Based on patches by clsid2 in ffdshow-tryout.
---
 configure               |    2 ++
 libavcodec/aacdec.c     |   31 +++++++++++++++++++++++++++----
 libavcodec/ac3dec.c     |   35 ++++++++++++++++++++++++++++++++++-
 libavcodec/dca.c        |   28 +++++++++++++++++++++++++---
 libavcodec/vorbis_dec.c |   23 ++++++++++++++++++++++-
 5 files changed, 110 insertions(+), 9 deletions(-)

diff --git a/configure b/configure
index dd44ba4..64779fd 100755
--- a/configure
+++ b/configure
@@ -95,6 +95,7 @@ Configuration options:
   --enable-x11grab         enable X11 grabbing [no]
   --disable-network        disable network support [no]
   --disable-mpegaudio-hp   faster (but less accurate) MPEG audio decoding [no]
+  --enable-audio-float     floating-point output from some audio decoders [no]
   --enable-gray            enable full grayscale support (slower color)
   --disable-swscale-alpha  disable alpha channel support in swscale
   --disable-fastdiv        disable table-based division
@@ -901,6 +902,7 @@ CONFIG_LIST="
     $COMPONENT_LIST
     aandct
     ac3dsp
+    audio_float
     avcodec
     avdevice
     avfilter
diff --git a/libavcodec/aacdec.c b/libavcodec/aacdec.c
index c9761a1..6db8252 100644
--- a/libavcodec/aacdec.c
+++ b/libavcodec/aacdec.c
@@ -549,7 +549,11 @@ static av_cold int aac_decode_init(AVCodecContext *avctx)
             return -1;
     }
 
+#if CONFIG_AUDIO_FLOAT
+    avctx->sample_fmt = AV_SAMPLE_FMT_FLT;
+#else
     avctx->sample_fmt = AV_SAMPLE_FMT_S16;
+#endif
 
     AAC_INIT_VLC_STATIC( 0, 304);
     AAC_INIT_VLC_STATIC( 1, 270);
@@ -574,7 +578,11 @@ static av_cold int aac_decode_init(AVCodecContext *avctx)
     // 60    - Required to scale values to the correct range [-32768,32767]
     //         for float to int16 conversion. (1 << (60 / 4)) == 32768
     ac->sf_scale  = 1. / -1024.;
+#if CONFIG_AUDIO_FLOAT
+    ac->sf_offset = 0;
+#else
     ac->sf_offset = 60;
+#endif
 
     ff_aac_tableinit();
 
@@ -2166,7 +2174,8 @@ static int aac_decode_frame_int(AVCodecContext *avctx, void *data,
         avctx->frame_size = samples;
     }
 
-    data_size_tmp = samples * avctx->channels * sizeof(int16_t);
+    data_size_tmp = samples * avctx->channels *
+                    (av_get_bits_per_sample_fmt(avctx->sample_fmt) / 8);
     if (*data_size < data_size_tmp) {
         av_log(avctx, AV_LOG_ERROR,
                "Output buffer too small (%d) or trying to output too many samples (%d) for this frame.\n",
@@ -2175,8 +2184,14 @@ static int aac_decode_frame_int(AVCodecContext *avctx, void *data,
     }
     *data_size = data_size_tmp;
 
-    if (samples)
+    if (samples) {
+#if CONFIG_AUDIO_FLOAT
+        ac->fmt_conv.float_interleave(data, (const float **)ac->output_data,
+                                      samples, avctx->channels);
+#else
         ac->fmt_conv.float_to_int16_interleave(data, (const float **)ac->output_data, samples, avctx->channels);
+#endif
+    }
 
     if (ac->output_configured)
         ac->output_configured = OC_LOCKED;
@@ -2494,7 +2509,11 @@ AVCodec ff_aac_decoder = {
     aac_decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("Advanced Audio Coding"),
     .sample_fmts = (const enum AVSampleFormat[]) {
-        AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE
+#if CONFIG_AUDIO_FLOAT
+        AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_NONE
+#else
+        AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE
+#endif
     },
     .channel_layouts = aac_channel_layout,
 };
@@ -2514,7 +2533,11 @@ AVCodec ff_aac_latm_decoder = {
     .decode = latm_decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("AAC LATM (Advanced Audio Codec LATM syntax)"),
     .sample_fmts = (const enum AVSampleFormat[]) {
-        AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE
+#if CONFIG_AUDIO_FLOAT
+        AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_NONE
+#else
+        AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE
+#endif
     },
     .channel_layouts = aac_channel_layout,
 };
diff --git a/libavcodec/ac3dec.c b/libavcodec/ac3dec.c
index 015ebae..1ae25ef 100644
--- a/libavcodec/ac3dec.c
+++ b/libavcodec/ac3dec.c
@@ -24,6 +24,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#define CONFIG_AUDIO_FLOAT 1
+
 #include <stdio.h>
 #include <stddef.h>
 #include <math.h>
@@ -189,7 +191,11 @@ static av_cold int ac3_decode_init(AVCodecContext *avctx)
     av_lfg_init(&s->dith_state, 0);
 
     /* set scale value for float to int16 conversion */
+#if CONFIG_AUDIO_FLOAT
+    s->mul_bias = 1.0f;
+#else
     s->mul_bias = 32767.0f;
+#endif
 
     /* allow downmixing to stereo or mono */
     if (avctx->channels > 0 && avctx->request_channels > 0 &&
@@ -204,7 +210,12 @@ static av_cold int ac3_decode_init(AVCodecContext *avctx)
         if (!s->input_buffer)
             return AVERROR(ENOMEM);
 
+#if CONFIG_AUDIO_FLOAT
+    avctx->sample_fmt = AV_SAMPLE_FMT_FLT;
+#else
     avctx->sample_fmt = AV_SAMPLE_FMT_S16;
+#endif
+
     return 0;
 }
 
@@ -1299,7 +1310,11 @@ static int ac3_decode_frame(AVCodecContext * avctx, void *data, int *data_size,
     const uint8_t *buf = avpkt->data;
     int buf_size = avpkt->size;
     AC3DecodeContext *s = avctx->priv_data;
+#if CONFIG_AUDIO_FLOAT
+    float *out_samples = data;
+#else
     int16_t *out_samples = (int16_t *)data;
+#endif
     int blk, ch, err;
     const uint8_t *channel_map;
     const float *output[AC3_MAX_CHANNELS];
@@ -1405,10 +1420,14 @@ static int ac3_decode_frame(AVCodecContext * avctx, void *data, int *data_size,
             av_log(avctx, AV_LOG_ERROR, "error decoding the audio block\n");
             err = 1;
         }
+#if CONFIG_AUDIO_FLOAT
+        s->fmt_conv.float_interleave(out_samples, output, 256, s->out_channels);
+#else
         s->fmt_conv.float_to_int16_interleave(out_samples, output, 256, s->out_channels);
+#endif
         out_samples += 256 * s->out_channels;
     }
-    *data_size = s->num_blocks * 256 * avctx->channels * sizeof (int16_t);
+    *data_size = s->num_blocks * 256 * avctx->channels * sizeof(*out_samples);
     return FFMIN(buf_size, s->frame_size);
 }
 
@@ -1435,6 +1454,13 @@ AVCodec ff_ac3_decoder = {
     .close = ac3_decode_end,
     .decode = ac3_decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("ATSC A/52A (AC-3)"),
+    .sample_fmts = (const enum AVSampleFormat[]) {
+#if CONFIG_AUDIO_FLOAT
+        AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_NONE
+#else
+        AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE
+#endif
+    },
 };
 
 #if CONFIG_EAC3_DECODER
@@ -1447,5 +1473,12 @@ AVCodec ff_eac3_decoder = {
     .close = ac3_decode_end,
     .decode = ac3_decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("ATSC A/52B (AC-3, E-AC-3)"),
+    .sample_fmts = (const enum AVSampleFormat[]) {
+#if CONFIG_AUDIO_FLOAT
+        AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_NONE
+#else
+        AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE
+#endif
+    },
 };
 #endif
diff --git a/libavcodec/dca.c b/libavcodec/dca.c
index e3c6466..457c5b4 100644
--- a/libavcodec/dca.c
+++ b/libavcodec/dca.c
@@ -1626,7 +1626,12 @@ static int dca_decode_frame(AVCodecContext * avctx,
     int lfe_samples;
     int num_core_channels = 0;
     int i;
+#if CONFIG_AUDIO_FLOAT
+    float *samples = data;
+#else
     int16_t *samples = data;
+#endif
+    int out_size;
     DCAContext *s = avctx->priv_data;
     int channels;
     int core_ss_end;
@@ -1812,9 +1817,10 @@ static int dca_decode_frame(AVCodecContext * avctx,
         return -1;
     }
 
-    if (*data_size < (s->sample_blocks / 8) * 256 * sizeof(int16_t) * channels)
+    out_size = 256 / 8 * s->sample_blocks * sizeof(*samples) * channels;
+    if (*data_size < out_size)
         return -1;
-    *data_size = 256 / 8 * s->sample_blocks * sizeof(int16_t) * channels;
+    *data_size = out_size;
 
     /* filter to get final output */
     for (i = 0; i < (s->sample_blocks / 8); i++) {
@@ -1833,7 +1839,11 @@ static int dca_decode_frame(AVCodecContext * avctx,
             }
         }
 
+#if CONFIG_AUDIO_FLOAT
+        s->fmt_conv.float_interleave(samples, s->samples_chanptr, 256, channels);
+#else
         s->fmt_conv.float_to_int16_interleave(samples, s->samples_chanptr, 256, channels);
+#endif
         samples += 256 * channels;
     }
 
@@ -1870,9 +1880,14 @@ static av_cold int dca_decode_init(AVCodecContext * avctx)
 
     for (i = 0; i < DCA_PRIM_CHANNELS_MAX+1; i++)
         s->samples_chanptr[i] = s->samples + i * 256;
-    avctx->sample_fmt = AV_SAMPLE_FMT_S16;
 
+#if CONFIG_AUDIO_FLOAT
+    avctx->sample_fmt = AV_SAMPLE_FMT_FLT;
+    s->scale_bias = 1.0 / 32768.0;
+#else
+    avctx->sample_fmt = AV_SAMPLE_FMT_S16;
     s->scale_bias = 1.0;
+#endif
 
     /* allow downmixing to stereo */
     if (avctx->channels > 0 && avctx->request_channels < avctx->channels &&
@@ -1909,5 +1924,12 @@ AVCodec ff_dca_decoder = {
     .close = dca_decode_end,
     .long_name = NULL_IF_CONFIG_SMALL("DCA (DTS Coherent Acoustics)"),
     .capabilities = CODEC_CAP_CHANNEL_CONF,
+    .sample_fmts = (const enum AVSampleFormat[]) {
+#if CONFIG_AUDIO_FLOAT
+        AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_NONE
+#else
+        AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE
+#endif
+    },
     .profiles = NULL_IF_CONFIG_SMALL(profiles),
 };
diff --git a/libavcodec/vorbis_dec.c b/libavcodec/vorbis_dec.c
index 5fa7be1..d55329b 100644
--- a/libavcodec/vorbis_dec.c
+++ b/libavcodec/vorbis_dec.c
@@ -962,7 +962,11 @@ static av_cold int vorbis_decode_init(AVCodecContext *avccontext)
     dsputil_init(&vc->dsp, avccontext);
     ff_fmt_convert_init(&vc->fmt_conv, avccontext);
 
+#if CONFIG_AUDIO_FLOAT
+    vc->scale_bias = 1.0f;
+#else
     vc->scale_bias = 32768.0f;
+#endif
 
     if (!headers_len) {
         av_log(avccontext, AV_LOG_ERROR, "Extradata missing.\n");
@@ -1007,7 +1011,12 @@ static av_cold int vorbis_decode_init(AVCodecContext *avccontext)
     avccontext->channels    = vc->audio_channels;
     avccontext->sample_rate = vc->audio_samplerate;
     avccontext->frame_size  = FFMIN(vc->blocksize[0], vc->blocksize[1]) >> 2;
+
+#if CONFIG_AUDIO_FLOAT
+    avccontext->sample_fmt  = AV_SAMPLE_FMT_FLT;
+#else
     avccontext->sample_fmt  = AV_SAMPLE_FMT_S16;
+#endif
 
     return 0 ;
 }
@@ -1635,9 +1644,14 @@ static int vorbis_decode_frame(AVCodecContext *avccontext,
                               len * ff_vorbis_channel_layout_offsets[vc->audio_channels - 1][i];
     }
 
+#if CONFIG_AUDIO_FLOAT
+    vc->fmt_conv.float_interleave(data, channel_ptrs, len, vc->audio_channels);
+#else
     vc->fmt_conv.float_to_int16_interleave(data, channel_ptrs, len,
                                            vc->audio_channels);
-    *data_size = len * 2 * vc->audio_channels;
+#endif
+    *data_size = len * vc->audio_channels *
+                 (av_get_bits_per_sample_fmt(avccontext->sample_fmt) / 8);
 
     return buf_size ;
 }
@@ -1664,5 +1678,12 @@ AVCodec ff_vorbis_decoder = {
     vorbis_decode_frame,
     .long_name = NULL_IF_CONFIG_SMALL("Vorbis"),
     .channel_layouts = ff_vorbis_channel_layouts,
+    .sample_fmts = (const enum AVSampleFormat[]) {
+#if CONFIG_AUDIO_FLOAT
+        AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_NONE
+#else
+        AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE
+#endif
+    },
 };
 
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to