresample_version(void);
  462. /**
  463. * Return the swr build-time configuration.
  464. *
  465. * @returns the build-time @c ./configure flags
  466. */
  467. const char *swresample_configuration(void);
  468. /**
  469. * Return the swr license.
  470. *
  471. * @returns the license of libswresample, determined at build-time
  472. */
  473. const char *swresample_license(void);
  474. /**
  475. * @}
  476. *
  477. * @name AVFrame based API
  478. * @{
  479. */
  480. /**
  481. * Convert the samples in the input AVFrame and write them to the output AVFrame.
  482. *
  483. * Input and output AVFrames must have channel_layout, sample_rate and format set.
  484. *
  485. * If the output AVFrame does not have the data pointers allocated the nb_samples
  486. * field will be set using av_frame_get_buffer()
  487. * is called to allocate the frame.
  488. *
  489. * The output AVFrame can be NULL or have fewer allocated samples than required.
  490. * In this case, any remaining samples not written to the output will be added
  491. * to an internal FIFO buffer, to be returned at the next call to this function
  492. * or to swr_convert().
  493. *
  494. * If converting sample rate, there may be data remaining in the internal
  495. * resampling delay buffer. swr_get_delay() tells the number of
  496. * remaining samples. To get this data as output, call this function or
  497. * swr_convert() with NULL input.
  498. *
  499. * If the SwrContext configuration does not match the output and
  500. * input AVFrame settings the conversion does not take place and depending on
  501. * which AVFrame is not matching AVERROR_OUTPUT_CHANGED, AVERROR_INPUT_CHANGED
  502. * or the result of a bitwise-OR of them is returned.
  503. *
  504. * @see swr_delay()
  505. * @see swr_convert()
  506. * @see swr_get_delay()
  507. *
  508. * @param swr audio resample context
  509. * @param output output AVFrame
  510. * @param input input AVFrame
  511. * @return 0 on success, AVERROR on failure or nonmatching
  512. * configuration.
  513. */
  514. int swr_convert_frame(SwrContext *swr,
  515. AVFrame *output, const AVFrame *input);
  516. /**
  517. * Configure or reconfigure the SwrContext using the information
  518. * provided by the AVFrames.
  519. *
  520. * The original resampling context is reset even on failure.
  521. * The function calls swr_close() internally if the context is open.
  522. *
  523. * @see swr_close();
  524. *
  525. * @param swr audio resample context
  526. * @param output output AVFrame
  527. * @param input input AVFrame
  528. * @return 0 on success, AVERROR on failure.
  529. */
  530. int swr_config_frame(SwrContext *swr, const AVFrame *out, const AVFrame *in);
  531. /**
  532. * @}
  533. * @}
  534. */
  535. #endif /* SWRESAMPLE_SWRESAMPLE_H */