summaryrefslogtreecommitdiff
path: root/libswresample/swresample.h (plain)
blob: a8db5c2490b632cd09f344d36415e055bbc2f5f6
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
21#ifndef SWRESAMPLE_SWRESAMPLE_H
22#define SWRESAMPLE_SWRESAMPLE_H
23
24/**
25 * @file
26 * @ingroup lswr
27 * libswresample public header
28 */
29
30/**
31 * @defgroup lswr libswresample
32 * @{
33 *
34 * Audio resampling, sample format conversion and mixing library.
35 *
36 * Interaction with lswr is done through SwrContext, which is
37 * allocated with swr_alloc() or swr_alloc_set_opts(). It is opaque, so all parameters
38 * must be set with the @ref avoptions API.
39 *
40 * The first thing you will need to do in order to use lswr is to allocate
41 * SwrContext. This can be done with swr_alloc() or swr_alloc_set_opts(). If you
42 * are using the former, you must set options through the @ref avoptions API.
43 * The latter function provides the same feature, but it allows you to set some
44 * common options in the same statement.
45 *
46 * For example the following code will setup conversion from planar float sample
47 * format to interleaved signed 16-bit integer, downsampling from 48kHz to
48 * 44.1kHz and downmixing from 5.1 channels to stereo (using the default mixing
49 * matrix). This is using the swr_alloc() function.
50 * @code
51 * SwrContext *swr = swr_alloc();
52 * av_opt_set_channel_layout(swr, "in_channel_layout", AV_CH_LAYOUT_5POINT1, 0);
53 * av_opt_set_channel_layout(swr, "out_channel_layout", AV_CH_LAYOUT_STEREO, 0);
54 * av_opt_set_int(swr, "in_sample_rate", 48000, 0);
55 * av_opt_set_int(swr, "out_sample_rate", 44100, 0);
56 * av_opt_set_sample_fmt(swr, "in_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
57 * av_opt_set_sample_fmt(swr, "out_sample_fmt", AV_SAMPLE_FMT_S16, 0);
58 * @endcode
59 *
60 * The same job can be done using swr_alloc_set_opts() as well:
61 * @code
62 * SwrContext *swr = swr_alloc_set_opts(NULL, // we're allocating a new context
63 * AV_CH_LAYOUT_STEREO, // out_ch_layout
64 * AV_SAMPLE_FMT_S16, // out_sample_fmt
65 * 44100, // out_sample_rate
66 * AV_CH_LAYOUT_5POINT1, // in_ch_layout
67 * AV_SAMPLE_FMT_FLTP, // in_sample_fmt
68 * 48000, // in_sample_rate
69 * 0, // log_offset
70 * NULL); // log_ctx
71 * @endcode
72 *
73 * Once all values have been set, it must be initialized with swr_init(). If
74 * you need to change the conversion parameters, you can change the parameters
75 * using @ref AVOptions, as described above in the first example; or by using
76 * swr_alloc_set_opts(), but with the first argument the allocated context.
77 * You must then call swr_init() again.
78 *
79 * The conversion itself is done by repeatedly calling swr_convert().
80 * Note that the samples may get buffered in swr if you provide insufficient
81 * output space or if sample rate conversion is done, which requires "future"
82 * samples. Samples that do not require future input can be retrieved at any
83 * time by using swr_convert() (in_count can be set to 0).
84 * At the end of conversion the resampling buffer can be flushed by calling
85 * swr_convert() with NULL in and 0 in_count.
86 *
87 * The samples used in the conversion process can be managed with the libavutil
88 * @ref lavu_sampmanip "samples manipulation" API, including av_samples_alloc()
89 * function used in the following example.
90 *
91 * The delay between input and output, can at any time be found by using
92 * swr_get_delay().
93 *
94 * The following code demonstrates the conversion loop assuming the parameters
95 * from above and caller-defined functions get_input() and handle_output():
96 * @code
97 * uint8_t **input;
98 * int in_samples;
99 *
100 * while (get_input(&input, &in_samples)) {
101 * uint8_t *output;
102 * int out_samples = av_rescale_rnd(swr_get_delay(swr, 48000) +
103 * in_samples, 44100, 48000, AV_ROUND_UP);
104 * av_samples_alloc(&output, NULL, 2, out_samples,
105 * AV_SAMPLE_FMT_S16, 0);
106 * out_samples = swr_convert(swr, &output, out_samples,
107 * input, in_samples);
108 * handle_output(output, out_samples);
109 * av_freep(&output);
110 * }
111 * @endcode
112 *
113 * When the conversion is finished, the conversion
114 * context and everything associated with it must be freed with swr_free().
115 * A swr_close() function is also available, but it exists mainly for
116 * compatibility with libavresample, and is not required to be called.
117 *
118 * There will be no memory leak if the data is not completely flushed before
119 * swr_free().
120 */
121
122#include <stdint.h>
123#include "libavutil/channel_layout.h"
124#include "libavutil/frame.h"
125#include "libavutil/samplefmt.h"
126
127#include "libswresample/version.h"
128
129#if LIBSWRESAMPLE_VERSION_MAJOR < 1
130#define SWR_CH_MAX 32 ///< Maximum number of channels
131#endif
132
133/**
134 * @name Option constants
135 * These constants are used for the @ref avoptions interface for lswr.
136 * @{
137 *
138 */
139
140#define SWR_FLAG_RESAMPLE 1 ///< Force resampling even if equal sample rate
141//TODO use int resample ?
142//long term TODO can we enable this dynamically?
143
144/** Dithering algorithms */
145enum SwrDitherType {
146 SWR_DITHER_NONE = 0,
147 SWR_DITHER_RECTANGULAR,
148 SWR_DITHER_TRIANGULAR,
149 SWR_DITHER_TRIANGULAR_HIGHPASS,
150
151 SWR_DITHER_NS = 64, ///< not part of API/ABI
152 SWR_DITHER_NS_LIPSHITZ,
153 SWR_DITHER_NS_F_WEIGHTED,
154 SWR_DITHER_NS_MODIFIED_E_WEIGHTED,
155 SWR_DITHER_NS_IMPROVED_E_WEIGHTED,
156 SWR_DITHER_NS_SHIBATA,
157 SWR_DITHER_NS_LOW_SHIBATA,
158 SWR_DITHER_NS_HIGH_SHIBATA,
159 SWR_DITHER_NB, ///< not part of API/ABI
160};
161
162/** Resampling Engines */
163enum SwrEngine {
164 SWR_ENGINE_SWR, /**< SW Resampler */
165 SWR_ENGINE_SOXR, /**< SoX Resampler */
166 SWR_ENGINE_NB, ///< not part of API/ABI
167};
168
169/** Resampling Filter Types */
170enum SwrFilterType {
171 SWR_FILTER_TYPE_CUBIC, /**< Cubic */
172 SWR_FILTER_TYPE_BLACKMAN_NUTTALL, /**< Blackman Nuttall windowed sinc */
173 SWR_FILTER_TYPE_KAISER, /**< Kaiser windowed sinc */
174};
175
176/**
177 * @}
178 */
179
180/**
181 * The libswresample context. Unlike libavcodec and libavformat, this structure
182 * is opaque. This means that if you would like to set options, you must use
183 * the @ref avoptions API and cannot directly set values to members of the
184 * structure.
185 */
186typedef struct SwrContext SwrContext;
187
188/**
189 * Get the AVClass for SwrContext. It can be used in combination with
190 * AV_OPT_SEARCH_FAKE_OBJ for examining options.
191 *
192 * @see av_opt_find().
193 * @return the AVClass of SwrContext
194 */
195const AVClass *swr_get_class(void);
196
197/**
198 * @name SwrContext constructor functions
199 * @{
200 */
201
202/**
203 * Allocate SwrContext.
204 *
205 * If you use this function you will need to set the parameters (manually or
206 * with swr_alloc_set_opts()) before calling swr_init().
207 *
208 * @see swr_alloc_set_opts(), swr_init(), swr_free()
209 * @return NULL on error, allocated context otherwise
210 */
211struct SwrContext *swr_alloc(void);
212
213/**
214 * Initialize context after user parameters have been set.
215 * @note The context must be configured using the AVOption API.
216 *
217 * @see av_opt_set_int()
218 * @see av_opt_set_dict()
219 *
220 * @param[in,out] s Swr context to initialize
221 * @return AVERROR error code in case of failure.
222 */
223int swr_init(struct SwrContext *s);
224
225/**
226 * Check whether an swr context has been initialized or not.
227 *
228 * @param[in] s Swr context to check
229 * @see swr_init()
230 * @return positive if it has been initialized, 0 if not initialized
231 */
232int swr_is_initialized(struct SwrContext *s);
233
234/**
235 * Allocate SwrContext if needed and set/reset common parameters.
236 *
237 * This function does not require s to be allocated with swr_alloc(). On the
238 * other hand, swr_alloc() can use swr_alloc_set_opts() to set the parameters
239 * on the allocated context.
240 *
241 * @param s existing Swr context if available, or NULL if not
242 * @param out_ch_layout output channel layout (AV_CH_LAYOUT_*)
243 * @param out_sample_fmt output sample format (AV_SAMPLE_FMT_*).
244 * @param out_sample_rate output sample rate (frequency in Hz)
245 * @param in_ch_layout input channel layout (AV_CH_LAYOUT_*)
246 * @param in_sample_fmt input sample format (AV_SAMPLE_FMT_*).
247 * @param in_sample_rate input sample rate (frequency in Hz)
248 * @param log_offset logging level offset
249 * @param log_ctx parent logging context, can be NULL
250 *
251 * @see swr_init(), swr_free()
252 * @return NULL on error, allocated context otherwise
253 */
254struct SwrContext *swr_alloc_set_opts(struct SwrContext *s,
255 int64_t out_ch_layout, enum AVSampleFormat out_sample_fmt, int out_sample_rate,
256 int64_t in_ch_layout, enum AVSampleFormat in_sample_fmt, int in_sample_rate,
257 int log_offset, void *log_ctx);
258
259/**
260 * @}
261 *
262 * @name SwrContext destructor functions
263 * @{
264 */
265
266/**
267 * Free the given SwrContext and set the pointer to NULL.
268 *
269 * @param[in] s a pointer to a pointer to Swr context
270 */
271void swr_free(struct SwrContext **s);
272
273/**
274 * Closes the context so that swr_is_initialized() returns 0.
275 *
276 * The context can be brought back to life by running swr_init(),
277 * swr_init() can also be used without swr_close().
278 * This function is mainly provided for simplifying the usecase
279 * where one tries to support libavresample and libswresample.
280 *
281 * @param[in,out] s Swr context to be closed
282 */
283void swr_close(struct SwrContext *s);
284
285/**
286 * @}
287 *
288 * @name Core conversion functions
289 * @{
290 */
291
292/** Convert audio.
293 *
294 * in and in_count can be set to 0 to flush the last few samples out at the
295 * end.
296 *
297 * If more input is provided than output space, then the input will be buffered.
298 * You can avoid this buffering by using swr_get_out_samples() to retrieve an
299 * upper bound on the required number of output samples for the given number of
300 * input samples. Conversion will run directly without copying whenever possible.
301 *
302 * @param s allocated Swr context, with parameters set
303 * @param out output buffers, only the first one need be set in case of packed audio
304 * @param out_count amount of space available for output in samples per channel
305 * @param in input buffers, only the first one need to be set in case of packed audio
306 * @param in_count number of input samples available in one channel
307 *
308 * @return number of samples output per channel, negative value on error
309 */
310int swr_convert(struct SwrContext *s, uint8_t **out, int out_count,
311 const uint8_t **in , int in_count);
312
313/**
314 * Convert the next timestamp from input to output
315 * timestamps are in 1/(in_sample_rate * out_sample_rate) units.
316 *
317 * @note There are 2 slightly differently behaving modes.
318 * @li When automatic timestamp compensation is not used, (min_compensation >= FLT_MAX)
319 * in this case timestamps will be passed through with delays compensated
320 * @li When automatic timestamp compensation is used, (min_compensation < FLT_MAX)
321 * in this case the output timestamps will match output sample numbers.
322 * See ffmpeg-resampler(1) for the two modes of compensation.
323 *
324 * @param s[in] initialized Swr context
325 * @param pts[in] timestamp for the next input sample, INT64_MIN if unknown
326 * @see swr_set_compensation(), swr_drop_output(), and swr_inject_silence() are
327 * function used internally for timestamp compensation.
328 * @return the output timestamp for the next output sample
329 */
330int64_t swr_next_pts(struct SwrContext *s, int64_t pts);
331
332/**
333 * @}
334 *
335 * @name Low-level option setting functions
336 * These functons provide a means to set low-level options that is not possible
337 * with the AVOption API.
338 * @{
339 */
340
341/**
342 * Activate resampling compensation ("soft" compensation). This function is
343 * internally called when needed in swr_next_pts().
344 *
345 * @param[in,out] s allocated Swr context. If it is not initialized,
346 * or SWR_FLAG_RESAMPLE is not set, swr_init() is
347 * called with the flag set.
348 * @param[in] sample_delta delta in PTS per sample
349 * @param[in] compensation_distance number of samples to compensate for
350 * @return >= 0 on success, AVERROR error codes if:
351 * @li @c s is NULL,
352 * @li @c compensation_distance is less than 0,
353 * @li @c compensation_distance is 0 but sample_delta is not,
354 * @li compensation unsupported by resampler, or
355 * @li swr_init() fails when called.
356 */
357int swr_set_compensation(struct SwrContext *s, int sample_delta, int compensation_distance);
358
359/**
360 * Set a customized input channel mapping.
361 *
362 * @param[in,out] s allocated Swr context, not yet initialized
363 * @param[in] channel_map customized input channel mapping (array of channel
364 * indexes, -1 for a muted channel)
365 * @return >= 0 on success, or AVERROR error code in case of failure.
366 */
367int swr_set_channel_mapping(struct SwrContext *s, const int *channel_map);
368
369/**
370 * Generate a channel mixing matrix.
371 *
372 * This function is the one used internally by libswresample for building the
373 * default mixing matrix. It is made public just as a utility function for
374 * building custom matrices.
375 *
376 * @param in_layout input channel layout
377 * @param out_layout output channel layout
378 * @param center_mix_level mix level for the center channel
379 * @param surround_mix_level mix level for the surround channel(s)
380 * @param lfe_mix_level mix level for the low-frequency effects channel
381 * @param rematrix_maxval if 1.0, coefficients will be normalized to prevent
382 * overflow. if INT_MAX, coefficients will not be
383 * normalized.
384 * @param[out] matrix mixing coefficients; matrix[i + stride * o] is
385 * the weight of input channel i in output channel o.
386 * @param stride distance between adjacent input channels in the
387 * matrix array
388 * @param matrix_encoding matrixed stereo downmix mode (e.g. dplii)
389 * @param log_ctx parent logging context, can be NULL
390 * @return 0 on success, negative AVERROR code on failure
391 */
392int swr_build_matrix(uint64_t in_layout, uint64_t out_layout,
393 double center_mix_level, double surround_mix_level,
394 double lfe_mix_level, double rematrix_maxval,
395 double rematrix_volume, double *matrix,
396 int stride, enum AVMatrixEncoding matrix_encoding,
397 void *log_ctx);
398
399/**
400 * Set a customized remix matrix.
401 *
402 * @param s allocated Swr context, not yet initialized
403 * @param matrix remix coefficients; matrix[i + stride * o] is
404 * the weight of input channel i in output channel o
405 * @param stride offset between lines of the matrix
406 * @return >= 0 on success, or AVERROR error code in case of failure.
407 */
408int swr_set_matrix(struct SwrContext *s, const double *matrix, int stride);
409
410/**
411 * @}
412 *
413 * @name Sample handling functions
414 * @{
415 */
416
417/**
418 * Drops the specified number of output samples.
419 *
420 * This function, along with swr_inject_silence(), is called by swr_next_pts()
421 * if needed for "hard" compensation.
422 *
423 * @param s allocated Swr context
424 * @param count number of samples to be dropped
425 *
426 * @return >= 0 on success, or a negative AVERROR code on failure
427 */
428int swr_drop_output(struct SwrContext *s, int count);
429
430/**
431 * Injects the specified number of silence samples.
432 *
433 * This function, along with swr_drop_output(), is called by swr_next_pts()
434 * if needed for "hard" compensation.
435 *
436 * @param s allocated Swr context
437 * @param count number of samples to be dropped
438 *
439 * @return >= 0 on success, or a negative AVERROR code on failure
440 */
441int swr_inject_silence(struct SwrContext *s, int count);
442
443/**
444 * Gets the delay the next input sample will experience relative to the next output sample.
445 *
446 * Swresample can buffer data if more input has been provided than available
447 * output space, also converting between sample rates needs a delay.
448 * This function returns the sum of all such delays.
449 * The exact delay is not necessarily an integer value in either input or
450 * output sample rate. Especially when downsampling by a large value, the
451 * output sample rate may be a poor choice to represent the delay, similarly
452 * for upsampling and the input sample rate.
453 *
454 * @param s swr context
455 * @param base timebase in which the returned delay will be:
456 * @li if it's set to 1 the returned delay is in seconds
457 * @li if it's set to 1000 the returned delay is in milliseconds
458 * @li if it's set to the input sample rate then the returned
459 * delay is in input samples
460 * @li if it's set to the output sample rate then the returned
461 * delay is in output samples
462 * @li if it's the least common multiple of in_sample_rate and
463 * out_sample_rate then an exact rounding-free delay will be
464 * returned
465 * @returns the delay in 1 / @c base units.
466 */
467int64_t swr_get_delay(struct SwrContext *s, int64_t base);
468
469/**
470 * Find an upper bound on the number of samples that the next swr_convert
471 * call will output, if called with in_samples of input samples. This
472 * depends on the internal state, and anything changing the internal state
473 * (like further swr_convert() calls) will may change the number of samples
474 * swr_get_out_samples() returns for the same number of input samples.
475 *
476 * @param in_samples number of input samples.
477 * @note any call to swr_inject_silence(), swr_convert(), swr_next_pts()
478 * or swr_set_compensation() invalidates this limit
479 * @note it is recommended to pass the correct available buffer size
480 * to all functions like swr_convert() even if swr_get_out_samples()
481 * indicates that less would be used.
482 * @returns an upper bound on the number of samples that the next swr_convert
483 * will output or a negative value to indicate an error
484 */
485int swr_get_out_samples(struct SwrContext *s, int in_samples);
486
487/**
488 * @}
489 *
490 * @name Configuration accessors
491 * @{
492 */
493
494/**
495 * Return the @ref LIBSWRESAMPLE_VERSION_INT constant.
496 *
497 * This is useful to check if the build-time libswresample has the same version
498 * as the run-time one.
499 *
500 * @returns the unsigned int-typed version
501 */
502unsigned swresample_version(void);
503
504/**
505 * Return the swr build-time configuration.
506 *
507 * @returns the build-time @c ./configure flags
508 */
509const char *swresample_configuration(void);
510
511/**
512 * Return the swr license.
513 *
514 * @returns the license of libswresample, determined at build-time
515 */
516const char *swresample_license(void);
517
518/**
519 * @}
520 *
521 * @name AVFrame based API
522 * @{
523 */
524
525/**
526 * Convert the samples in the input AVFrame and write them to the output AVFrame.
527 *
528 * Input and output AVFrames must have channel_layout, sample_rate and format set.
529 *
530 * If the output AVFrame does not have the data pointers allocated the nb_samples
531 * field will be set using av_frame_get_buffer()
532 * is called to allocate the frame.
533 *
534 * The output AVFrame can be NULL or have fewer allocated samples than required.
535 * In this case, any remaining samples not written to the output will be added
536 * to an internal FIFO buffer, to be returned at the next call to this function
537 * or to swr_convert().
538 *
539 * If converting sample rate, there may be data remaining in the internal
540 * resampling delay buffer. swr_get_delay() tells the number of
541 * remaining samples. To get this data as output, call this function or
542 * swr_convert() with NULL input.
543 *
544 * If the SwrContext configuration does not match the output and
545 * input AVFrame settings the conversion does not take place and depending on
546 * which AVFrame is not matching AVERROR_OUTPUT_CHANGED, AVERROR_INPUT_CHANGED
547 * or the result of a bitwise-OR of them is returned.
548 *
549 * @see swr_delay()
550 * @see swr_convert()
551 * @see swr_get_delay()
552 *
553 * @param swr audio resample context
554 * @param output output AVFrame
555 * @param input input AVFrame
556 * @return 0 on success, AVERROR on failure or nonmatching
557 * configuration.
558 */
559int swr_convert_frame(SwrContext *swr,
560 AVFrame *output, const AVFrame *input);
561
562/**
563 * Configure or reconfigure the SwrContext using the information
564 * provided by the AVFrames.
565 *
566 * The original resampling context is reset even on failure.
567 * The function calls swr_close() internally if the context is open.
568 *
569 * @see swr_close();
570 *
571 * @param swr audio resample context
572 * @param output output AVFrame
573 * @param input input AVFrame
574 * @return 0 on success, AVERROR on failure.
575 */
576int swr_config_frame(SwrContext *swr, const AVFrame *out, const AVFrame *in);
577
578/**
579 * @}
580 * @}
581 */
582
583#endif /* SWRESAMPLE_SWRESAMPLE_H */
584
generated by cgit at 2024-05-17 03:15:10 (GMT)