FFmpeg
af_anlms.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2019 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
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 #include "libavutil/avassert.h"
23 #include "libavutil/common.h"
24 #include "libavutil/float_dsp.h"
25 #include "libavutil/opt.h"
26 
27 #include "audio.h"
28 #include "avfilter.h"
29 #include "formats.h"
30 #include "filters.h"
31 #include "internal.h"
32 
33 enum OutModes {
39 };
40 
41 typedef struct AudioNLMSContext {
42  const AVClass *class;
43 
44  int order;
45  float mu;
46  float eps;
47  float leakage;
49 
55 
57 
60 
61 #define OFFSET(x) offsetof(AudioNLMSContext, x)
62 #define A AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
63 #define AT AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_RUNTIME_PARAM
64 
65 static const AVOption anlms_options[] = {
66  { "order", "set the filter order", OFFSET(order), AV_OPT_TYPE_INT, {.i64=256}, 1, INT16_MAX, A },
67  { "mu", "set the filter mu", OFFSET(mu), AV_OPT_TYPE_FLOAT, {.dbl=0.75}, 0, 2, AT },
68  { "eps", "set the filter eps", OFFSET(eps), AV_OPT_TYPE_FLOAT, {.dbl=1}, 0, 1, AT },
69  { "leakage", "set the filter leakage", OFFSET(leakage), AV_OPT_TYPE_FLOAT, {.dbl=0}, 0, 1, AT },
70  { "out_mode", "set output mode", OFFSET(output_mode), AV_OPT_TYPE_INT, {.i64=OUT_MODE}, 0, NB_OMODES-1, AT, "mode" },
71  { "i", "input", 0, AV_OPT_TYPE_CONST, {.i64=IN_MODE}, 0, 0, AT, "mode" },
72  { "d", "desired", 0, AV_OPT_TYPE_CONST, {.i64=DESIRED_MODE}, 0, 0, AT, "mode" },
73  { "o", "output", 0, AV_OPT_TYPE_CONST, {.i64=OUT_MODE}, 0, 0, AT, "mode" },
74  { "n", "noise", 0, AV_OPT_TYPE_CONST, {.i64=NOISE_MODE}, 0, 0, AT, "mode" },
75  { NULL }
76 };
77 
79 
81 {
84  static const enum AVSampleFormat sample_fmts[] = {
87  };
88  int ret;
89 
90  layouts = ff_all_channel_counts();
91  if (!layouts)
92  return AVERROR(ENOMEM);
93  ret = ff_set_common_channel_layouts(ctx, layouts);
94  if (ret < 0)
95  return ret;
96 
97  formats = ff_make_format_list(sample_fmts);
98  if (!formats)
99  return AVERROR(ENOMEM);
100  ret = ff_set_common_formats(ctx, formats);
101  if (ret < 0)
102  return ret;
103 
104  formats = ff_all_samplerates();
105  if (!formats)
106  return AVERROR(ENOMEM);
107  return ff_set_common_samplerates(ctx, formats);
108 }
109 
110 static float fir_sample(AudioNLMSContext *s, float sample, float *delay,
111  float *coeffs, float *tmp, int *offset)
112 {
113  const int order = s->order;
114  float output;
115 
116  delay[*offset] = sample;
117 
118  memcpy(tmp, coeffs + order - *offset, order * sizeof(float));
119 
120  output = s->fdsp->scalarproduct_float(delay, tmp, s->kernel_size);
121 
122  if (--(*offset) < 0)
123  *offset = order - 1;
124 
125  return output;
126 }
127 
128 static float process_sample(AudioNLMSContext *s, float input, float desired,
129  float *delay, float *coeffs, float *tmp, int *offsetp)
130 {
131  const int order = s->order;
132  const float leakage = s->leakage;
133  const float mu = s->mu;
134  const float a = 1.f - leakage * mu;
135  float sum, output, e, norm, b;
136  int offset = *offsetp;
137 
138  delay[offset + order] = input;
139 
140  output = fir_sample(s, input, delay, coeffs, tmp, offsetp);
141  e = desired - output;
142 
143  sum = s->fdsp->scalarproduct_float(delay, delay, s->kernel_size);
144 
145  norm = s->eps + sum;
146  b = mu * e / norm;
147 
148  memcpy(tmp, delay + offset, order * sizeof(float));
149 
150  s->fdsp->vector_fmul_scalar(coeffs, coeffs, a, s->kernel_size);
151 
152  s->fdsp->vector_fmac_scalar(coeffs, tmp, b, s->kernel_size);
153 
154  memcpy(coeffs + order, coeffs, order * sizeof(float));
155 
156  switch (s->output_mode) {
157  case IN_MODE: output = input; break;
158  case DESIRED_MODE: output = desired; break;
159  case OUT_MODE: /*output = output;*/ break;
160  case NOISE_MODE: output = desired - output; break;
161  }
162  return output;
163 }
164 
165 static int process_channels(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
166 {
167  AudioNLMSContext *s = ctx->priv;
168  AVFrame *out = arg;
169  const int start = (out->channels * jobnr) / nb_jobs;
170  const int end = (out->channels * (jobnr+1)) / nb_jobs;
171 
172  for (int c = start; c < end; c++) {
173  const float *input = (const float *)s->frame[0]->extended_data[c];
174  const float *desired = (const float *)s->frame[1]->extended_data[c];
175  float *delay = (float *)s->delay->extended_data[c];
176  float *coeffs = (float *)s->coeffs->extended_data[c];
177  float *tmp = (float *)s->tmp->extended_data[c];
178  int *offset = (int *)s->offset->extended_data[c];
179  float *output = (float *)out->extended_data[c];
180 
181  for (int n = 0; n < out->nb_samples; n++)
182  output[n] = process_sample(s, input[n], desired[n], delay, coeffs, tmp, offset);
183  }
184 
185  return 0;
186 }
187 
189 {
190  AudioNLMSContext *s = ctx->priv;
191  int i, ret, status;
192  int nb_samples;
193  int64_t pts;
194 
196 
197  nb_samples = FFMIN(ff_inlink_queued_samples(ctx->inputs[0]),
199  for (i = 0; i < ctx->nb_inputs && nb_samples > 0; i++) {
200  if (s->frame[i])
201  continue;
202 
203  if (ff_inlink_check_available_samples(ctx->inputs[i], nb_samples) > 0) {
204  ret = ff_inlink_consume_samples(ctx->inputs[i], nb_samples, nb_samples, &s->frame[i]);
205  if (ret < 0)
206  return ret;
207  }
208  }
209 
210  if (s->frame[0] && s->frame[1]) {
211  AVFrame *out;
212 
213  out = ff_get_audio_buffer(ctx->outputs[0], s->frame[0]->nb_samples);
214  if (!out) {
215  av_frame_free(&s->frame[0]);
216  av_frame_free(&s->frame[1]);
217  return AVERROR(ENOMEM);
218  }
219 
220  ctx->internal->execute(ctx, process_channels, out, NULL, FFMIN(ctx->outputs[0]->channels,
222 
223  out->pts = s->frame[0]->pts;
224 
225  av_frame_free(&s->frame[0]);
226  av_frame_free(&s->frame[1]);
227 
228  ret = ff_filter_frame(ctx->outputs[0], out);
229  if (ret < 0)
230  return ret;
231  }
232 
233  if (!nb_samples) {
234  for (i = 0; i < 2; i++) {
235  if (ff_inlink_acknowledge_status(ctx->inputs[i], &status, &pts)) {
236  ff_outlink_set_status(ctx->outputs[0], status, pts);
237  return 0;
238  }
239  }
240  }
241 
242  if (ff_outlink_frame_wanted(ctx->outputs[0])) {
243  for (i = 0; i < 2; i++) {
244  if (ff_inlink_queued_samples(ctx->inputs[i]) > 0)
245  continue;
247  return 0;
248  }
249  }
250  return 0;
251 }
252 
253 static int config_output(AVFilterLink *outlink)
254 {
255  AVFilterContext *ctx = outlink->src;
256  AudioNLMSContext *s = ctx->priv;
257 
258  s->kernel_size = FFALIGN(s->order, 16);
259 
260  if (!s->offset)
261  s->offset = ff_get_audio_buffer(outlink, 1);
262  if (!s->delay)
263  s->delay = ff_get_audio_buffer(outlink, 2 * s->kernel_size);
264  if (!s->coeffs)
265  s->coeffs = ff_get_audio_buffer(outlink, 2 * s->kernel_size);
266  if (!s->tmp)
267  s->tmp = ff_get_audio_buffer(outlink, s->kernel_size);
268  if (!s->delay || !s->coeffs || !s->offset || !s->tmp)
269  return AVERROR(ENOMEM);
270 
271  return 0;
272 }
273 
275 {
276  AudioNLMSContext *s = ctx->priv;
277 
279  if (!s->fdsp)
280  return AVERROR(ENOMEM);
281 
282  return 0;
283 }
284 
286 {
287  AudioNLMSContext *s = ctx->priv;
288 
289  av_freep(&s->fdsp);
290  av_frame_free(&s->delay);
291  av_frame_free(&s->coeffs);
292  av_frame_free(&s->offset);
293  av_frame_free(&s->tmp);
294 }
295 
296 static const AVFilterPad inputs[] = {
297  {
298  .name = "input",
299  .type = AVMEDIA_TYPE_AUDIO,
300  },
301  {
302  .name = "desired",
303  .type = AVMEDIA_TYPE_AUDIO,
304  },
305  { NULL }
306 };
307 
308 static const AVFilterPad outputs[] = {
309  {
310  .name = "default",
311  .type = AVMEDIA_TYPE_AUDIO,
312  .config_props = config_output,
313  },
314  { NULL }
315 };
316 
318  .name = "anlms",
319  .description = NULL_IF_CONFIG_SMALL("Apply Normalized Least-Mean-Squares algorithm to first audio stream."),
320  .priv_size = sizeof(AudioNLMSContext),
321  .priv_class = &anlms_class,
322  .init = init,
323  .uninit = uninit,
324  .activate = activate,
326  .inputs = inputs,
327  .outputs = outputs,
330 };
float, planar
Definition: samplefmt.h:69
float(* scalarproduct_float)(const float *v1, const float *v2, int len)
Calculate the scalar product of two vectors of floats.
Definition: float_dsp.h:175
#define NULL
Definition: coverity.c:32
int ff_set_common_channel_layouts(AVFilterContext *ctx, AVFilterChannelLayouts *layouts)
A helper for query_formats() which sets all links to the same list of channel layouts/sample rates...
Definition: formats.c:581
This structure describes decoded (raw) audio or video data.
Definition: frame.h:308
AVOption.
Definition: opt.h:248
AVFrame * coeffs
Definition: af_anlms.c:53
static float fir_sample(AudioNLMSContext *s, float sample, float *delay, float *coeffs, float *tmp, int *offset)
Definition: af_anlms.c:110
#define FF_FILTER_FORWARD_STATUS_BACK_ALL(outlink, filter)
Forward the status on an output link to all input links.
Definition: filters.h:212
Main libavfilter public API header.
static int query_formats(AVFilterContext *ctx)
Definition: af_anlms.c:80
static const AVFilterPad inputs[]
Definition: af_anlms.c:296
The reader does not expect b to be semantically here and if the code is changed by maybe adding a a division or other the signedness will almost certainly be mistaken To avoid this confusion a new type was SUINT is the C unsigned type but it holds a signed int to use the same example SUINT a
Definition: undefined.txt:36
static av_cold void uninit(AVFilterContext *ctx)
Definition: af_anlms.c:285
static const AVOption anlms_options[]
Definition: af_anlms.c:65
static int config_output(AVFilterLink *outlink)
Definition: af_anlms.c:253
int ff_inlink_check_available_samples(AVFilterLink *link, unsigned min)
Test if enough samples are available on the link.
Definition: avfilter.c:1477
AVFILTER_DEFINE_CLASS(anlms)
#define sample
static void ff_outlink_set_status(AVFilterLink *link, int status, int64_t pts)
Set the status field of a link from the source filter.
Definition: filters.h:189
void ff_inlink_request_frame(AVFilterLink *link)
Mark that a frame is wanted on the link.
Definition: avfilter.c:1618
AVFilterFormats * ff_make_format_list(const int *fmts)
Create a list of supported formats.
Definition: formats.c:283
const char * name
Pad name.
Definition: internal.h:60
AVFilterLink ** inputs
array of pointers to input links
Definition: avfilter.h:346
void(* vector_fmac_scalar)(float *dst, const float *src, float mul, int len)
Multiply a vector of floats by a scalar float and add to destination vector.
Definition: float_dsp.h:54
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.
#define AT
Definition: af_anlms.c:63
static av_cold int end(AVCodecContext *avctx)
Definition: avrndec.c:90
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
Definition: undefined.txt:32
filter_frame For filters that do not use the this method is called when a frame is pushed to the filter s input It can be called at any time except in a reentrant way If the input frame is enough to produce output
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:401
the definition of that something depends on the semantic of the filter The callback must examine the status of the filter s links and proceed accordingly The status of output links is stored in the status_in and status_out fields and tested by the ff_outlink_frame_wanted() function.If this function returns true
#define FFALIGN(x, a)
Definition: macros.h:48
A filter pad used for either input or output.
Definition: internal.h:54
int ff_inlink_acknowledge_status(AVFilterLink *link, int *rstatus, int64_t *rpts)
Test and acknowledge the change of status on the link.
Definition: avfilter.c:1447
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:600
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
Definition: float_dsp.c:135
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
static int activate(AVFilterContext *ctx)
Definition: af_anlms.c:188
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:203
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:153
int ff_filter_process_command(AVFilterContext *ctx, const char *cmd, const char *arg, char *res, int res_len, int flags)
Generic processing of user supplied commands that are set in the same way as the filter options...
Definition: avfilter.c:885
void * priv
private data for use by the filter
Definition: avfilter.h:353
#define AVFILTER_FLAG_SLICE_THREADS
The filter supports multithreading by splitting frames into multiple parts and processing them concur...
Definition: avfilter.h:116
AVFrame * offset
Definition: af_anlms.c:51
const char * arg
Definition: jacosubdec.c:66
simple assert() macros that are a bit more flexible than ISO C assert().
static av_cold int init(AVFilterContext *ctx)
Definition: af_anlms.c:274
#define b
Definition: input.c:41
int channels
number of audio channels, only used for audio.
Definition: frame.h:614
audio channel layout utility functions
int ff_filter_get_nb_threads(AVFilterContext *ctx)
Get number of threads for current filter instance.
Definition: avfilter.c:800
unsigned nb_inputs
number of input pads
Definition: avfilter.h:347
#define FFMIN(a, b)
Definition: common.h:96
static float process_sample(AudioNLMSContext *s, float input, float desired, float *delay, float *coeffs, float *tmp, int *offsetp)
Definition: af_anlms.c:128
int ff_inlink_queued_samples(AVFilterLink *link)
Definition: avfilter.c:1472
#define A
Definition: af_anlms.c:62
AVFormatContext * ctx
Definition: movenc.c:48
#define s(width, name)
Definition: cbs_vp9.c:257
OutModes
Definition: af_afftdn.c:37
A list of supported channel layouts.
Definition: formats.h:85
#define OFFSET(x)
Definition: af_anlms.c:61
static const AVFilterPad outputs[]
Definition: af_anlms.c:308
AVSampleFormat
Audio sample formats.
Definition: samplefmt.h:58
int ff_inlink_consume_samples(AVFilterLink *link, unsigned min, unsigned max, AVFrame **rframe)
Take samples from the link&#39;s FIFO and update the link&#39;s stats.
Definition: avfilter.c:1511
Describe the class of an AVClass context structure.
Definition: log.h:67
Filter definition.
Definition: avfilter.h:144
they must not be accessed directly The fifo field contains the frames that are queued in the input for processing by the filter The status_in and status_out fields contains the queued status(EOF or error) of the link
void(* vector_fmul_scalar)(float *dst, const float *src, float mul, int len)
Multiply a vector of floats by a scalar float.
Definition: float_dsp.h:85
const char * name
Filter name.
Definition: avfilter.h:148
and forward the test the status of outputs and forward it to the corresponding return FFERROR_NOT_READY If the filters stores internally one or a few frame for some input
static int process_channels(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
Definition: af_anlms.c:165
AVFilterLink ** outputs
array of pointers to output links
Definition: avfilter.h:350
enum MovChannelLayoutTag * layouts
Definition: mov_chan.c:434
AVFilterFormats * ff_all_samplerates(void)
Definition: formats.c:425
static int64_t pts
#define flags(name, subs,...)
Definition: cbs_av1.c:560
AVFilterInternal * internal
An opaque struct for libavfilter internal use.
Definition: avfilter.h:378
common internal and external API header
AVFloatDSPContext * fdsp
Definition: af_anlms.c:58
AVFrame * delay
Definition: af_anlms.c:52
avfilter_execute_func * execute
Definition: internal.h:144
AVFrame * frame[2]
Definition: af_anlms.c:56
AVFilter ff_af_anlms
Definition: af_anlms.c:317
A list of supported formats for one end of a filter link.
Definition: formats.h:64
An instance of a filter.
Definition: avfilter.h:338
static enum AVSampleFormat sample_fmts[]
Definition: adpcmenc.c:723
FILE * out
Definition: movenc.c:54
static int process_command(AVFilterContext *ctx, const char *cmd, const char *args, char *res, int res_len, int flags)
Definition: af_afftdn.c:1374
#define av_freep(p)
AVFrame * tmp
Definition: af_anlms.c:54
formats
Definition: signature.h:48
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
AVFilterChannelLayouts * ff_all_channel_counts(void)
Construct an AVFilterChannelLayouts coding for any channel layout, with known or unknown disposition...
Definition: formats.c:440
uint8_t ** extended_data
pointers to the data planes/channels.
Definition: frame.h:355
int nb_samples
number of audio samples (per channel) described by this frame
Definition: frame.h:374
for(j=16;j >0;--j)
int ff_set_common_samplerates(AVFilterContext *ctx, AVFilterFormats *samplerates)
Definition: formats.c:588
int i
Definition: input.c:406