FFmpeg
asrc_afirsrc.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2020 Paul B Mahol
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 License
8  * 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
14  * GNU Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public License
17  * along with FFmpeg; if not, write to the Free Software Foundation, Inc.,
18  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "libavutil/eval.h"
22 #include "libavutil/opt.h"
23 #include "libavutil/tx.h"
24 #include "audio.h"
25 #include "avfilter.h"
26 #include "internal.h"
27 #include "window_func.h"
28 
29 typedef struct AudioFIRSourceContext {
30  const AVClass *class;
31 
34  char *phase_str;
35  int nb_taps;
38  int win_func;
39 
41  float *freq;
42  float *magnitude;
43  float *phase;
44  int freq_size;
47  int nb_freq;
49  int nb_phase;
50 
51  float *taps;
52  float *win;
53  int64_t pts;
54 
58 
59 #define OFFSET(x) offsetof(AudioFIRSourceContext, x)
60 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
61 
62 static const AVOption afirsrc_options[] = {
63  { "taps", "set number of taps", OFFSET(nb_taps), AV_OPT_TYPE_INT, {.i64=1025}, 9, UINT16_MAX, FLAGS },
64  { "t", "set number of taps", OFFSET(nb_taps), AV_OPT_TYPE_INT, {.i64=1025}, 9, UINT16_MAX, FLAGS },
65  { "frequency", "set frequency points", OFFSET(freq_points_str), AV_OPT_TYPE_STRING, {.str="0 1"}, 0, 0, FLAGS },
66  { "f", "set frequency points", OFFSET(freq_points_str), AV_OPT_TYPE_STRING, {.str="0 1"}, 0, 0, FLAGS },
67  { "magnitude", "set magnitude values", OFFSET(magnitude_str), AV_OPT_TYPE_STRING, {.str="1 1"}, 0, 0, FLAGS },
68  { "m", "set magnitude values", OFFSET(magnitude_str), AV_OPT_TYPE_STRING, {.str="1 1"}, 0, 0, FLAGS },
69  { "phase", "set phase values", OFFSET(phase_str), AV_OPT_TYPE_STRING, {.str="0 0"}, 0, 0, FLAGS },
70  { "p", "set phase values", OFFSET(phase_str), AV_OPT_TYPE_STRING, {.str="0 0"}, 0, 0, FLAGS },
71  { "sample_rate", "set sample rate", OFFSET(sample_rate), AV_OPT_TYPE_INT, {.i64=44100}, 1, INT_MAX, FLAGS },
72  { "r", "set sample rate", OFFSET(sample_rate), AV_OPT_TYPE_INT, {.i64=44100}, 1, INT_MAX, FLAGS },
73  { "nb_samples", "set the number of samples per requested frame", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 1024}, 1, INT_MAX, FLAGS },
74  { "n", "set the number of samples per requested frame", OFFSET(nb_samples), AV_OPT_TYPE_INT, {.i64 = 1024}, 1, INT_MAX, FLAGS },
75  { "win_func", "set window function", OFFSET(win_func), AV_OPT_TYPE_INT, {.i64=WFUNC_BLACKMAN}, 0, NB_WFUNC-1, FLAGS, "win_func" },
76  { "w", "set window function", OFFSET(win_func), AV_OPT_TYPE_INT, {.i64=WFUNC_BLACKMAN}, 0, NB_WFUNC-1, FLAGS, "win_func" },
77  { "rect", "Rectangular", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_RECT}, 0, 0, FLAGS, "win_func" },
78  { "bartlett", "Bartlett", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BARTLETT}, 0, 0, FLAGS, "win_func" },
79  { "hanning", "Hanning", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HANNING}, 0, 0, FLAGS, "win_func" },
80  { "hamming", "Hamming", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_HAMMING}, 0, 0, FLAGS, "win_func" },
81  { "blackman", "Blackman", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BLACKMAN}, 0, 0, FLAGS, "win_func" },
82  { "welch", "Welch", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_WELCH}, 0, 0, FLAGS, "win_func" },
83  { "flattop", "Flat-top", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_FLATTOP}, 0, 0, FLAGS, "win_func" },
84  { "bharris", "Blackman-Harris", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHARRIS}, 0, 0, FLAGS, "win_func" },
85  { "bnuttall", "Blackman-Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BNUTTALL}, 0, 0, FLAGS, "win_func" },
86  { "bhann", "Bartlett-Hann", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BHANN}, 0, 0, FLAGS, "win_func" },
87  { "sine", "Sine", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_SINE}, 0, 0, FLAGS, "win_func" },
88  { "nuttall", "Nuttall", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_NUTTALL}, 0, 0, FLAGS, "win_func" },
89  { "lanczos", "Lanczos", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_LANCZOS}, 0, 0, FLAGS, "win_func" },
90  { "gauss", "Gauss", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_GAUSS}, 0, 0, FLAGS, "win_func" },
91  { "tukey", "Tukey", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_TUKEY}, 0, 0, FLAGS, "win_func" },
92  { "dolph", "Dolph-Chebyshev", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_DOLPH}, 0, 0, FLAGS, "win_func" },
93  { "cauchy", "Cauchy", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_CAUCHY}, 0, 0, FLAGS, "win_func" },
94  { "parzen", "Parzen", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_PARZEN}, 0, 0, FLAGS, "win_func" },
95  { "poisson", "Poisson", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_POISSON}, 0, 0, FLAGS, "win_func" },
96  { "bohman" , "Bohman", 0, AV_OPT_TYPE_CONST, {.i64=WFUNC_BOHMAN}, 0, 0, FLAGS, "win_func" },
97  {NULL}
98 };
99 
100 AVFILTER_DEFINE_CLASS(afirsrc);
101 
103 {
104  AudioFIRSourceContext *s = ctx->priv;
105 
106  if (!(s->nb_taps & 1)) {
107  av_log(s, AV_LOG_WARNING, "Number of taps %d must be odd length.\n", s->nb_taps);
108  s->nb_taps |= 1;
109  }
110 
111  return 0;
112 }
113 
115 {
116  AudioFIRSourceContext *s = ctx->priv;
117 
118  av_freep(&s->win);
119  av_freep(&s->taps);
120  av_freep(&s->freq);
121  av_freep(&s->magnitude);
122  av_freep(&s->phase);
123  av_freep(&s->complexf);
124  av_tx_uninit(&s->tx_ctx);
125 }
126 
128 {
129  AudioFIRSourceContext *s = ctx->priv;
130  static const int64_t chlayouts[] = { AV_CH_LAYOUT_MONO, -1 };
131  int sample_rates[] = { s->sample_rate, -1 };
132  static const enum AVSampleFormat sample_fmts[] = {
135  };
136 
139  int ret;
140 
141  formats = ff_make_format_list(sample_fmts);
142  if (!formats)
143  return AVERROR(ENOMEM);
144  ret = ff_set_common_formats (ctx, formats);
145  if (ret < 0)
146  return ret;
147 
148  layouts = ff_make_format64_list(chlayouts);
149  if (!layouts)
150  return AVERROR(ENOMEM);
151  ret = ff_set_common_channel_layouts(ctx, layouts);
152  if (ret < 0)
153  return ret;
154 
155  formats = ff_make_format_list(sample_rates);
156  if (!formats)
157  return AVERROR(ENOMEM);
158  return ff_set_common_samplerates(ctx, formats);
159 }
160 
161 static int parse_string(char *str, float **items, int *nb_items, int *items_size)
162 {
163  float *new_items;
164  char *tail;
165 
166  new_items = av_fast_realloc(NULL, items_size, 1 * sizeof(float));
167  if (!new_items)
168  return AVERROR(ENOMEM);
169  *items = new_items;
170 
171  tail = str;
172  if (!tail)
173  return AVERROR(EINVAL);
174 
175  do {
176  (*items)[(*nb_items)++] = av_strtod(tail, &tail);
177  new_items = av_fast_realloc(*items, items_size, (*nb_items + 1) * sizeof(float));
178  if (!new_items)
179  return AVERROR(ENOMEM);
180  *items = new_items;
181  if (tail && *tail)
182  tail++;
183  } while (tail && *tail);
184 
185  return 0;
186 }
187 
189  const float *freq,
190  const float *magnitude,
191  const float *phase,
192  int m, int minterp)
193 {
194  for (int i = 0; i < minterp; i++) {
195  for (int j = 1; j < m; j++) {
196  const float x = i / (float)minterp;
197 
198  if (x <= freq[j]) {
199  const float mg = (x - freq[j-1]) / (freq[j] - freq[j-1]) * (magnitude[j] - magnitude[j-1]) + magnitude[j-1];
200  const float ph = (x - freq[j-1]) / (freq[j] - freq[j-1]) * (phase[j] - phase[j-1]) + phase[j-1];
201 
202  complexf[i].re = mg * cosf(ph);
203  complexf[i].im = mg * sinf(ph);
204  break;
205  }
206  }
207  }
208 }
209 
210 static av_cold int config_output(AVFilterLink *outlink)
211 {
212  AVFilterContext *ctx = outlink->src;
213  AudioFIRSourceContext *s = ctx->priv;
214  float overlap, scale = 1.f, compensation;
215  int fft_size, middle, ret;
216 
217  s->nb_freq = s->nb_magnitude = s->nb_phase = 0;
218 
219  ret = parse_string(s->freq_points_str, &s->freq, &s->nb_freq, &s->freq_size);
220  if (ret < 0)
221  return ret;
222 
224  if (ret < 0)
225  return ret;
226 
227  ret = parse_string(s->phase_str, &s->phase, &s->nb_phase, &s->phase_size);
228  if (ret < 0)
229  return ret;
230 
231  if (s->nb_freq != s->nb_magnitude && s->nb_freq != s->nb_phase && s->nb_freq >= 2) {
232  av_log(ctx, AV_LOG_ERROR, "Number of frequencies, magnitudes and phases must be same and >= 2.\n");
233  return AVERROR(EINVAL);
234  }
235 
236  for (int i = 0; i < s->nb_freq; i++) {
237  if (i == 0 && s->freq[i] != 0.f) {
238  av_log(ctx, AV_LOG_ERROR, "First frequency must be 0.\n");
239  return AVERROR(EINVAL);
240  }
241 
242  if (i == s->nb_freq - 1 && s->freq[i] != 1.f) {
243  av_log(ctx, AV_LOG_ERROR, "Last frequency must be 1.\n");
244  return AVERROR(EINVAL);
245  }
246 
247  if (i && s->freq[i] < s->freq[i-1]) {
248  av_log(ctx, AV_LOG_ERROR, "Frequencies must be in increasing order.\n");
249  return AVERROR(EINVAL);
250  }
251  }
252 
253  fft_size = 1 << (av_log2(s->nb_taps) + 1);
254  s->complexf = av_calloc(fft_size * 2, sizeof(*s->complexf));
255  if (!s->complexf)
256  return AVERROR(ENOMEM);
257 
258  ret = av_tx_init(&s->tx_ctx, &s->tx_fn, AV_TX_FLOAT_FFT, 1, fft_size, &scale, 0);
259  if (ret < 0)
260  return ret;
261 
262  s->taps = av_calloc(s->nb_taps, sizeof(*s->taps));
263  if (!s->taps)
264  return AVERROR(ENOMEM);
265 
266  s->win = av_calloc(s->nb_taps, sizeof(*s->win));
267  if (!s->win)
268  return AVERROR(ENOMEM);
269 
270  generate_window_func(s->win, s->nb_taps, s->win_func, &overlap);
271 
272  lininterp(s->complexf, s->freq, s->magnitude, s->phase, s->nb_freq, fft_size / 2);
273 
274  s->tx_fn(s->tx_ctx, s->complexf + fft_size, s->complexf, sizeof(float));
275 
276  compensation = 2.f / fft_size;
277  middle = s->nb_taps / 2;
278 
279  for (int i = 0; i <= middle; i++) {
280  s->taps[ i] = s->complexf[fft_size + middle - i].re * compensation * s->win[i];
281  s->taps[middle + i] = s->complexf[fft_size + i].re * compensation * s->win[middle + i];
282  }
283 
284  s->pts = 0;
285 
286  return 0;
287 }
288 
289 static int request_frame(AVFilterLink *outlink)
290 {
291  AVFilterContext *ctx = outlink->src;
292  AudioFIRSourceContext *s = ctx->priv;
293  AVFrame *frame;
294  int nb_samples;
295 
296  nb_samples = FFMIN(s->nb_samples, s->nb_taps - s->pts);
297  if (!nb_samples)
298  return AVERROR_EOF;
299 
300  if (!(frame = ff_get_audio_buffer(outlink, nb_samples)))
301  return AVERROR(ENOMEM);
302 
303  memcpy(frame->data[0], s->taps + s->pts, nb_samples * sizeof(float));
304 
305  frame->pts = s->pts;
306  s->pts += nb_samples;
307  return ff_filter_frame(outlink, frame);
308 }
309 
310 static const AVFilterPad afirsrc_outputs[] = {
311  {
312  .name = "default",
313  .type = AVMEDIA_TYPE_AUDIO,
314  .request_frame = request_frame,
315  .config_props = config_output,
316  },
317  { NULL }
318 };
319 
321  .name = "afirsrc",
322  .description = NULL_IF_CONFIG_SMALL("Generate a FIR coefficients audio stream."),
323  .query_formats = query_formats,
324  .init = init,
325  .uninit = uninit,
326  .priv_size = sizeof(AudioFIRSourceContext),
327  .inputs = NULL,
328  .outputs = afirsrc_outputs,
329  .priv_class = &afirsrc_class,
330 };
av_cold void av_tx_uninit(AVTXContext **ctx)
Frees a context and sets ctx to NULL, does nothing when ctx == NULL.
Definition: tx.c:110
#define NULL
Definition: coverity.c:32
static av_cold int config_output(AVFilterLink *outlink)
Definition: asrc_afirsrc.c:210
This structure describes decoded (raw) audio or video data.
Definition: frame.h:308
AVOption.
Definition: opt.h:248
AVFilterChannelLayouts * ff_make_format64_list(const int64_t *fmts)
Definition: formats.c:296
#define FLAGS
Definition: asrc_afirsrc.c:60
static void lininterp(AVComplexFloat *complexf, const float *freq, const float *magnitude, const float *phase, int m, int minterp)
Definition: asrc_afirsrc.c:188
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:200
Main libavfilter public API header.
int av_log2(unsigned v)
Definition: intmath.c:26
int ff_set_common_channel_layouts(AVFilterContext *ctx, AVFilterChannelLayouts *channel_layouts)
A helper for query_formats() which sets all links to the same list of channel layouts/sample rates...
Definition: formats.c:569
static void generate_window_func(float *lut, int N, int win_func, float *overlap)
Definition: window_func.h:36
void * av_calloc(size_t nmemb, size_t size)
Non-inlined equivalent of av_mallocz_array().
Definition: mem.c:245
AVFilterFormats * ff_make_format_list(const int *fmts)
Create a list of supported formats.
Definition: formats.c:287
static av_cold void uninit(AVFilterContext *ctx)
Definition: asrc_afirsrc.c:114
AVFilter ff_asrc_afirsrc
Definition: asrc_afirsrc.c:320
const char * name
Pad name.
Definition: internal.h:60
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1091
#define av_cold
Definition: attributes.h:88
AVOptions.
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:401
#define cosf(x)
Definition: libm.h:78
float re
Definition: tx.h:28
#define AVERROR_EOF
End of file.
Definition: error.h:55
#define av_log(a,...)
A filter pad used for either input or output.
Definition: internal.h:54
static int request_frame(AVFilterLink *outlink)
Definition: asrc_afirsrc.c:289
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:194
int ff_set_common_formats(AVFilterContext *ctx, AVFilterFormats *formats)
A helper for query_formats() which sets all links to the same list of formats.
Definition: formats.c:588
void(* av_tx_fn)(AVTXContext *s, void *out, void *in, ptrdiff_t stride)
Function pointer to a function to perform the transform.
Definition: tx.h:92
AVFrame * ff_get_audio_buffer(AVFilterLink *link, int nb_samples)
Request an audio samples buffer with a specific set of permissions.
Definition: audio.c:86
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:153
void * priv
private data for use by the filter
Definition: avfilter.h:354
#define OFFSET(x)
Definition: asrc_afirsrc.c:59
Standard complex to complex FFT with sample data type AVComplexFloat.
Definition: tx.h:45
static int parse_string(char *str, float **items, int *nb_items, int *items_size)
Definition: asrc_afirsrc.c:161
#define FFMIN(a, b)
Definition: common.h:96
AVFormatContext * ctx
Definition: movenc.c:48
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several the filter must be ready for frames arriving randomly on any input any filter with several inputs will most likely require some kind of queuing mechanism It is perfectly acceptable to have a limited queue and to drop frames when the inputs are too unbalanced request_frame For filters that do not use the this method is called when a frame is wanted on an output For a it should directly call filter_frame on the corresponding output For a if there are queued frames already one of these frames should be pushed If the filter should request a frame on one of its repeatedly until at least one frame has been pushed Return or at least make progress towards producing a frame
#define s(width, name)
Definition: cbs_vp9.c:257
void * av_fast_realloc(void *ptr, unsigned int *size, size_t min_size)
Reallocate the given buffer if it is not large enough, otherwise do nothing.
Definition: mem.c:478
static const AVFilterPad outputs[]
Definition: af_acontrast.c:203
A list of supported channel layouts.
Definition: formats.h:86
double av_strtod(const char *numstr, char **tail)
Parse the string in numstr and return its value as a double.
Definition: eval.c:106
#define sinf(x)
Definition: libm.h:419
float im
Definition: tx.h:28
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several inputs
Describe the class of an AVClass context structure.
Definition: log.h:67
Filter definition.
Definition: avfilter.h:145
const char * name
Filter name.
Definition: avfilter.h:149
av_cold int av_tx_init(AVTXContext **ctx, av_tx_fn *tx, enum AVTXType type, int inv, int len, const void *scale, uint64_t flags)
Initialize a transform context with the given configuration Currently power of two lengths from 2 to ...
Definition: tx.c:123
enum MovChannelLayoutTag * layouts
Definition: mov_chan.c:434
AVFILTER_DEFINE_CLASS(afirsrc)
AVTXContext * tx_ctx
Definition: asrc_afirsrc.c:55
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:322
sample_rates
#define mg
static const AVOption afirsrc_options[]
Definition: asrc_afirsrc.c:62
A list of supported formats for one end of a filter link.
Definition: formats.h:65
static const AVFilterPad afirsrc_outputs[]
Definition: asrc_afirsrc.c:310
static av_cold int query_formats(AVFilterContext *ctx)
Definition: asrc_afirsrc.c:127
An instance of a filter.
Definition: avfilter.h:339
static enum AVSampleFormat sample_fmts[]
Definition: adpcmenc.c:836
#define av_freep(p)
formats
Definition: signature.h:48
AVComplexFloat * complexf
Definition: asrc_afirsrc.c:40
internal API functions
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later.That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another.Frame references ownership and permissions
#define AV_CH_LAYOUT_MONO
int ff_set_common_samplerates(AVFilterContext *ctx, AVFilterFormats *samplerates)
Definition: formats.c:576
int i
Definition: input.c:407
static av_cold int init(AVFilterContext *ctx)
Definition: asrc_afirsrc.c:102
simple arithmetic expression evaluator