Go to most recent revision | Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
4349 | Serge | 1 | /* |
2 | * Copyright (c) 2012 Justin Ruggles |
||
3 | * |
||
4 | * This file is part of FFmpeg. |
||
5 | * |
||
6 | * FFmpeg 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 | * FFmpeg 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 FFmpeg; if not, write to the Free Software |
||
18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||
19 | */ |
||
20 | |||
21 | #ifndef AVRESAMPLE_AVRESAMPLE_H |
||
22 | #define AVRESAMPLE_AVRESAMPLE_H |
||
23 | |||
24 | /** |
||
25 | * @file |
||
26 | * @ingroup lavr |
||
27 | * external API header |
||
28 | */ |
||
29 | |||
30 | /** |
||
31 | * @defgroup lavr Libavresample |
||
32 | * @{ |
||
33 | * |
||
34 | * Libavresample (lavr) is a library that handles audio resampling, sample |
||
35 | * format conversion and mixing. |
||
36 | * |
||
37 | * Interaction with lavr is done through AVAudioResampleContext, which is |
||
38 | * allocated with avresample_alloc_context(). It is opaque, so all parameters |
||
39 | * must be set with the @ref avoptions API. |
||
40 | * |
||
41 | * For example the following code will setup conversion from planar float sample |
||
42 | * format to interleaved signed 16-bit integer, downsampling from 48kHz to |
||
43 | * 44.1kHz and downmixing from 5.1 channels to stereo (using the default mixing |
||
44 | * matrix): |
||
45 | * @code |
||
46 | * AVAudioResampleContext *avr = avresample_alloc_context(); |
||
47 | * av_opt_set_int(avr, "in_channel_layout", AV_CH_LAYOUT_5POINT1, 0); |
||
48 | * av_opt_set_int(avr, "out_channel_layout", AV_CH_LAYOUT_STEREO, 0); |
||
49 | * av_opt_set_int(avr, "in_sample_rate", 48000, 0); |
||
50 | * av_opt_set_int(avr, "out_sample_rate", 44100, 0); |
||
51 | * av_opt_set_int(avr, "in_sample_fmt", AV_SAMPLE_FMT_FLTP, 0); |
||
52 | * av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_S16, 0); |
||
53 | * @endcode |
||
54 | * |
||
55 | * Once the context is initialized, it must be opened with avresample_open(). If |
||
56 | * you need to change the conversion parameters, you must close the context with |
||
57 | * avresample_close(), change the parameters as described above, then reopen it |
||
58 | * again. |
||
59 | * |
||
60 | * The conversion itself is done by repeatedly calling avresample_convert(). |
||
61 | * Note that the samples may get buffered in two places in lavr. The first one |
||
62 | * is the output FIFO, where the samples end up if the output buffer is not |
||
63 | * large enough. The data stored in there may be retrieved at any time with |
||
64 | * avresample_read(). The second place is the resampling delay buffer, |
||
65 | * applicable only when resampling is done. The samples in it require more input |
||
66 | * before they can be processed. Their current amount is returned by |
||
67 | * avresample_get_delay(). At the end of conversion the resampling buffer can be |
||
68 | * flushed by calling avresample_convert() with NULL input. |
||
69 | * |
||
70 | * The following code demonstrates the conversion loop assuming the parameters |
||
71 | * from above and caller-defined functions get_input() and handle_output(): |
||
72 | * @code |
||
73 | * uint8_t **input; |
||
74 | * int in_linesize, in_samples; |
||
75 | * |
||
76 | * while (get_input(&input, &in_linesize, &in_samples)) { |
||
77 | * uint8_t *output |
||
78 | * int out_linesize; |
||
79 | * int out_samples = avresample_available(avr) + |
||
80 | * av_rescale_rnd(avresample_get_delay(avr) + |
||
81 | * in_samples, 44100, 48000, AV_ROUND_UP); |
||
82 | * av_samples_alloc(&output, &out_linesize, 2, out_samples, |
||
83 | * AV_SAMPLE_FMT_S16, 0); |
||
84 | * out_samples = avresample_convert(avr, &output, out_linesize, out_samples, |
||
85 | * input, in_linesize, in_samples); |
||
86 | * handle_output(output, out_linesize, out_samples); |
||
87 | * av_freep(&output); |
||
88 | * } |
||
89 | * @endcode |
||
90 | * |
||
91 | * When the conversion is finished and the FIFOs are flushed if required, the |
||
92 | * conversion context and everything associated with it must be freed with |
||
93 | * avresample_free(). |
||
94 | */ |
||
95 | |||
96 | #include "libavutil/avutil.h" |
||
97 | #include "libavutil/channel_layout.h" |
||
98 | #include "libavutil/dict.h" |
||
99 | #include "libavutil/log.h" |
||
100 | |||
101 | #include "libavresample/version.h" |
||
102 | |||
103 | #define AVRESAMPLE_MAX_CHANNELS 32 |
||
104 | |||
105 | typedef struct AVAudioResampleContext AVAudioResampleContext; |
||
106 | |||
107 | /** Mixing Coefficient Types */ |
||
108 | enum AVMixCoeffType { |
||
109 | AV_MIX_COEFF_TYPE_Q8, /** 16-bit 8.8 fixed-point */ |
||
110 | AV_MIX_COEFF_TYPE_Q15, /** 32-bit 17.15 fixed-point */ |
||
111 | AV_MIX_COEFF_TYPE_FLT, /** floating-point */ |
||
112 | AV_MIX_COEFF_TYPE_NB, /** Number of coeff types. Not part of ABI */ |
||
113 | }; |
||
114 | |||
115 | /** Resampling Filter Types */ |
||
116 | enum AVResampleFilterType { |
||
117 | AV_RESAMPLE_FILTER_TYPE_CUBIC, /**< Cubic */ |
||
118 | AV_RESAMPLE_FILTER_TYPE_BLACKMAN_NUTTALL, /**< Blackman Nuttall Windowed Sinc */ |
||
119 | AV_RESAMPLE_FILTER_TYPE_KAISER, /**< Kaiser Windowed Sinc */ |
||
120 | }; |
||
121 | |||
122 | enum AVResampleDitherMethod { |
||
123 | AV_RESAMPLE_DITHER_NONE, /**< Do not use dithering */ |
||
124 | AV_RESAMPLE_DITHER_RECTANGULAR, /**< Rectangular Dither */ |
||
125 | AV_RESAMPLE_DITHER_TRIANGULAR, /**< Triangular Dither*/ |
||
126 | AV_RESAMPLE_DITHER_TRIANGULAR_HP, /**< Triangular Dither with High Pass */ |
||
127 | AV_RESAMPLE_DITHER_TRIANGULAR_NS, /**< Triangular Dither with Noise Shaping */ |
||
128 | AV_RESAMPLE_DITHER_NB, /**< Number of dither types. Not part of ABI. */ |
||
129 | }; |
||
130 | |||
131 | /** |
||
132 | * Return the LIBAVRESAMPLE_VERSION_INT constant. |
||
133 | */ |
||
134 | unsigned avresample_version(void); |
||
135 | |||
136 | /** |
||
137 | * Return the libavresample build-time configuration. |
||
138 | * @return configure string |
||
139 | */ |
||
140 | const char *avresample_configuration(void); |
||
141 | |||
142 | /** |
||
143 | * Return the libavresample license. |
||
144 | */ |
||
145 | const char *avresample_license(void); |
||
146 | |||
147 | /** |
||
148 | * Get the AVClass for AVAudioResampleContext. |
||
149 | * |
||
150 | * Can be used in combination with AV_OPT_SEARCH_FAKE_OBJ for examining options |
||
151 | * without allocating a context. |
||
152 | * |
||
153 | * @see av_opt_find(). |
||
154 | * |
||
155 | * @return AVClass for AVAudioResampleContext |
||
156 | */ |
||
157 | const AVClass *avresample_get_class(void); |
||
158 | |||
159 | /** |
||
160 | * Allocate AVAudioResampleContext and set options. |
||
161 | * |
||
162 | * @return allocated audio resample context, or NULL on failure |
||
163 | */ |
||
164 | AVAudioResampleContext *avresample_alloc_context(void); |
||
165 | |||
166 | /** |
||
167 | * Initialize AVAudioResampleContext. |
||
168 | * |
||
169 | * @param avr audio resample context |
||
170 | * @return 0 on success, negative AVERROR code on failure |
||
171 | */ |
||
172 | int avresample_open(AVAudioResampleContext *avr); |
||
173 | |||
174 | /** |
||
175 | * Close AVAudioResampleContext. |
||
176 | * |
||
177 | * This closes the context, but it does not change the parameters. The context |
||
178 | * can be reopened with avresample_open(). It does, however, clear the output |
||
179 | * FIFO and any remaining leftover samples in the resampling delay buffer. If |
||
180 | * there was a custom matrix being used, that is also cleared. |
||
181 | * |
||
182 | * @see avresample_convert() |
||
183 | * @see avresample_set_matrix() |
||
184 | * |
||
185 | * @param avr audio resample context |
||
186 | */ |
||
187 | void avresample_close(AVAudioResampleContext *avr); |
||
188 | |||
189 | /** |
||
190 | * Free AVAudioResampleContext and associated AVOption values. |
||
191 | * |
||
192 | * This also calls avresample_close() before freeing. |
||
193 | * |
||
194 | * @param avr audio resample context |
||
195 | */ |
||
196 | void avresample_free(AVAudioResampleContext **avr); |
||
197 | |||
198 | /** |
||
199 | * Generate a channel mixing matrix. |
||
200 | * |
||
201 | * This function is the one used internally by libavresample for building the |
||
202 | * default mixing matrix. It is made public just as a utility function for |
||
203 | * building custom matrices. |
||
204 | * |
||
205 | * @param in_layout input channel layout |
||
206 | * @param out_layout output channel layout |
||
207 | * @param center_mix_level mix level for the center channel |
||
208 | * @param surround_mix_level mix level for the surround channel(s) |
||
209 | * @param lfe_mix_level mix level for the low-frequency effects channel |
||
210 | * @param normalize if 1, coefficients will be normalized to prevent |
||
211 | * overflow. if 0, coefficients will not be |
||
212 | * normalized. |
||
213 | * @param[out] matrix mixing coefficients; matrix[i + stride * o] is |
||
214 | * the weight of input channel i in output channel o. |
||
215 | * @param stride distance between adjacent input channels in the |
||
216 | * matrix array |
||
217 | * @param matrix_encoding matrixed stereo downmix mode (e.g. dplii) |
||
218 | * @return 0 on success, negative AVERROR code on failure |
||
219 | */ |
||
220 | int avresample_build_matrix(uint64_t in_layout, uint64_t out_layout, |
||
221 | double center_mix_level, double surround_mix_level, |
||
222 | double lfe_mix_level, int normalize, double *matrix, |
||
223 | int stride, enum AVMatrixEncoding matrix_encoding); |
||
224 | |||
225 | /** |
||
226 | * Get the current channel mixing matrix. |
||
227 | * |
||
228 | * If no custom matrix has been previously set or the AVAudioResampleContext is |
||
229 | * not open, an error is returned. |
||
230 | * |
||
231 | * @param avr audio resample context |
||
232 | * @param matrix mixing coefficients; matrix[i + stride * o] is the weight of |
||
233 | * input channel i in output channel o. |
||
234 | * @param stride distance between adjacent input channels in the matrix array |
||
235 | * @return 0 on success, negative AVERROR code on failure |
||
236 | */ |
||
237 | int avresample_get_matrix(AVAudioResampleContext *avr, double *matrix, |
||
238 | int stride); |
||
239 | |||
240 | /** |
||
241 | * Set channel mixing matrix. |
||
242 | * |
||
243 | * Allows for setting a custom mixing matrix, overriding the default matrix |
||
244 | * generated internally during avresample_open(). This function can be called |
||
245 | * anytime on an allocated context, either before or after calling |
||
246 | * avresample_open(), as long as the channel layouts have been set. |
||
247 | * avresample_convert() always uses the current matrix. |
||
248 | * Calling avresample_close() on the context will clear the current matrix. |
||
249 | * |
||
250 | * @see avresample_close() |
||
251 | * |
||
252 | * @param avr audio resample context |
||
253 | * @param matrix mixing coefficients; matrix[i + stride * o] is the weight of |
||
254 | * input channel i in output channel o. |
||
255 | * @param stride distance between adjacent input channels in the matrix array |
||
256 | * @return 0 on success, negative AVERROR code on failure |
||
257 | */ |
||
258 | int avresample_set_matrix(AVAudioResampleContext *avr, const double *matrix, |
||
259 | int stride); |
||
260 | |||
261 | /** |
||
262 | * Set a customized input channel mapping. |
||
263 | * |
||
264 | * This function can only be called when the allocated context is not open. |
||
265 | * Also, the input channel layout must have already been set. |
||
266 | * |
||
267 | * Calling avresample_close() on the context will clear the channel mapping. |
||
268 | * |
||
269 | * The map for each input channel specifies the channel index in the source to |
||
270 | * use for that particular channel, or -1 to mute the channel. Source channels |
||
271 | * can be duplicated by using the same index for multiple input channels. |
||
272 | * |
||
273 | * Examples: |
||
274 | * |
||
275 | * Reordering 5.1 AAC order (C,L,R,Ls,Rs,LFE) to FFmpeg order (L,R,C,LFE,Ls,Rs): |
||
276 | * { 1, 2, 0, 5, 3, 4 } |
||
277 | * |
||
278 | * Muting the 3rd channel in 4-channel input: |
||
279 | * { 0, 1, -1, 3 } |
||
280 | * |
||
281 | * Duplicating the left channel of stereo input: |
||
282 | * { 0, 0 } |
||
283 | * |
||
284 | * @param avr audio resample context |
||
285 | * @param channel_map customized input channel mapping |
||
286 | * @return 0 on success, negative AVERROR code on failure |
||
287 | */ |
||
288 | int avresample_set_channel_mapping(AVAudioResampleContext *avr, |
||
289 | const int *channel_map); |
||
290 | |||
291 | /** |
||
292 | * Set compensation for resampling. |
||
293 | * |
||
294 | * This can be called anytime after avresample_open(). If resampling is not |
||
295 | * automatically enabled because of a sample rate conversion, the |
||
296 | * "force_resampling" option must have been set to 1 when opening the context |
||
297 | * in order to use resampling compensation. |
||
298 | * |
||
299 | * @param avr audio resample context |
||
300 | * @param sample_delta compensation delta, in samples |
||
301 | * @param compensation_distance compensation distance, in samples |
||
302 | * @return 0 on success, negative AVERROR code on failure |
||
303 | */ |
||
304 | int avresample_set_compensation(AVAudioResampleContext *avr, int sample_delta, |
||
305 | int compensation_distance); |
||
306 | |||
307 | /** |
||
308 | * Convert input samples and write them to the output FIFO. |
||
309 | * |
||
310 | * The upper bound on the number of output samples is given by |
||
311 | * avresample_available() + (avresample_get_delay() + number of input samples) * |
||
312 | * output sample rate / input sample rate. |
||
313 | * |
||
314 | * The output data can be NULL or have fewer allocated samples than required. |
||
315 | * In this case, any remaining samples not written to the output will be added |
||
316 | * to an internal FIFO buffer, to be returned at the next call to this function |
||
317 | * or to avresample_read(). |
||
318 | * |
||
319 | * If converting sample rate, there may be data remaining in the internal |
||
320 | * resampling delay buffer. avresample_get_delay() tells the number of remaining |
||
321 | * samples. To get this data as output, call avresample_convert() with NULL |
||
322 | * input. |
||
323 | * |
||
324 | * At the end of the conversion process, there may be data remaining in the |
||
325 | * internal FIFO buffer. avresample_available() tells the number of remaining |
||
326 | * samples. To get this data as output, either call avresample_convert() with |
||
327 | * NULL input or call avresample_read(). |
||
328 | * |
||
329 | * @see avresample_available() |
||
330 | * @see avresample_read() |
||
331 | * @see avresample_get_delay() |
||
332 | * |
||
333 | * @param avr audio resample context |
||
334 | * @param output output data pointers |
||
335 | * @param out_plane_size output plane size, in bytes. |
||
336 | * This can be 0 if unknown, but that will lead to |
||
337 | * optimized functions not being used directly on the |
||
338 | * output, which could slow down some conversions. |
||
339 | * @param out_samples maximum number of samples that the output buffer can hold |
||
340 | * @param input input data pointers |
||
341 | * @param in_plane_size input plane size, in bytes |
||
342 | * This can be 0 if unknown, but that will lead to |
||
343 | * optimized functions not being used directly on the |
||
344 | * input, which could slow down some conversions. |
||
345 | * @param in_samples number of input samples to convert |
||
346 | * @return number of samples written to the output buffer, |
||
347 | * not including converted samples added to the internal |
||
348 | * output FIFO |
||
349 | */ |
||
350 | int avresample_convert(AVAudioResampleContext *avr, uint8_t **output, |
||
351 | int out_plane_size, int out_samples, uint8_t **input, |
||
352 | int in_plane_size, int in_samples); |
||
353 | |||
354 | /** |
||
355 | * Return the number of samples currently in the resampling delay buffer. |
||
356 | * |
||
357 | * When resampling, there may be a delay between the input and output. Any |
||
358 | * unconverted samples in each call are stored internally in a delay buffer. |
||
359 | * This function allows the user to determine the current number of samples in |
||
360 | * the delay buffer, which can be useful for synchronization. |
||
361 | * |
||
362 | * @see avresample_convert() |
||
363 | * |
||
364 | * @param avr audio resample context |
||
365 | * @return number of samples currently in the resampling delay buffer |
||
366 | */ |
||
367 | int avresample_get_delay(AVAudioResampleContext *avr); |
||
368 | |||
369 | /** |
||
370 | * Return the number of available samples in the output FIFO. |
||
371 | * |
||
372 | * During conversion, if the user does not specify an output buffer or |
||
373 | * specifies an output buffer that is smaller than what is needed, remaining |
||
374 | * samples that are not written to the output are stored to an internal FIFO |
||
375 | * buffer. The samples in the FIFO can be read with avresample_read() or |
||
376 | * avresample_convert(). |
||
377 | * |
||
378 | * @see avresample_read() |
||
379 | * @see avresample_convert() |
||
380 | * |
||
381 | * @param avr audio resample context |
||
382 | * @return number of samples available for reading |
||
383 | */ |
||
384 | int avresample_available(AVAudioResampleContext *avr); |
||
385 | |||
386 | /** |
||
387 | * Read samples from the output FIFO. |
||
388 | * |
||
389 | * During conversion, if the user does not specify an output buffer or |
||
390 | * specifies an output buffer that is smaller than what is needed, remaining |
||
391 | * samples that are not written to the output are stored to an internal FIFO |
||
392 | * buffer. This function can be used to read samples from that internal FIFO. |
||
393 | * |
||
394 | * @see avresample_available() |
||
395 | * @see avresample_convert() |
||
396 | * |
||
397 | * @param avr audio resample context |
||
398 | * @param output output data pointers. May be NULL, in which case |
||
399 | * nb_samples of data is discarded from output FIFO. |
||
400 | * @param nb_samples number of samples to read from the FIFO |
||
401 | * @return the number of samples written to output |
||
402 | */ |
||
403 | int avresample_read(AVAudioResampleContext *avr, uint8_t **output, int nb_samples); |
||
404 | |||
405 | /** |
||
406 | * @} |
||
407 | */ |
||
408 | |||
409 | #endif /* AVRESAMPLE_AVRESAMPLE_H */>>>>>>>>> |