FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
swresample.h
Go to the documentation of this file.
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  * Libswresample (lswr) is a library that handles audio resampling, sample
35  * format conversion and mixing.
36  *
37  * Interaction with lswr is done through SwrContext, which is
38  * allocated with swr_alloc() or swr_alloc_set_opts(). It is opaque, so all parameters
39  * must be set with the @ref avoptions API.
40  *
41  * The first thing you will need to do in order to use lswr is to allocate
42  * SwrContext. This can be done with swr_alloc() or swr_alloc_set_opts(). If you
43  * are using the former, you must set options through the @ref avoptions API.
44  * The latter function provides the same feature, but it allows you to set some
45  * common options in the same statement.
46  *
47  * For example the following code will setup conversion from planar float sample
48  * format to interleaved signed 16-bit integer, downsampling from 48kHz to
49  * 44.1kHz and downmixing from 5.1 channels to stereo (using the default mixing
50  * matrix). This is using the swr_alloc() function.
51  * @code
52  * SwrContext *swr = swr_alloc();
53  * av_opt_set_channel_layout(swr, "in_channel_layout", AV_CH_LAYOUT_5POINT1, 0);
54  * av_opt_set_channel_layout(swr, "out_channel_layout", AV_CH_LAYOUT_STEREO, 0);
55  * av_opt_set_int(swr, "in_sample_rate", 48000, 0);
56  * av_opt_set_int(swr, "out_sample_rate", 44100, 0);
57  * av_opt_set_sample_fmt(swr, "in_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
58  * av_opt_set_sample_fmt(swr, "out_sample_fmt", AV_SAMPLE_FMT_S16, 0);
59  * @endcode
60  *
61  * The same job can be done using swr_alloc_set_opts() as well:
62  * @code
63  * SwrContext *swr = swr_alloc_set_opts(NULL, // we're allocating a new context
64  * AV_CH_LAYOUT_STEREO, // out_ch_layout
65  * AV_SAMPLE_FMT_S16, // out_sample_fmt
66  * 44100, // out_sample_rate
67  * AV_CH_LAYOUT_5POINT1, // in_ch_layout
68  * AV_SAMPLE_FMT_FLTP, // in_sample_fmt
69  * 48000, // in_sample_rate
70  * 0, // log_offset
71  * NULL); // log_ctx
72  * @endcode
73  *
74  * Once all values have been set, it must be initialized with swr_init(). If
75  * you need to change the conversion parameters, you can change the parameters
76  * using @ref AVOptions, as described above in the first example; or by using
77  * swr_alloc_set_opts(), but with the first argument the allocated context.
78  * You must then call swr_init() again.
79  *
80  * The conversion itself is done by repeatedly calling swr_convert().
81  * Note that the samples may get buffered in swr if you provide insufficient
82  * output space or if sample rate conversion is done, which requires "future"
83  * samples. Samples that do not require future input can be retrieved at any
84  * time by using swr_convert() (in_count can be set to 0).
85  * At the end of conversion the resampling buffer can be flushed by calling
86  * swr_convert() with NULL in and 0 in_count.
87  *
88  * The samples used in the conversion process can be managed with the libavutil
89  * @ref lavu_sampmanip "samples manipulation" API, including av_samples_alloc()
90  * function used in the following example.
91  *
92  * The delay between input and output, can at any time be found by using
93  * swr_get_delay().
94  *
95  * The following code demonstrates the conversion loop assuming the parameters
96  * from above and caller-defined functions get_input() and handle_output():
97  * @code
98  * uint8_t **input;
99  * int in_samples;
100  *
101  * while (get_input(&input, &in_samples)) {
102  * uint8_t *output;
103  * int out_samples = av_rescale_rnd(swr_get_delay(swr, 48000) +
104  * in_samples, 44100, 48000, AV_ROUND_UP);
105  * av_samples_alloc(&output, NULL, 2, out_samples,
106  * AV_SAMPLE_FMT_S16, 0);
107  * out_samples = swr_convert(swr, &output, out_samples,
108  * input, in_samples);
109  * handle_output(output, out_samples);
110  * av_freep(&output);
111  * }
112  * @endcode
113  *
114  * When the conversion is finished, the conversion
115  * context and everything associated with it must be freed with swr_free().
116  * A swr_close() function is also available, but it exists mainly for
117  * compatibility with libavresample, and is not required to be called.
118  *
119  * There will be no memory leak if the data is not completely flushed before
120  * swr_free().
121  */
122 
123 #include <stdint.h>
124 #include "libavutil/samplefmt.h"
125 
126 #include "libswresample/version.h"
127 
128 #if LIBSWRESAMPLE_VERSION_MAJOR < 1
129 #define SWR_CH_MAX 32 ///< Maximum number of channels
130 #endif
131 
132 /**
133  * @name Option constants
134  * These constants are used for the @ref avoptions interface for lswr.
135  * @{
136  *
137  */
138 
139 #define SWR_FLAG_RESAMPLE 1 ///< Force resampling even if equal sample rate
140 //TODO use int resample ?
141 //long term TODO can we enable this dynamically?
142 
143 /** Dithering algorithms */
149 
150  SWR_DITHER_NS = 64, ///< not part of API/ABI
158  SWR_DITHER_NB, ///< not part of API/ABI
159 };
160 
161 /** Resampling Engines */
162 enum SwrEngine {
163  SWR_ENGINE_SWR, /**< SW Resampler */
164  SWR_ENGINE_SOXR, /**< SoX Resampler */
165  SWR_ENGINE_NB, ///< not part of API/ABI
166 };
167 
168 /** Resampling Filter Types */
170  SWR_FILTER_TYPE_CUBIC, /**< Cubic */
171  SWR_FILTER_TYPE_BLACKMAN_NUTTALL, /**< Blackman Nuttall Windowed Sinc */
172  SWR_FILTER_TYPE_KAISER, /**< Kaiser Windowed Sinc */
173 };
174 
175 /**
176  * @}
177  */
178 
179 /**
180  * The libswresample context. Unlike libavcodec and libavformat, this structure
181  * is opaque. This means that if you would like to set options, you must use
182  * the @ref avoptions API and cannot directly set values to members of the
183  * structure.
184  */
185 typedef struct SwrContext SwrContext;
186 
187 /**
188  * Get the AVClass for SwrContext. It can be used in combination with
189  * AV_OPT_SEARCH_FAKE_OBJ for examining options.
190  *
191  * @see av_opt_find().
192  * @return the AVClass of SwrContext
193  */
194 const AVClass *swr_get_class(void);
195 
196 /**
197  * @name SwrContext constructor functions
198  * @{
199  */
200 
201 /**
202  * Allocate SwrContext.
203  *
204  * If you use this function you will need to set the parameters (manually or
205  * with swr_alloc_set_opts()) before calling swr_init().
206  *
207  * @see swr_alloc_set_opts(), swr_init(), swr_free()
208  * @return NULL on error, allocated context otherwise
209  */
210 struct SwrContext *swr_alloc(void);
211 
212 /**
213  * Initialize context after user parameters have been set.
214  *
215  * @param[in,out] s Swr context to initialize
216  * @return AVERROR error code in case of failure.
217  */
218 int swr_init(struct SwrContext *s);
219 
220 /**
221  * Check whether an swr context has been initialized or not.
222  *
223  * @param[in] s Swr context to check
224  * @see swr_init()
225  * @return positive if it has been initialized, 0 if not initialized
226  */
227 int swr_is_initialized(struct SwrContext *s);
228 
229 /**
230  * Allocate SwrContext if needed and set/reset common parameters.
231  *
232  * This function does not require s to be allocated with swr_alloc(). On the
233  * other hand, swr_alloc() can use swr_alloc_set_opts() to set the parameters
234  * on the allocated context.
235  *
236  * @param s existing Swr context if available, or NULL if not
237  * @param out_ch_layout output channel layout (AV_CH_LAYOUT_*)
238  * @param out_sample_fmt output sample format (AV_SAMPLE_FMT_*).
239  * @param out_sample_rate output sample rate (frequency in Hz)
240  * @param in_ch_layout input channel layout (AV_CH_LAYOUT_*)
241  * @param in_sample_fmt input sample format (AV_SAMPLE_FMT_*).
242  * @param in_sample_rate input sample rate (frequency in Hz)
243  * @param log_offset logging level offset
244  * @param log_ctx parent logging context, can be NULL
245  *
246  * @see swr_init(), swr_free()
247  * @return NULL on error, allocated context otherwise
248  */
249 struct SwrContext *swr_alloc_set_opts(struct SwrContext *s,
252  int log_offset, void *log_ctx);
253 
254 /**
255  * @}
256  *
257  * @name SwrContext destructor functions
258  * @{
259  */
260 
261 /**
262  * Free the given SwrContext and set the pointer to NULL.
263  *
264  * @param[in] s a pointer to a pointer to Swr context
265  */
266 void swr_free(struct SwrContext **s);
267 
268 /**
269  * Closes the context so that swr_is_initialized() returns 0.
270  *
271  * The context can be brought back to life by running swr_init(),
272  * swr_init() can also be used without swr_close().
273  * This function is mainly provided for simplifying the usecase
274  * where one tries to support libavresample and libswresample.
275  *
276  * @param[in,out] s Swr context to be closed
277  */
278 void swr_close(struct SwrContext *s);
279 
280 /**
281  * @}
282  *
283  * @name Core conversion functions
284  * @{
285  */
286 
287 /** Convert audio.
288  *
289  * in and in_count can be set to 0 to flush the last few samples out at the
290  * end.
291  *
292  * If more input is provided than output space then the input will be buffered.
293  * You can avoid this buffering by providing more output space than input.
294  * Conversion will run directly without copying whenever possible.
295  *
296  * @param s allocated Swr context, with parameters set
297  * @param out output buffers, only the first one need be set in case of packed audio
298  * @param out_count amount of space available for output in samples per channel
299  * @param in input buffers, only the first one need to be set in case of packed audio
300  * @param in_count number of input samples available in one channel
301  *
302  * @return number of samples output per channel, negative value on error
303  */
304 int swr_convert(struct SwrContext *s, uint8_t **out, int out_count,
305  const uint8_t **in , int in_count);
306 
307 /**
308  * Convert the next timestamp from input to output
309  * timestamps are in 1/(in_sample_rate * out_sample_rate) units.
310  *
311  * @note There are 2 slightly differently behaving modes.
312  * @li When automatic timestamp compensation is not used, (min_compensation >= FLT_MAX)
313  * in this case timestamps will be passed through with delays compensated
314  * @li When automatic timestamp compensation is used, (min_compensation < FLT_MAX)
315  * in this case the output timestamps will match output sample numbers.
316  * See ffmpeg-resampler(1) for the two modes of compensation.
317  *
318  * @param s[in] initialized Swr context
319  * @param pts[in] timestamp for the next input sample, INT64_MIN if unknown
320  * @see swr_set_compensation(), swr_drop_output(), and swr_inject_silence() are
321  * function used internally for timestamp compensation.
322  * @return the output timestamp for the next output sample
323  */
324 int64_t swr_next_pts(struct SwrContext *s, int64_t pts);
325 
326 /**
327  * @}
328  *
329  * @name Low-level option setting functions
330  * These functons provide a means to set low-level options that is not possible
331  * with the AVOption API.
332  * @{
333  */
334 
335 /**
336  * Activate resampling compensation ("soft" compensation). This function is
337  * internally called when needed in swr_next_pts().
338  *
339  * @param[in,out] s allocated Swr context. If it is not initialized,
340  * or SWR_FLAG_RESAMPLE is not set, swr_init() is
341  * called with the flag set.
342  * @param[in] sample_delta delta in PTS per sample
343  * @param[in] compensation_distance number of samples to compensate for
344  * @return >= 0 on success, AVERROR error codes if:
345  * @li @c s is NULL,
346  * @li @c compensation_distance is less than 0,
347  * @li @c compensation_distance is 0 but sample_delta is not,
348  * @li compensation unsupported by resampler, or
349  * @li swr_init() fails when called.
350  */
351 int swr_set_compensation(struct SwrContext *s, int sample_delta, int compensation_distance);
352 
353 /**
354  * Set a customized input channel mapping.
355  *
356  * @param[in,out] s allocated Swr context, not yet initialized
357  * @param[in] channel_map customized input channel mapping (array of channel
358  * indexes, -1 for a muted channel)
359  * @return >= 0 on success, or AVERROR error code in case of failure.
360  */
361 int swr_set_channel_mapping(struct SwrContext *s, const int *channel_map);
362 
363 /**
364  * Set a customized remix matrix.
365  *
366  * @param s allocated Swr context, not yet initialized
367  * @param matrix remix coefficients; matrix[i + stride * o] is
368  * the weight of input channel i in output channel o
369  * @param stride offset between lines of the matrix
370  * @return >= 0 on success, or AVERROR error code in case of failure.
371  */
372 int swr_set_matrix(struct SwrContext *s, const double *matrix, int stride);
373 
374 /**
375  * @}
376  *
377  * @name Sample handling functions
378  * @{
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 /**
395  * Injects the specified number of silence samples.
396  *
397  * This function, along with swr_drop_output(), is called by swr_next_pts()
398  * if needed for "hard" compensation.
399  *
400  * @param s allocated Swr context
401  * @param count number of samples to be dropped
402  *
403  * @return >= 0 on success, or a negative AVERROR code on failure
404  */
405 int swr_inject_silence(struct SwrContext *s, int count);
406 
407 /**
408  * Gets the delay the next input sample will experience relative to the next output sample.
409  *
410  * Swresample can buffer data if more input has been provided than available
411  * output space, also converting between sample rates needs a delay.
412  * This function returns the sum of all such delays.
413  * The exact delay is not necessarily an integer value in either input or
414  * output sample rate. Especially when downsampling by a large value, the
415  * output sample rate may be a poor choice to represent the delay, similarly
416  * for upsampling and the input sample rate.
417  *
418  * @param s swr context
419  * @param base timebase in which the returned delay will be:
420  * @li if it's set to 1 the returned delay is in seconds
421  * @li if it's set to 1000 the returned delay is in milliseconds
422  * @li if it's set to the input sample rate then the returned
423  * delay is in input samples
424  * @li if it's set to the output sample rate then the returned
425  * delay is in output samples
426  * @li if it's the least common multiple of in_sample_rate and
427  * out_sample_rate then an exact rounding-free delay will be
428  * returned
429  * @returns the delay in 1 / @c base units.
430  */
431 int64_t swr_get_delay(struct SwrContext *s, int64_t base);
432 
433 /**
434  * @}
435  *
436  * @name Configuration accessors
437  * @{
438  */
439 
440 /**
441  * Return the @ref LIBSWRESAMPLE_VERSION_INT constant.
442  *
443  * This is useful to check if the build-time libswresample has the same version
444  * as the run-time one.
445  *
446  * @returns the unsigned int-typed version
447  */
448 unsigned swresample_version(void);
449 
450 /**
451  * Return the swr build-time configuration.
452  *
453  * @returns the build-time @c ./configure flags
454  */
455 const char *swresample_configuration(void);
456 
457 /**
458  * Return the swr license.
459  *
460  * @returns the license of libswresample, determined at build-time
461  */
462 const char *swresample_license(void);
463 
464 /**
465  * @}
466  * @}
467  */
468 
469 #endif /* SWRESAMPLE_SWRESAMPLE_H */