[FFmpeg-devel] [PATCH] avfilter/af_silencedetect: support more sample formats
Clément Bœsch
u at pkh.me
Fri Oct 4 20:11:15 CEST 2013
Thanks for this patch.
On Fri, Oct 04, 2013 at 06:00:33PM +0000, Paul B Mahol wrote:
> Signed-off-by: Paul B Mahol <onemda at gmail.com>
> ---
> libavfilter/af_silencedetect.c | 112 +++++++++++++++++++++++++++++------------
> 1 file changed, 79 insertions(+), 33 deletions(-)
>
> diff --git a/libavfilter/af_silencedetect.c b/libavfilter/af_silencedetect.c
> index a9b9358..a510912 100644
> --- a/libavfilter/af_silencedetect.c
> +++ b/libavfilter/af_silencedetect.c
> @@ -32,13 +32,17 @@
> #include "avfilter.h"
> #include "internal.h"
>
> -typedef struct {
> +typedef struct SilenceDetectContext {
Unrelated, and I don't like it, so please drop.
> const AVClass *class;
> double noise; ///< noise amplitude ratio
> double duration; ///< minimum duration of silence until notification
> int64_t nb_null_samples; ///< current number of continuous zero samples
> int64_t start; ///< if silence is detected, this value contains the time of the first zero sample
> int last_sample_rate; ///< last sample rate to check for sample rate changes
> +
> + void (*silencedetect)(struct SilenceDetectContext *s, AVFrame *insamples,
> + int nb_samples, int64_t nb_samples_notify,
> + AVRational time_base);
> } SilenceDetectContext;
>
> #define OFFSET(x) offsetof(SilenceDetectContext, x)
> @@ -59,9 +63,76 @@ static char *get_metadata_val(AVFrame *insamples, const char *key)
> return e && e->value ? e->value : NULL;
> }
>
> +static av_always_inline void update(SilenceDetectContext *s, AVFrame *insamples,
> + int is_silence, int64_t nb_samples_notify,
> + AVRational time_base)
> +{
> + if (is_silence) {
> + if (!s->start) {
> + s->nb_null_samples++;
> + if (s->nb_null_samples >= nb_samples_notify) {
> + s->start = insamples->pts - (int64_t)(s->duration / av_q2d(time_base) + .5);
> + av_dict_set(&insamples->metadata, "lavfi.silence_start",
> + av_ts2timestr(s->start, &time_base), 0);
setlocal cinoptions=(0
same several times below
> + av_log(s, AV_LOG_INFO, "silence_start: %s\n",
> + get_metadata_val(insamples, "lavfi.silence_start"));
> + }
> + }
> + } else {
> + if (s->start) {
> + av_dict_set(&insamples->metadata, "lavfi.silence_end",
> + av_ts2timestr(insamples->pts, &time_base), 0);
> + av_dict_set(&insamples->metadata, "lavfi.silence_duration",
> + av_ts2timestr(insamples->pts - s->start, &time_base), 0);
> + av_log(s, AV_LOG_INFO,
> + "silence_end: %s | silence_duration: %s\n",
> + get_metadata_val(insamples, "lavfi.silence_end"),
> + get_metadata_val(insamples, "lavfi.silence_duration"));
> + }
> + s->nb_null_samples = s->start = 0;
> + }
> +}
> +
> +#define SILENCE_DETECT(name, type) \
> +static void silencedetect_##name(SilenceDetectContext *s, AVFrame *insamples, \
> + int nb_samples, int64_t nb_samples_notify, \
> + AVRational time_base) \
> +{ \
> + type *p = (type *)insamples->data[0]; \
> + const type noise = s->noise; \
> + int i; \
> + \
> + for (i = 0; i < nb_samples; i++, p++) \
> + update(s, insamples, *p < noise && *p > -noise, \
> + nb_samples_notify, time_base); \
> +}
> +
> +SILENCE_DETECT(dbl, double);
> +SILENCE_DETECT(flt, float);
> +SILENCE_DETECT(s32, int32_t);
> +SILENCE_DETECT(s16, int16_t);
you can/should drop the trailing ;
> +
> +static int config_input(AVFilterLink *inlink)
> +{
> + AVFilterContext *ctx = inlink->dst;
> + SilenceDetectContext *s = ctx->priv;
> +
> + switch (inlink->format) {
> + case AV_SAMPLE_FMT_DBL: s->silencedetect = silencedetect_dbl; break;
> + case AV_SAMPLE_FMT_FLT: s->silencedetect = silencedetect_flt; break;
> + case AV_SAMPLE_FMT_S32:
> + s->noise *= INT32_MAX;
> + s->silencedetect = silencedetect_s32; break;
please move the break below
> + case AV_SAMPLE_FMT_S16:
> + s->noise *= INT16_MAX;
> + s->silencedetect = silencedetect_s16; break;
> + };
trailing ;
> +
> + return 0;
> +}
> +
> static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
> {
> - int i;
> SilenceDetectContext *s = inlink->dst->priv;
> const int nb_channels = inlink->channels;
> const int srate = inlink->sample_rate;
> @@ -73,38 +144,9 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *insamples)
> s->nb_null_samples = srate * s->nb_null_samples / s->last_sample_rate;
> s->last_sample_rate = srate;
>
> - // TODO: support more sample formats
> // TODO: document metadata
> - if (insamples->format == AV_SAMPLE_FMT_DBL) {
> - double *p = (double *)insamples->data[0];
> -
> - for (i = 0; i < nb_samples; i++, p++) {
> - if (*p < s->noise && *p > -s->noise) {
> - if (!s->start) {
> - s->nb_null_samples++;
> - if (s->nb_null_samples >= nb_samples_notify) {
> - s->start = insamples->pts - (int64_t)(s->duration / av_q2d(inlink->time_base) + .5);
> - av_dict_set(&insamples->metadata, "lavfi.silence_start",
> - av_ts2timestr(s->start, &inlink->time_base), 0);
> - av_log(s, AV_LOG_INFO, "silence_start: %s\n",
> - get_metadata_val(insamples, "lavfi.silence_start"));
> - }
> - }
> - } else {
> - if (s->start) {
> - av_dict_set(&insamples->metadata, "lavfi.silence_end",
> - av_ts2timestr(insamples->pts, &inlink->time_base), 0);
> - av_dict_set(&insamples->metadata, "lavfi.silence_duration",
> - av_ts2timestr(insamples->pts - s->start, &inlink->time_base), 0);
> - av_log(s, AV_LOG_INFO,
> - "silence_end: %s | silence_duration: %s\n",
> - get_metadata_val(insamples, "lavfi.silence_end"),
> - get_metadata_val(insamples, "lavfi.silence_duration"));
> - }
> - s->nb_null_samples = s->start = 0;
> - }
> - }
> - }
> + s->silencedetect(s, insamples, nb_samples, nb_samples_notify,
> + inlink->time_base);
align
>
> return ff_filter_frame(inlink->dst->outputs[0], insamples);
> }
> @@ -115,6 +157,9 @@ static int query_formats(AVFilterContext *ctx)
> AVFilterChannelLayouts *layouts = NULL;
> static const enum AVSampleFormat sample_fmts[] = {
> AV_SAMPLE_FMT_DBL,
> + AV_SAMPLE_FMT_FLT,
> + AV_SAMPLE_FMT_S32,
> + AV_SAMPLE_FMT_S16,
> AV_SAMPLE_FMT_NONE
> };
>
> @@ -140,6 +185,7 @@ static const AVFilterPad silencedetect_inputs[] = {
> {
> .name = "default",
> .type = AVMEDIA_TYPE_AUDIO,
> + .config_props = config_input,
> .filter_frame = filter_frame,
> },
> { NULL }
It looks fine; I'm assuming you tried with the FATE sample and compared
the results.
--
Clément B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 490 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20131004/93de8c6a/attachment.asc>
More information about the ffmpeg-devel
mailing list