Go to the documentation of this file.
36 #define FREQUENCY_DOMAIN 1
101 char *
arg, *tokenizer, *p;
102 uint64_t used_channels = 0;
106 uint64_t out_channel;
113 if (used_channels & out_channel) {
117 used_channels |= out_channel;
118 s->mapping[
s->nb_irs] = out_channel;
123 s->nb_hrir_inputs = 1;
125 s->nb_hrir_inputs =
s->nb_irs;
146 int *write = &
td->write[jobnr];
147 const float *
const ir =
td->ir[jobnr];
148 int *n_clippings = &
td->n_clippings[jobnr];
149 float *ringbuffer =
td->ringbuffer[jobnr];
150 float *temp_src =
td->temp_src[jobnr];
151 const int ir_len =
s->ir_len;
152 const int air_len =
s->air_len;
153 const float *
src = (
const float *)in->
data[0];
154 float *dst = (
float *)
out->data[0];
155 const int in_channels = in->
channels;
156 const int buffer_length =
s->buffer_length;
157 const uint32_t modulo = (uint32_t)buffer_length - 1;
164 for (l = 0; l < in_channels; l++) {
165 buffer[l] = ringbuffer + l * buffer_length;
169 const float *cur_ir = ir;
172 for (l = 0; l < in_channels; l++) {
176 for (l = 0; l < in_channels; cur_ir += air_len, l++) {
177 const float *
const bptr =
buffer[l];
179 if (l ==
s->lfe_channel) {
180 *dst += *(
buffer[
s->lfe_channel] + wr) *
s->gain_lfe;
184 read = (wr - (ir_len - 1)) & modulo;
186 if (read + ir_len < buffer_length) {
187 memcpy(temp_src, bptr + read, ir_len *
sizeof(*temp_src));
189 int len =
FFMIN(air_len - (read % ir_len), buffer_length - read);
191 memcpy(temp_src, bptr + read,
len *
sizeof(*temp_src));
192 memcpy(temp_src +
len, bptr, (air_len -
len) *
sizeof(*temp_src));
195 dst[0] +=
s->scalarproduct_float(cur_ir, temp_src,
FFALIGN(ir_len, 32));
198 if (
fabsf(dst[0]) > 1)
203 wr = (wr + 1) & modulo;
217 int *write = &
td->write[jobnr];
219 int *n_clippings = &
td->n_clippings[jobnr];
220 float *ringbuffer =
td->ringbuffer[jobnr];
221 const int ir_len =
s->ir_len;
222 const float *
src = (
const float *)in->
data[0];
223 float *dst = (
float *)
out->data[0];
224 const int in_channels = in->
channels;
225 const int buffer_length =
s->buffer_length;
226 const uint32_t modulo = (uint32_t)buffer_length - 1;
234 const int n_fft =
s->n_fft;
235 const float fft_scale = 1.0f /
s->n_fft;
244 for (j = 0; j < n_read; j++) {
245 dst[2 * j] = ringbuffer[wr];
246 ringbuffer[wr] = 0.0;
247 wr = (wr + 1) & modulo;
256 for (
i = 0;
i < in_channels;
i++) {
257 if (
i ==
s->lfe_channel) {
259 dst[2 * j] +=
src[
i + j * in_channels] *
s->gain_lfe;
265 hrtf_offset = hrtf +
offset;
270 fft_in[j].
re =
src[j * in_channels +
i];
273 tx_fn(fft, fft_out, fft_in,
sizeof(
float));
275 for (j = 0; j < n_fft; j++) {
277 const float re = fft_out[j].
re;
278 const float im = fft_out[j].
im;
280 fft_acc[j].
re +=
re * hcomplex->
re -
im * hcomplex->
im;
281 fft_acc[j].
im +=
re * hcomplex->
im +
im * hcomplex->
re;
285 itx_fn(ifft, fft_out, fft_acc,
sizeof(
float));
288 dst[2 * j] += fft_out[j].
re * fft_scale;
289 if (
fabsf(dst[2 * j]) > 1)
293 for (j = 0; j < ir_len - 1; j++) {
294 int write_pos = (wr + j) & modulo;
296 *(ringbuffer + write_pos) += fft_out[in->
nb_samples + j].
re * fft_scale;
308 int ir_len, max_ir_len;
312 if (ir_len > max_ir_len) {
316 s->hrir_in[input_number].ir_len = ir_len;
317 s->ir_len =
FFMAX(ir_len,
s->ir_len);
325 int n_clippings[2] = { 0 };
336 td.in = in;
td.out =
out;
td.write =
s->write;
337 td.ir =
s->data_ir;
td.n_clippings = n_clippings;
338 td.ringbuffer =
s->ringbuffer;
td.temp_src =
s->temp_src;
339 td.out_fft =
s->out_fft;
340 td.in_fft =
s->in_fft;
341 td.temp_afft =
s->temp_afft;
350 if (n_clippings[0] + n_clippings[1] > 0) {
352 n_clippings[0] + n_clippings[1],
out->nb_samples * 2);
363 int nb_input_channels =
ctx->inputs[0]->channels;
364 float gain_lin =
expf((
s->gain - 3 * nb_input_channels) / 20 *
M_LN10);
374 s->buffer_length = 1 << (32 -
ff_clz(
s->air_len));
393 if (!
s->fft[0] || !
s->fft[1] || !
s->ifft[0] || !
s->ifft[1]) {
401 s->ringbuffer[0] =
av_calloc(
s->buffer_length,
sizeof(
float) * nb_input_channels);
402 s->ringbuffer[1] =
av_calloc(
s->buffer_length,
sizeof(
float) * nb_input_channels);
404 s->ringbuffer[0] =
av_calloc(
s->buffer_length,
sizeof(
float));
405 s->ringbuffer[1] =
av_calloc(
s->buffer_length,
sizeof(
float));
412 if (!
s->in_fft[0] || !
s->in_fft[1] ||
413 !
s->out_fft[0] || !
s->out_fft[1] ||
414 !
s->temp_afft[0] || !
s->temp_afft[1]) {
420 if (!
s->ringbuffer[0] || !
s->ringbuffer[1]) {
426 s->temp_src[0] =
av_calloc(
s->air_len,
sizeof(
float));
427 s->temp_src[1] =
av_calloc(
s->air_len,
sizeof(
float));
429 s->data_ir[0] =
av_calloc(nb_input_channels *
s->air_len,
sizeof(*
s->data_ir[0]));
430 s->data_ir[1] =
av_calloc(nb_input_channels *
s->air_len,
sizeof(*
s->data_ir[1]));
431 if (!
s->data_ir[0] || !
s->data_ir[1] || !
s->temp_src[0] || !
s->temp_src[1]) {
436 s->data_hrtf[0] =
av_calloc(
n_fft,
sizeof(*
s->data_hrtf[0]) * nb_input_channels);
437 s->data_hrtf[1] =
av_calloc(
n_fft,
sizeof(*
s->data_hrtf[1]) * nb_input_channels);
438 if (!
s->data_hrtf[0] || !
s->data_hrtf[1]) {
445 int len =
s->hrir_in[
i].ir_len;
451 ptr = (
float *)
frame->extended_data[0];
459 float *data_ir_l =
s->data_ir[0] + idx *
s->air_len;
460 float *data_ir_r =
s->data_ir[1] + idx *
s->air_len;
462 for (j = 0; j <
len; j++) {
463 data_ir_l[j] = ptr[
len * 2 - j * 2 - 2] * gain_lin;
464 data_ir_r[j] = ptr[
len * 2 - j * 2 - 1] * gain_lin;
472 for (j = 0; j <
len; j++) {
473 fft_in_l[j].
re = ptr[j * 2 ] * gain_lin;
474 fft_in_r[j].
re = ptr[j * 2 + 1] * gain_lin;
477 s->tx_fn[0](
s->fft[0], fft_out_l, fft_in_l,
sizeof(float));
478 s->tx_fn[0](
s->fft[0], fft_out_r, fft_in_r,
sizeof(float));
481 int I,
N =
ctx->inputs[1]->channels;
483 for (k = 0; k <
N / 2; k++) {
491 float *data_ir_l =
s->data_ir[0] + idx *
s->air_len;
492 float *data_ir_r =
s->data_ir[1] + idx *
s->air_len;
494 for (j = 0; j <
len; j++) {
495 data_ir_l[j] = ptr[
len *
N - j *
N -
N + I ] * gain_lin;
496 data_ir_r[j] = ptr[
len *
N - j *
N -
N + I + 1] * gain_lin;
504 for (j = 0; j <
len; j++) {
505 fft_in_l[j].
re = ptr[j *
N + I ] * gain_lin;
506 fft_in_r[j].
re = ptr[j *
N + I + 1] * gain_lin;
509 s->tx_fn[0](
s->fft[0], fft_out_l, fft_in_l,
sizeof(float));
510 s->tx_fn[0](
s->fft[0], fft_out_r, fft_in_r,
sizeof(float));
533 for (
i = 0;
i <
s->nb_hrir_inputs;
i++) {
536 if (
s->hrir_in[
i].eof)
545 "HRIR stream %d.\n",
i);
548 s->hrir_in[
i].eof = 1;
562 }
else if (!
s->have_hrirs)
620 for (
i = 1;
i <=
s->nb_hrir_inputs;
i++) {
635 if (
s->nb_irs <
inlink->channels) {
665 for (
i = 0;
i <
s->nb_hrir_inputs;
i++) {
732 #define OFFSET(x) offsetof(HeadphoneContext, x)
733 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
761 .description =
NULL_IF_CONFIG_SMALL(
"Apply headphone binaural spatialization with HRTFs in additional streams."),
763 .priv_class = &headphone_class,
static int convert_coeffs(AVFilterContext *ctx, AVFilterLink *inlink)
AVFrame * ff_get_audio_buffer(AVFilterLink *link, int nb_samples)
Request an audio samples buffer with a specific set of permissions.
A list of supported channel layouts.
#define AV_LOG_WARNING
Something somehow does not look correct.
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf default minimum maximum flags name is the option name
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
AVComplexFloat ** out_fft
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
enum MovChannelLayoutTag * layouts
#define AVERROR_EOF
End of file.
static int parse_channel_name(const char *arg, uint64_t *rchannel)
AVComplexFloat * temp_afft[2]
The exact code depends on how similar the blocks are and how related they are to the and needs to apply these operations to the correct inlink or outlink if there are several Macros are available to factor that when no extra processing is inlink
char * av_asprintf(const char *fmt,...)
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
This structure describes decoded (raw) audio or video data.
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
#define FILTER_QUERY_FUNC(func)
uint64_t av_get_channel_layout(const char *name)
Return a channel layout id that matches name, or 0 if no match is found.
const char * name
Filter name.
A link between two filters.
int channels
Number of channels.
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 (i)MDCTs with an odd length are currently...
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
AVComplexFloat * data_hrtf[2]
#define FF_FILTER_FORWARD_STATUS_BACK_ALL(outlink, filter)
Forward the status on an output link to all input links.
int ff_append_inpad(AVFilterContext *f, AVFilterPad *p)
Append a new input/output pad to the filter's list of such pads.
float(* scalarproduct_float)(const float *v1, const float *v2, int len)
static int activate(AVFilterContext *ctx)
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf type
static av_always_inline float scale(float x, float s)
#define AVFILTER_FLAG_DYNAMIC_INPUTS
The number of the filter inputs is not determined just by AVFilter.inputs.
static __device__ float fabsf(float a)
#define AV_CH_LAYOUT_STEREO
static int config_input(AVFilterLink *inlink)
A filter pad used for either input or output.
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
void(* av_tx_fn)(AVTXContext *s, void *out, void *in, ptrdiff_t stride)
Function pointer to a function to perform the transform.
#define AV_CH_LOW_FREQUENCY
float(* scalarproduct_float)(const float *v1, const float *v2, int len)
Calculate the scalar product of two vectors of floats.
void ff_inlink_request_frame(AVFilterLink *link)
Mark that a frame is wanted on the link.
int channels
number of audio channels, only used for audio.
char * av_strtok(char *s, const char *delim, char **saveptr)
Split the string into several tokens which can be accessed by successive calls to av_strtok().
@ AV_TX_FLOAT_FFT
Standard complex to complex FFT with sample data type AVComplexFloat.
static av_cold void uninit(AVFilterContext *ctx)
static int headphone_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
Describe the class of an AVClass context structure.
int ff_inlink_consume_samples(AVFilterLink *link, unsigned min, unsigned max, AVFrame **rframe)
Take samples from the link's FIFO and update the link's stats.
static void parse_map(AVFilterContext *ctx)
int ff_append_inpad_free_name(AVFilterContext *f, AVFilterPad *p)
static const AVFilterPad outputs[]
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
static int query_formats(AVFilterContext *ctx)
int av_get_channel_layout_nb_channels(uint64_t channel_layout)
Return the number of channels in the channel layout.
AVFILTER_DEFINE_CLASS(headphone)
static int headphone_frame(HeadphoneContext *s, AVFrame *in, AVFilterLink *outlink)
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
AVFilterContext * src
source filter
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf offset
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
av_cold void av_tx_uninit(AVTXContext **ctx)
Frees a context and sets ctx to NULL, does nothing when ctx == NULL.
static const AVOption headphone_options[]
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 layout
int nb_samples
number of audio samples (per channel) described by this frame
#define i(width, name, range_min, range_max)
static av_cold int init(AVFilterContext *ctx)
int av_get_channel_layout_channel_index(uint64_t channel_layout, uint64_t channel)
Get the index of a channel in channel_layout.
Used for passing data between threads.
const char * name
Pad name.
int ff_inlink_queued_samples(AVFilterLink *link)
void * av_calloc(size_t nmemb, size_t size)
AVComplexFloat * in_fft[2]
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
AVComplexFloat ** temp_afft
the frame and frame reference mechanism is intended to as much as expensive copies of that data while still allowing the filters to produce correct results The data is stored in buffers represented by AVFrame structures Several references can point to the same frame buffer
static int headphone_fast_convolute(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
int ff_outlink_get_status(AVFilterLink *link)
Get the status on an output link.
#define AVFILTER_FLAG_SLICE_THREADS
The filter supports multithreading by splitting frames into multiple parts and processing them concur...
const VDPAUPixFmtMap * map
FF_FILTER_FORWARD_STATUS(inlink, outlink)
#define FILTER_OUTPUTS(array)
const AVFilter ff_af_headphone
av_cold AVFloatDSPContext * avpriv_float_dsp_alloc(int bit_exact)
Allocate a float DSP context.
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
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
static av_always_inline int ff_filter_execute(AVFilterContext *ctx, avfilter_action_func *func, void *arg, int *ret, int nb_jobs)
struct HeadphoneContext::hrir_inputs hrir_in[64]
static int config_output(AVFilterLink *outlink)
AVComplexFloat * out_fft[2]
static int check_ir(AVFilterLink *inlink, int input_number)