[FFmpeg-devel] [PATCH 5/5] lavfi: add audio resample filter

Stefano Sabatini stefano.sabatini-lala at poste.it
Tue Aug 16 00:05:57 CEST 2011


On date Friday 2011-08-12 12:41:19 +0300, Mina Nagy Zaki encoded:
> Updated after change to the ff_parse_* functions.

> From 37a57549dfc1cb40d132c5fb21d7cd4f9a93badc Mon Sep 17 00:00:00 2001
> From: Stefano Sabatini <stefano.sabatini-lala at poste.it>
> Date: Sun, 13 Feb 2011 18:00:41 +0100
> Subject: [PATCH 06/16] lavfi: add audio resample filter
> 
> ---
>  doc/filters.texi           |   13 ++
>  libavfilter/Makefile       |    2 +
>  libavfilter/af_aresample.c |  350 ++++++++++++++++++++++++++++++++++++++++++++
>  libavfilter/allfilters.c   |    1 +
>  4 files changed, 366 insertions(+), 0 deletions(-)
>  create mode 100644 libavfilter/af_aresample.c
> 
> diff --git a/doc/filters.texi b/doc/filters.texi
> index 9ce3fed..c2335bc 100644
> --- a/doc/filters.texi
> +++ b/doc/filters.texi
> @@ -124,6 +124,19 @@ aformat=s16:mono\\,stereo:all
>  
>  Pass the audio source unchanged to the output.
>  
> + at section aresample
> +
> +Resample the input audio to the specified sample rate.
> +
> +The filter accepts exactly one parameter, the output sample rate. If not
> +specified then the filter will automatically convert between its input and
> +output sample rates
> +
> + at example
> +# resample to 44100Hz
> +aresample=44100
> + at end example
> +
>  @c man end AUDIO FILTERS
>  
>  @chapter Audio Sources
> diff --git a/libavfilter/Makefile b/libavfilter/Makefile
> index ca54bd3..d233009 100644
> --- a/libavfilter/Makefile
> +++ b/libavfilter/Makefile
> @@ -3,6 +3,7 @@ include $(SUBDIR)../config.mak
>  NAME = avfilter
>  FFLIBS = avutil
>  FFLIBS-$(CONFIG_ACONVERT_FILTER) += avcodec
> +FFLIBS-$(CONFIG_ARESAMPLE_FILTER) += avcodec
>  FFLIBS-$(CONFIG_MOVIE_FILTER) += avformat avcodec
>  FFLIBS-$(CONFIG_SCALE_FILTER) += swscale
>  FFLIBS-$(CONFIG_MP_FILTER) += avcodec
> @@ -22,6 +23,7 @@ OBJS-$(CONFIG_AVCODEC)                       += avcodec.o
>  OBJS-$(CONFIG_ACONVERT_FILTER)               += af_aconvert.o
>  OBJS-$(CONFIG_AFORMAT_FILTER)                += af_aformat.o
>  OBJS-$(CONFIG_ANULL_FILTER)                  += af_anull.o
> +OBJS-$(CONFIG_ARESAMPLE_FILTER)              += af_aresample.o
>  
>  OBJS-$(CONFIG_ANULLSRC_FILTER)               += asrc_anullsrc.o
>  
> diff --git a/libavfilter/af_aresample.c b/libavfilter/af_aresample.c
> new file mode 100644
> index 0000000..9505286
> --- /dev/null
> +++ b/libavfilter/af_aresample.c
> @@ -0,0 +1,350 @@
> +/*
> + * Copyright (c) 2011 Stefano Sabatini
> + * Copyright (c) 2011 Mina Nagy Zaki
> + *
> + * This file is part of FFmpeg.
> + *
> + * FFmpeg is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * FFmpeg is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with FFmpeg; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
> + */
> +
> +/**
> + * @file
> + * resampling audio filter
> + */
> +
> +#include "libavutil/eval.h"
> +#include "libavcodec/avcodec.h"
> +#include "avfilter.h"
> +#include "internal.h"
> +
> +typedef struct {
> +    struct AVResampleContext *resample;
> +    int out_rate;
> +    double ratio;
> +    AVFilterBufferRef *outsamplesref;
> +    int unconsumed_nb_samples,
> +        max_cached_nb_samples;
> +    int16_t *cached_data[8],
> +            *resampled_data[8];
> +} AResampleContext;
> +
> +static av_cold int init(AVFilterContext *ctx, const char *args, void *opaque)
> +{
> +    AResampleContext *aresample = ctx->priv;
> +    int ret;
> +
> +    if (args) {
> +        if ((ret = ff_parse_sample_rate(&aresample->out_rate, args, ctx)) < 0)
> +            return ret;
> +    } else {
> +        aresample->out_rate = -1;
> +    }
> +
> +    return 0;
> +}
> +
> +static av_cold void uninit(AVFilterContext *ctx)
> +{
> +    AResampleContext *aresample = ctx->priv;
> +    if (aresample->outsamplesref) {
> +        int nb_channels =
> +            av_get_channel_layout_nb_channels(
> +                aresample->outsamplesref->audio->channel_layout);
> +        avfilter_unref_buffer(aresample->outsamplesref);
> +        while (nb_channels--) {
> +            av_freep(&(aresample->cached_data[nb_channels]));
> +            av_freep(&(aresample->resampled_data[nb_channels]));
> +        }
> +    }
> +
> +    if (aresample->resample)
> +        av_resample_close(aresample->resample);
> +}
> +
> +static int config_output(AVFilterLink *outlink)
> +{
> +    AVFilterContext *ctx = outlink->src;
> +    AVFilterLink *inlink = ctx->inputs[0];
> +    AResampleContext *aresample = ctx->priv;
> +
> +    if (aresample->out_rate == -1)
> +        aresample->out_rate = outlink->sample_rate;
> +    else
> +        outlink->sample_rate = aresample->out_rate;
> +

> +    //FIXME make the resampling parameters configurable??
> +    aresample->resample = av_resample_init(aresample->out_rate, inlink->sample_rate,
> +                                          16, 10, 0, 0.8);

This is more a todo than a fixme, weird reindent

> +
> +    aresample->ratio = (double)outlink->sample_rate / inlink->sample_rate;
> +
> +    av_log(ctx, AV_LOG_INFO, "r:%"PRId64"Hz -> r:%"PRId64"Hz\n",
> +           inlink->sample_rate, outlink->sample_rate);
> +    return 0;
> +}
> +
> +static int query_formats(AVFilterContext *ctx)
> +{
> +    AVFilterFormats *formats = NULL;
> +
> +    avfilter_add_format(&formats, AV_SAMPLE_FMT_S16);
> +    if (!formats)
> +        return AVERROR(ENOMEM);
> +    avfilter_set_common_sample_formats(ctx, formats);
> +

> +    formats = avfilter_all_channel_layouts();
> +    if (!formats)
> +        return AVERROR(ENOMEM);
> +    avfilter_set_common_channel_layouts(ctx, formats);
> +
> +    formats = avfilter_all_packing_formats();
> +    if (!formats)
> +        return AVERROR(ENOMEM);
> +    avfilter_set_common_packing_formats(ctx, formats);

Note: maybe we may add some convenience function
(e.g. set_common_audio_formats(fmts, chlayouts, packing_fmts)) if we
see we use this code again and again.

> +
> +    return 0;
> +}
> +
> +static void deinterleave(int16_t **outp, int16_t *in,
> +                         int nb_channels, int nb_samples)
> +{
> +    int16_t *out[8];
> +    memcpy(out, outp, nb_channels * sizeof(int16_t*));
> +
> +    switch (nb_channels) {
> +    case 2:
> +        while (nb_samples--) {
> +            *out[0]++ = *in++;
> +            *out[1]++ = *in++;
> +        }
> +        break;
> +    case 3:
> +        while (nb_samples--) {
> +            *out[0]++ = *in++;
> +            *out[1]++ = *in++;
> +            *out[2]++ = *in++;
> +        }
> +        break;
> +    case 4:
> +        while (nb_samples--) {
> +            *out[0]++ = *in++;
> +            *out[1]++ = *in++;
> +            *out[2]++ = *in++;
> +            *out[3]++ = *in++;
> +        }
> +        break;
> +    case 5:
> +        while (nb_samples--) {
> +            *out[0]++ = *in++;
> +            *out[1]++ = *in++;
> +            *out[2]++ = *in++;
> +            *out[3]++ = *in++;
> +            *out[4]++ = *in++;
> +        }
> +        break;
> +    case 6:
> +        while (nb_samples--) {
> +            *out[0]++ = *in++;
> +            *out[1]++ = *in++;
> +            *out[2]++ = *in++;
> +            *out[3]++ = *in++;
> +            *out[4]++ = *in++;
> +            *out[5]++ = *in++;
> +        }
> +        break;
> +    case 8:
> +        while (nb_samples--) {
> +            *out[0]++ = *in++;
> +            *out[1]++ = *in++;
> +            *out[2]++ = *in++;
> +            *out[3]++ = *in++;
> +            *out[4]++ = *in++;
> +            *out[5]++ = *in++;
> +            *out[6]++ = *in++;
> +            *out[7]++ = *in++;
> +        }
> +        break;
> +    }
> +}
> +
> +static void interleave(int16_t *out, int16_t **inp,
> +        int nb_channels, int nb_samples)
> +{
> +    int16_t *in[8];
> +    memcpy(in, inp, nb_channels * sizeof(int16_t*));
> +
> +    switch (nb_channels) {
> +    case 2:
> +        while (nb_samples--) {
> +            *out++ = *in[0]++;
> +            *out++ = *in[1]++;
> +        }
> +        break;
> +    case 3:
> +        while (nb_samples--) {
> +            *out++ = *in[0]++;
> +            *out++ = *in[1]++;
> +            *out++ = *in[2]++;
> +        }
> +        break;
> +    case 4:
> +        while (nb_samples--) {
> +            *out++ = *in[0]++;
> +            *out++ = *in[1]++;
> +            *out++ = *in[2]++;
> +            *out++ = *in[3]++;
> +        }
> +        break;
> +    case 5:
> +        while (nb_samples--) {
> +            *out++ = *in[0]++;
> +            *out++ = *in[1]++;
> +            *out++ = *in[2]++;
> +            *out++ = *in[3]++;
> +            *out++ = *in[4]++;
> +        }
> +        break;
> +    case 6:
> +        while (nb_samples--) {
> +            *out++ = *in[0]++;
> +            *out++ = *in[1]++;
> +            *out++ = *in[2]++;
> +            *out++ = *in[3]++;
> +            *out++ = *in[4]++;
> +            *out++ = *in[5]++;
> +        }
> +        break;
> +    case 8:
> +        while (nb_samples--) {
> +            *out++ = *in[0]++;
> +            *out++ = *in[1]++;
> +            *out++ = *in[2]++;
> +            *out++ = *in[3]++;
> +            *out++ = *in[4]++;
> +            *out++ = *in[5]++;
> +            *out++ = *in[6]++;
> +            *out++ = *in[7]++;
> +        }
> +        break;
> +    }
> +}
> +
> +static void filter_samples(AVFilterLink *inlink, AVFilterBufferRef *insamplesref)
> +{
> +    AResampleContext *aresample  = inlink->dst->priv;
> +    AVFilterLink * const outlink = inlink->dst->outputs[0];
> +    int i,
> +        in_nb_samples            = insamplesref->audio->nb_samples,
> +        cached_nb_samples        = in_nb_samples + aresample->unconsumed_nb_samples,
> +        requested_out_nb_samples = aresample->ratio * cached_nb_samples,
> +        nb_channels              =
> +            av_get_channel_layout_nb_channels(inlink->channel_layout);
> +
> +    if (cached_nb_samples > aresample->max_cached_nb_samples) {
> +        for (i = 0; i < nb_channels; i++) {
> +            aresample->cached_data[i]    =
> +                av_realloc(aresample->cached_data[i], cached_nb_samples * sizeof(int16_t));
> +            aresample->resampled_data[i] =
> +                av_realloc(aresample->resampled_data[i],
> +                           FFALIGN(sizeof(int16_t) * requested_out_nb_samples, 16));
> +
> +            if (aresample->cached_data[i] == NULL || aresample->resampled_data[i] == NULL)
> +                return;
> +        }
> +        aresample->max_cached_nb_samples = cached_nb_samples;
> +
> +        if (aresample->outsamplesref)
> +            avfilter_unref_buffer(aresample->outsamplesref);
> +
> +        aresample->outsamplesref = avfilter_get_audio_buffer(outlink,
> +                                                            AV_PERM_WRITE | AV_PERM_REUSE2,
> +                                                            inlink->format,
> +                                                            requested_out_nb_samples,
> +                                                            insamplesref->audio->channel_layout,
> +                                                            insamplesref->audio->planar);
> +
> +        avfilter_copy_buffer_ref_props(aresample->outsamplesref, insamplesref);
> +        aresample->outsamplesref->pts =
> +            insamplesref->pts / inlink->sample_rate * outlink->sample_rate;
> +        aresample->outsamplesref->audio->sample_rate = outlink->sample_rate;
> +        outlink->out_buf = aresample->outsamplesref;
> +    }
> +
> +    /* av_resample() works with planar audio buffers */
> +    if (!inlink->planar && nb_channels > 1) {
> +        int16_t *out[8];
> +        for (i = 0; i < nb_channels; i++)
> +            out[i] = aresample->cached_data[i] + aresample->unconsumed_nb_samples;
> +
> +        deinterleave(out, (int16_t *)insamplesref->data[0],
> +                     nb_channels, in_nb_samples);
> +    } else {
> +        for (i = 0; i < nb_channels; i++)
> +            memcpy(aresample->cached_data[i] + aresample->unconsumed_nb_samples,
> +                   insamplesref->data[i],
> +                   in_nb_samples * sizeof(int16_t));
> +    }
> +

> +    for (i = 0; i < nb_channels; i++) {
> +        int consumed;
> +        const int is_last = i+1 == nb_channels;
> +
> +        aresample->outsamplesref->audio->nb_samples =
> +            av_resample(aresample->resample,
> +                        aresample->resampled_data[i], aresample->cached_data[i],
> +                        &consumed,
> +                        cached_nb_samples,
> +                        requested_out_nb_samples, is_last);
> +
> +        /* move unconsumed data back to the beginning of the cache */
> +        aresample->unconsumed_nb_samples = cached_nb_samples - consumed;
> +        memmove(aresample->cached_data[i], aresample->cached_data[i] + consumed,
> +                aresample->unconsumed_nb_samples * sizeof(int16_t));
> +    }

nit++: consumed -> consumed_nb_samples

> +

> +
> +    /* copy resampled data to the output samplesref */
> +    if (!inlink->planar && nb_channels > 1) {
> +        interleave((int16_t *)aresample->outsamplesref->data[0],
> +                   aresample->resampled_data,
> +                   nb_channels, aresample->outsamplesref->audio->nb_samples);
> +    } else {
> +        for (i = 0; i < nb_channels; i++)
> +            memcpy(aresample->outsamplesref->data[i], aresample->resampled_data[i],
> +                   aresample->outsamplesref->audio->nb_samples * sizeof(int16_t));
> +    }

I wonder if there is some way to directly process data in place, with
no memcpy/memmove (but I don't think so...).

> +
> +    avfilter_filter_samples(outlink, avfilter_ref_buffer(aresample->outsamplesref, ~0));
> +    avfilter_unref_buffer(insamplesref);
> +}
> +
> +AVFilter avfilter_af_aresample = {
> +    .name          = "aresample",
> +    .description   = NULL_IF_CONFIG_SMALL("Resample audio data."),
> +    .init          = init,
> +    .uninit        = uninit,
> +    .query_formats = query_formats,
> +    .priv_size     = sizeof(AResampleContext),
> +
> +    .inputs    = (AVFilterPad[]) {{ .name            = "default",
> +                                    .type            = AVMEDIA_TYPE_AUDIO,
> +                                    .filter_samples  = filter_samples,
> +                                    .min_perms       = AV_PERM_READ, },
> +                                  { .name = NULL}},
> +    .outputs   = (AVFilterPad[]) {{ .name            = "default",
> +                                    .config_props    = config_output,
> +                                    .type            = AVMEDIA_TYPE_AUDIO, },
> +                                  { .name = NULL}},
> +};

Anyway I'm fine with the patch, maybe you could add a test for it,
I'll apply in a few days if I see no more comments.
-- 
FFmpeg = Free Free MultiPurpose Eager Gadget


More information about the ffmpeg-devel mailing list