lavfi: add audio convert filter

Add aconvert filter to perform sample format, channel layout, and
packing format conversion.

The aconvert code depends on audio conversion code in libavcodec, so
this requires a dependency on libavcodec.

Based on previous work by S.N. Hemanth Meenakshisundaram and Mina Nagy
Zaki, performed for the GSoC 2010 and 2011.

Changelog

@@ -55,6 +55,7 @@ easier to use. The changes are:
 - H.264 Decoding on Android via Stagefright
 - Prores decoder
 - BIN/XBIN/ADF/IDF text file decoder
+- aconvert audio filter added
 
 
 version 0.8:

configure

@@ -1575,6 +1575,7 @@ udp_protocol_deps="network"
 
 # filters
 abuffer_filter_deps="strtok_r"
+aconvert_filter_deps="strtok_r"
 aformat_filter_deps="strtok_r"
 amovie_filter_deps="avcodec avformat"
 blackframe_filter_deps="gpl"

doc/filters.texi

@@ -99,6 +99,42 @@ build.
 
 Below is a description of the currently available audio filters.
 
+@section aconvert
+
+Convert the input audio format to the specified formats.
+
+The filter accepts a string of the form:
+"@var{sample_format}:@var{channel_layout}:@var{packing_format}".
+
+@var{sample_format} specifies the sample format, and can be a string or
+the corresponding numeric value defined in @file{libavutil/samplefmt.h}.
+
+@var{channel_layout} specifies the channel layout, and can be a string
+or the corresponding numer value defined in @file{libavutil/chlayout.h}.
+
+@var{packing_format} specifies the type of packing in output, can be one
+of "planar" or "packed", or the corresponding numeric values "0" or "1".
+
+The special parameter "auto", signifies that the filter will
+automatically select the output format depending on the output filter.
+
+Some examples follow.
+
+@itemize
+@item
+Convert input to unsigned 8-bit, stereo, packed:
+@example
+aconvert=u8:stereo:packed
+@end example
+
+@item
+Convert input to unsigned 8-bit, automatically select out channel layout
+and packing format:
+@example
+aconvert=u8:auto:auto
+@end example
+@end itemize
+
 @section aformat
 
 Convert the input audio to one of the specified formats. The framework will

libavfilter/Makefile

@@ -2,6 +2,8 @@ include $(SUBDIR)../config.mak
 
 NAME = avfilter
 FFLIBS = avutil
+
+FFLIBS-$(CONFIG_ACONVERT_FILTER) += avcodec
 FFLIBS-$(CONFIG_AMOVIE_FILTER) += avformat avcodec
 FFLIBS-$(CONFIG_ARESAMPLE_FILTER) += avcodec
 FFLIBS-$(CONFIG_MOVIE_FILTER) += avformat avcodec
@@ -20,6 +22,7 @@ OBJS = allfilters.o                                                     \
 
 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

libavfilter/af_aconvert.c

+/*
+ * Copyright (c) 2010 S.N. Hemanth Meenakshisundaram <smeenaks@ucsd.edu>
+ * 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
+ * sample format and channel layout conversion audio filter
+ * based on code in libavcodec/resample.c by Fabrice Bellard and
+ * libavcodec/audioconvert.c by Michael Niedermayer
+ */
+
+#include "libavutil/audioconvert.h"
+#include "libavcodec/audioconvert.h"
+#include "avfilter.h"
+#include "internal.h"
+
+typedef struct {
+    enum AVSampleFormat  out_sample_fmt,  in_sample_fmt;   ///< in/out sample formats
+    int64_t              out_chlayout,    in_chlayout;     ///< in/out channel layout
+    int                  out_nb_channels, in_nb_channels;  ///< number of in/output channels
+    enum AVFilterPacking out_packing_fmt, in_packing_fmt;  ///< output packing format
+
+    int max_nb_samples;                     ///< maximum number of buffered samples
+    AVFilterBufferRef *mix_samplesref;      ///< rematrixed buffer
+    AVFilterBufferRef *out_samplesref;      ///< output buffer after required conversions
+
+    uint8_t *in_mix[8], *out_mix[8];        ///< input/output for rematrixing functions
+    uint8_t *packed_data[8];                ///< pointers for packing conversion
+    int out_strides[8], in_strides[8];      ///< input/output strides for av_audio_convert
+    uint8_t **in_conv, **out_conv;          ///< input/output for av_audio_convert
+
+    AVAudioConvert *audioconvert_ctx;       ///< context for conversion to output sample format
+
+    void (*convert_chlayout)();             ///< function to do the requested rematrixing
+} AConvertContext;
+
+#define REMATRIX_FUNC_SIG(NAME) static void REMATRIX_FUNC_NAME(NAME) \
+    (FMT_TYPE *outp[], FMT_TYPE *inp[], int nb_samples, AConvertContext *aconvert)
+
+#define FMT_TYPE uint8_t
+#define REMATRIX_FUNC_NAME(NAME) NAME ## _u8
+#include "af_aconvert_rematrix.c"
+
+#define FMT_TYPE int16_t
+#define REMATRIX_FUNC_NAME(NAME) NAME ## _s16
+#include "af_aconvert_rematrix.c"
+
+#define FMT_TYPE int32_t
+#define REMATRIX_FUNC_NAME(NAME) NAME ## _s32
+#include "af_aconvert_rematrix.c"
+
+#define FLOATING
+
+#define FMT_TYPE float
+#define REMATRIX_FUNC_NAME(NAME) NAME ## _flt
+#include "af_aconvert_rematrix.c"
+
+#define FMT_TYPE double
+#define REMATRIX_FUNC_NAME(NAME) NAME ## _dbl
+#include "af_aconvert_rematrix.c"
+
+#define FMT_TYPE uint8_t
+#define REMATRIX_FUNC_NAME(NAME) NAME
+REMATRIX_FUNC_SIG(stereo_remix_planar)
+{
+    int size = av_get_bytes_per_sample(aconvert->in_sample_fmt) * nb_samples;
+
+    memcpy(outp[0], inp[0], size);
+    memcpy(outp[1], inp[aconvert->in_nb_channels == 1 ? 0 : 1], size);
+}
+
+#define REGISTER_FUNC_PACKING(INCHLAYOUT, OUTCHLAYOUT, FUNC, PACKING)   \
+    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_U8,  FUNC##_u8},   \
+    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_S16, FUNC##_s16},  \
+    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_S32, FUNC##_s32},  \
+    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_FLT, FUNC##_flt},  \
+    {INCHLAYOUT, OUTCHLAYOUT, PACKING, AV_SAMPLE_FMT_DBL, FUNC##_dbl},
+
+#define REGISTER_FUNC(INCHLAYOUT, OUTCHLAYOUT, FUNC)                                \
+    REGISTER_FUNC_PACKING(INCHLAYOUT, OUTCHLAYOUT, FUNC##_packed, AVFILTER_PACKED)  \
+    REGISTER_FUNC_PACKING(INCHLAYOUT, OUTCHLAYOUT, FUNC##_planar, AVFILTER_PLANAR)
+
+static struct RematrixFunctionInfo {
+    int64_t in_chlayout, out_chlayout;
+    int planar, sfmt;
+    void (*func)();
+} rematrix_funcs[] = {
+    REGISTER_FUNC        (AV_CH_LAYOUT_STEREO,  AV_CH_LAYOUT_5POINT1, stereo_to_surround_5p1)
+    REGISTER_FUNC        (AV_CH_LAYOUT_5POINT1, AV_CH_LAYOUT_STEREO,  surround_5p1_to_stereo)
+    REGISTER_FUNC_PACKING(AV_CH_LAYOUT_STEREO,  AV_CH_LAYOUT_MONO,    stereo_to_mono_packed, AVFILTER_PACKED)
+    REGISTER_FUNC_PACKING(AV_CH_LAYOUT_MONO,    AV_CH_LAYOUT_STEREO,  mono_to_stereo_packed, AVFILTER_PACKED)
+    REGISTER_FUNC        (0,                    AV_CH_LAYOUT_MONO,    mono_downmix)
+    REGISTER_FUNC_PACKING(0,                    AV_CH_LAYOUT_STEREO,  stereo_downmix_packed, AVFILTER_PACKED)
+
+    // This function works for all sample formats
+    {0, AV_CH_LAYOUT_STEREO, AVFILTER_PLANAR, -1, stereo_remix_planar}
+};
+
+static av_cold int init(AVFilterContext *ctx, const char *args0, void *opaque)
+{
+    AConvertContext *aconvert = ctx->priv;
+    char *arg, *ptr = NULL;
+    int ret = 0;
+    char *args = av_strdup(args0);
+
+    aconvert->out_sample_fmt  = AV_SAMPLE_FMT_NONE;
+    aconvert->out_chlayout    = 0;
+    aconvert->out_packing_fmt = -1;
+
+    if ((arg = strtok_r(args, ":", &ptr)) && strcmp(arg, "auto")) {
+        if ((ret = ff_parse_sample_format(&aconvert->out_sample_fmt, arg, ctx)) < 0)
+            goto end;
+    }
+    if ((arg = strtok_r(NULL, ":", &ptr)) && strcmp(arg, "auto")) {
+        if ((ret = ff_parse_channel_layout(&aconvert->out_chlayout, arg, ctx)) < 0)
+            goto end;
+    }
+    if ((arg = strtok_r(NULL, ":", &ptr)) && strcmp(arg, "auto")) {
+        if ((ret = ff_parse_packing_format((int *)&aconvert->out_packing_fmt, arg, ctx)) < 0)
+            goto end;
+    }
+
+end:
+    av_freep(&args);
+    return ret;
+}
+
+static av_cold void uninit(AVFilterContext *ctx)
+{
+    AConvertContext *aconvert = ctx->priv;
+    avfilter_unref_buffer(aconvert->mix_samplesref);
+    avfilter_unref_buffer(aconvert->out_samplesref);
+    if (aconvert->audioconvert_ctx)
+        av_audio_convert_free(aconvert->audioconvert_ctx);
+}
+
+static int query_formats(AVFilterContext *ctx)
+{
+    AVFilterFormats *formats = NULL;
+    AConvertContext *aconvert = ctx->priv;
+    AVFilterLink *inlink  = ctx->inputs[0];
+    AVFilterLink *outlink = ctx->outputs[0];
+
+    avfilter_formats_ref(avfilter_make_all_formats(AVMEDIA_TYPE_AUDIO),
+                         &inlink->out_formats);
+    if (aconvert->out_sample_fmt != AV_SAMPLE_FMT_NONE) {
+        formats = NULL;
+        avfilter_add_format(&formats, aconvert->out_sample_fmt);
+        avfilter_formats_ref(formats, &outlink->in_formats);
+    } else
+        avfilter_formats_ref(avfilter_make_all_formats(AVMEDIA_TYPE_AUDIO),
+                             &outlink->in_formats);
+
+    avfilter_formats_ref(avfilter_make_all_channel_layouts(),
+                         &inlink->out_chlayouts);
+    if (aconvert->out_chlayout != 0) {
+        formats = NULL;
+        avfilter_add_format(&formats, aconvert->out_chlayout);
+        avfilter_formats_ref(formats, &outlink->in_chlayouts);
+    } else
+        avfilter_formats_ref(avfilter_make_all_channel_layouts(),
+                             &outlink->in_chlayouts);
+
+    avfilter_formats_ref(avfilter_make_all_packing_formats(),
+                         &inlink->out_packing);
+    if (aconvert->out_packing_fmt != -1) {
+        formats = NULL;
+        avfilter_add_format(&formats, aconvert->out_packing_fmt);
+        avfilter_formats_ref(formats, &outlink->in_packing);
+    } else
+        avfilter_formats_ref(avfilter_make_all_packing_formats(),
+                             &outlink->in_packing);
+
+    return 0;
+}
+
+static int config_output(AVFilterLink *outlink)
+{
+    AVFilterLink *inlink = outlink->src->inputs[0];
+    AConvertContext *aconvert = outlink->src->priv;
+    char buf1[64], buf2[64];
+
+    aconvert->in_sample_fmt  = inlink->format;
+    aconvert->in_packing_fmt = inlink->planar;
+    if (aconvert->out_packing_fmt == -1)
+        aconvert->out_packing_fmt = outlink->planar;
+    aconvert->in_chlayout    = inlink->channel_layout;
+    aconvert->in_nb_channels =
+        av_get_channel_layout_nb_channels(inlink->channel_layout);
+
+    /* if not specified in args, use the format and layout of the output */
+    if (aconvert->out_sample_fmt == AV_SAMPLE_FMT_NONE)
+        aconvert->out_sample_fmt = outlink->format;
+    if (aconvert->out_chlayout   == 0)
+        aconvert->out_chlayout   = outlink->channel_layout;
+    aconvert->out_nb_channels  =
+        av_get_channel_layout_nb_channels(outlink->channel_layout);
+
+    av_get_channel_layout_string(buf1, sizeof(buf1),
+                                 -1, inlink ->channel_layout);
+    av_get_channel_layout_string(buf2, sizeof(buf2),
+                                 -1, outlink->channel_layout);
+    av_log(outlink->src, AV_LOG_INFO,
+           "fmt:%s cl:%s planar:%i -> fmt:%s cl:%s planar:%i\n",
+           av_get_sample_fmt_name(inlink ->format), buf1, inlink ->planar,
+           av_get_sample_fmt_name(outlink->format), buf2, outlink->planar);
+
+    /* compute which channel layout conversion to use */
+    if (inlink->channel_layout != outlink->channel_layout) {
+        int i;
+        for (i = 0; i < sizeof(rematrix_funcs); i++) {
+            const struct RematrixFunctionInfo *f = &rematrix_funcs[i];
+            if ((f->in_chlayout  == 0 || f->in_chlayout  == inlink ->channel_layout) &&
+                (f->out_chlayout == 0 || f->out_chlayout == outlink->channel_layout) &&
+                (f->planar == -1 || f->planar == inlink->planar) &&
+                (f->sfmt   == -1 || f->sfmt   == inlink->format)
+               ) {
+                aconvert->convert_chlayout = f->func;
+                break;
+            }
+        }
+        if (!aconvert->convert_chlayout) {
+            av_log(outlink->src, AV_LOG_ERROR,
+                   "Unsupported channel layout conversion '%s -> %s' requested!\n",
+                   buf1, buf2);
+            return AVERROR(EINVAL);
+        }
+    }
+
+    return 0;
+}
+
+static int init_buffers(AVFilterLink *inlink, int nb_samples)
+{
+    AConvertContext *aconvert = inlink->dst->priv;
+    AVFilterLink * const outlink = inlink->dst->outputs[0];
+    int i, packed_stride = 0;
+    const unsigned
+        packing_conv = inlink->planar != outlink->planar &&
+                       aconvert->out_nb_channels != 1,
+        format_conv  = inlink->format != outlink->format;
+    int nb_channels  = aconvert->out_nb_channels;
+
+    uninit(inlink->dst);
+    aconvert->max_nb_samples = nb_samples;
+
+    if (aconvert->convert_chlayout) {
+        /* allocate buffer for storing intermediary mixing samplesref */
+        uint8_t *data[8];
+        int linesize[8];
+        int nb_channels = av_get_channel_layout_nb_channels(outlink->channel_layout);
+
+        if (av_samples_alloc(data, linesize, nb_channels, nb_samples,
+                             inlink->format, inlink->planar, 16) < 0)
+            goto fail_no_mem;
+        aconvert->mix_samplesref =
+            avfilter_get_audio_buffer_ref_from_arrays(data, linesize, AV_PERM_WRITE,
+                                                      nb_samples, inlink->format,
+                                                      outlink->channel_layout,
+                                                      inlink->planar);
+        if (!aconvert->mix_samplesref)
+            goto fail_no_mem;
+    }
+
+    // if there's a format/packing conversion we need an audio_convert context
+    if (format_conv || packing_conv) {
+        aconvert->out_samplesref =
+            avfilter_get_audio_buffer(outlink, AV_PERM_WRITE, nb_samples);
+        if (!aconvert->out_samplesref)
+            goto fail_no_mem;
+
+        aconvert->in_strides [0] = av_get_bytes_per_sample(inlink ->format);
+        aconvert->out_strides[0] = av_get_bytes_per_sample(outlink->format);
+
+        aconvert->out_conv = aconvert->out_samplesref->data;
+        if (aconvert->mix_samplesref)
+            aconvert->in_conv = aconvert->mix_samplesref->data;
+
+        if (packing_conv) {
+            // packed -> planar
+            if (outlink->planar == AVFILTER_PLANAR) {
+                if (aconvert->mix_samplesref)
+                    aconvert->packed_data[0] = aconvert->mix_samplesref->data[0];
+                aconvert->in_conv         = aconvert->packed_data;
+                packed_stride             = aconvert->in_strides[0];
+                aconvert->in_strides[0]  *= nb_channels;
+            // planar -> packed
+            } else {
+                aconvert->packed_data[0]  = aconvert->out_samplesref->data[0];
+                aconvert->out_conv        = aconvert->packed_data;
+                packed_stride             = aconvert->out_strides[0];
+                aconvert->out_strides[0] *= nb_channels;
+            }
+        } else if (outlink->planar == AVFILTER_PACKED) {
+            /* If there's no packing conversion, and the stream is packed
+             * then we treat the entire stream as one big channel
+             */
+            nb_channels = 1;
+        }
+
+        for (i = 1; i < nb_channels; i++) {
+            aconvert->packed_data[i] = aconvert->packed_data[i-1] + packed_stride;
+            aconvert->in_strides[i]  = aconvert->in_strides[0];
+            aconvert->out_strides[i] = aconvert->out_strides[0];
+        }
+
+        aconvert->audioconvert_ctx =
+                av_audio_convert_alloc(outlink->format, nb_channels,
+                                       inlink->format,  nb_channels, NULL, 0);
+        if (!aconvert->audioconvert_ctx)
+            goto fail_no_mem;
+    }
+
+    return 0;
+
+fail_no_mem:
+    av_log(inlink->dst, AV_LOG_ERROR, "Could not allocate memory.\n");
+    return AVERROR(ENOMEM);
+}
+
+static void filter_samples(AVFilterLink *inlink, AVFilterBufferRef *insamplesref)
+{
+    AConvertContext *aconvert = inlink->dst->priv;
+    AVFilterBufferRef *curbuf = insamplesref;
+    AVFilterLink * const outlink = inlink->dst->outputs[0];
+    int chan_mult;
+
+    /* in/reinint the internal buffers if this is the first buffer
+     * provided or it is needed to use a bigger one */
+    if (!aconvert->max_nb_samples ||
+        (curbuf->audio->nb_samples > aconvert->max_nb_samples))
+        if (init_buffers(inlink, curbuf->audio->nb_samples) < 0) {
+            av_log(inlink->dst, AV_LOG_ERROR, "Could not initialize buffers.\n");
+            return;
+        }
+
+    /* if channel mixing is required */
+    if (aconvert->mix_samplesref) {
+        memcpy(aconvert->in_mix,  curbuf->data, sizeof(aconvert->in_mix));
+        memcpy(aconvert->out_mix, aconvert->mix_samplesref->data, sizeof(aconvert->out_mix));
+        aconvert->convert_chlayout(aconvert->out_mix,
+                                   aconvert->in_mix,
+                                   curbuf->audio->nb_samples,
+                                   aconvert);
+        curbuf = aconvert->mix_samplesref;
+    }
+
+    if (aconvert->audioconvert_ctx) {
+        if (!aconvert->mix_samplesref) {
+            if (aconvert->in_conv == aconvert->packed_data) {
+                int i, packed_stride = av_get_bytes_per_sample(inlink->format);
+                aconvert->packed_data[0] = curbuf->data[0];
+                for (i = 1; i < aconvert->out_nb_channels; i++)
+                    aconvert->packed_data[i] = aconvert->packed_data[i-1] + packed_stride;
+            } else {
+                aconvert->in_conv = curbuf->data;
+            }
+        }
+
+        chan_mult = inlink->planar == outlink->planar && inlink->planar == 0 ?
+            aconvert->out_nb_channels : 1;
+
+        av_audio_convert(aconvert->audioconvert_ctx,
+                         (void * const *) aconvert->out_conv,
+                         aconvert->out_strides,
+                         (const void * const *) aconvert->in_conv,
+                         aconvert->in_strides,
+                         curbuf->audio->nb_samples * chan_mult);
+
+        curbuf = aconvert->out_samplesref;
+    }
+
+    avfilter_copy_buffer_ref_props(curbuf, insamplesref);
+    curbuf->audio->channel_layout = outlink->channel_layout;
+    curbuf->audio->planar         = outlink->planar;
+
+    avfilter_filter_samples(inlink->dst->outputs[0],
+                            avfilter_ref_buffer(curbuf, ~0));
+    avfilter_unref_buffer(insamplesref);
+}
+
+AVFilter avfilter_af_aconvert = {
+    .name          = "aconvert",
+    .description   = NULL_IF_CONFIG_SMALL("Convert the input audio to sample_fmt:channel_layout:packed_fmt."),
+    .priv_size     = sizeof(AConvertContext),
+    .init          = init,
+    .uninit        = uninit,
+    .query_formats = query_formats,
+
+    .inputs    = (AVFilterPad[]) {{ .name            = "default",
+                                    .type            = AVMEDIA_TYPE_AUDIO,
+                                    .filter_samples  = filter_samples,
+                                    .min_perms       = AV_PERM_READ, },
+                                  { .name = NULL}},
+    .outputs   = (AVFilterPad[]) {{ .name            = "default",
+                                    .type            = AVMEDIA_TYPE_AUDIO,
+                                    .config_props    = config_output, },
+                                  { .name = NULL}},
+};

libavfilter/af_aconvert_rematrix.c

+/*
+ * 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
+ * audio rematrixing functions, based on functions from libavcodec/resample.c
+ */
+
+#if defined(FLOATING)
+# define DIV2 /2
+#else
+# define DIV2 >>1
+#endif
+
+REMATRIX_FUNC_SIG(stereo_to_mono_packed)
+{
+    while (nb_samples >= 4) {
+        outp[0][0] = (inp[0][0] + inp[0][1]) DIV2;
+        outp[0][1] = (inp[0][2] + inp[0][3]) DIV2;
+        outp[0][2] = (inp[0][4] + inp[0][5]) DIV2;
+        outp[0][3] = (inp[0][6] + inp[0][7]) DIV2;
+        outp[0] += 4;
+        inp[0]  += 8;
+        nb_samples -= 4;
+    }
+    while (nb_samples--) {
+        outp[0][0] = (inp[0][0] + inp[0][1]) DIV2;
+        outp[0]++;
+        inp[0] += 2;
+    }
+}
+
+REMATRIX_FUNC_SIG(stereo_downmix_packed)
+{
+    while (nb_samples--) {
+        *outp[0]++ = inp[0][0];
+        *outp[0]++ = inp[0][1];
+        inp[0] += aconvert->in_nb_channels;
+    }
+}
+
+REMATRIX_FUNC_SIG(mono_to_stereo_packed)
+{
+    while (nb_samples >= 4) {
+        outp[0][0] = outp[0][1] = inp[0][0];
+        outp[0][2] = outp[0][3] = inp[0][1];
+        outp[0][4] = outp[0][5] = inp[0][2];
+        outp[0][6] = outp[0][7] = inp[0][3];
+        outp[0] += 8;
+        inp[0]  += 4;
+        nb_samples -= 4;
+    }
+    while (nb_samples--) {
+        outp[0][0] = outp[0][1] = inp[0][0];
+        outp[0] += 2;
+        inp[0]  += 1;
+    }
+}
+
+/**
+ * This is for when we have more than 2 input channels, need to downmix to mono
+ * and do not have a conversion formula available.  We just use first two input
+ * channels - left and right. This is a placeholder until more conversion
+ * functions are written.
+ */
+REMATRIX_FUNC_SIG(mono_downmix_packed)
+{
+    while (nb_samples--) {
+        outp[0][0] = (inp[0][0] + inp[0][1]) DIV2;
+        inp[0] += aconvert->in_nb_channels;
+        outp[0]++;
+    }
+}
+
+REMATRIX_FUNC_SIG(mono_downmix_planar)
+{
+    FMT_TYPE *out = outp[0];
+
+    while (nb_samples >= 4) {
+        out[0] = (inp[0][0] + inp[1][0]) DIV2;
+        out[1] = (inp[0][1] + inp[1][1]) DIV2;
+        out[2] = (inp[0][2] + inp[1][2]) DIV2;
+        out[3] = (inp[0][3] + inp[1][3]) DIV2;
+        out    += 4;
+        inp[0] += 4;
+        inp[1] += 4;
+        nb_samples -= 4;
+    }
+    while (nb_samples--) {
+        out[0] = (inp[0][0] + inp[1][0]) DIV2;
+        out++;
+        inp[0]++;
+        inp[1]++;
+    }
+}
+
+/* Stereo to 5.1 output */
+REMATRIX_FUNC_SIG(stereo_to_surround_5p1_packed)
+{
+    while (nb_samples--) {
+      outp[0][0] = inp[0][0];  /* left */
+      outp[0][1] = inp[0][1];  /* right */
+      outp[0][2] = (inp[0][0] + inp[0][1]) DIV2; /* center */
+      outp[0][3] = 0;          /* low freq */
+      outp[0][4] = 0;          /* FIXME: left surround: -3dB or -6dB or -9dB of stereo left  */
+      outp[0][5] = 0;          /* FIXME: right surroud: -3dB or -6dB or -9dB of stereo right */
+      inp[0]  += 2;
+      outp[0] += 6;
+    }
+}
+
+REMATRIX_FUNC_SIG(stereo_to_surround_5p1_planar)
+{
+    while (nb_samples--) {
+      *outp[0]++ = *inp[0];    /* left */
+      *outp[1]++ = *inp[1];    /* right */
+      *outp[2]++ = (*inp[0] + *inp[1]) DIV2; /* center */
+      *outp[3]++ = 0;          /* low freq */
+      *outp[4]++ = 0;          /* FIXME: left surround: -3dB or -6dB or -9dB of stereo left  */
+      *outp[5]++ = 0;          /* FIXME: right surroud: -3dB or -6dB or -9dB of stereo right */
+      inp[0]++; inp[1]++;
+    }
+}
+
+
+/*
+5.1 to stereo input: [fl, fr, c, lfe, rl, rr]
+- Left = front_left + rear_gain * rear_left + center_gain * center
+- Right = front_right + rear_gain * rear_right + center_gain * center
+Where rear_gain is usually around 0.5-1.0 and
+      center_gain is almost always 0.7 (-3 dB)
+*/
+REMATRIX_FUNC_SIG(surround_5p1_to_stereo_packed)
+{
+    while (nb_samples--) {
+        *outp[0]++ = inp[0][0] + (0.5 * inp[0][4]) + (0.7 * inp[0][2]); //FIXME CLIPPING!
+        *outp[0]++ = inp[0][1] + (0.5 * inp[0][5]) + (0.7 * inp[0][2]); //FIXME CLIPPING!
+
+        inp[0] += 6;
+    }
+}
+
+REMATRIX_FUNC_SIG(surround_5p1_to_stereo_planar)
+{
+    while (nb_samples--) {
+        *outp[0]++ = *inp[0] + (0.5 * *inp[4]) + (0.7 * *inp[2]); //FIXME CLIPPING!
+        *outp[1]++ = *inp[1] + (0.5 * *inp[5]) + (0.7 * *inp[2]); //FIXME CLIPPING!
+
+        inp[0]++; inp[1]++; inp[2]++; inp[3]++; inp[4]++; inp[5]++;
+    }
+}
+
+#undef DIV2
+#undef REMATRIX_FUNC_NAME
+#undef FMT_TYPE

libavfilter/allfilters.c

@@ -34,6 +34,7 @@ void avfilter_register_all(void)
         return;
     initialized = 1;
 
+    REGISTER_FILTER (ACONVERT,    aconvert,    af);
     REGISTER_FILTER (AFORMAT,     aformat,     af);
     REGISTER_FILTER (ANULL,       anull,       af);
     REGISTER_FILTER (ARESAMPLE,   aresample,   af);

libavfilter/avfilter.h

@@ -29,7 +29,7 @@
 #include "libavutil/rational.h"
 
 #define LIBAVFILTER_VERSION_MAJOR  2
-#define LIBAVFILTER_VERSION_MINOR 42
+#define LIBAVFILTER_VERSION_MINOR 43
 #define LIBAVFILTER_VERSION_MICRO  0
 
 #define LIBAVFILTER_VERSION_INT AV_VERSION_INT(LIBAVFILTER_VERSION_MAJOR, \