Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
* [FFmpeg-devel] PATCH - libmad MP3 decoding support
@ 2022-05-01 23:16 David Fletcher
  2022-05-02  7:19 ` Paul B Mahol
  2022-05-02  9:47 ` Nicolas George
  0 siblings, 2 replies; 14+ messages in thread
From: David Fletcher @ 2022-05-01 23:16 UTC (permalink / raw)
  To: ffmpeg-devel

[-- Attachment #1: Type: text/plain, Size: 773 bytes --]

Please find attached a patch adding support for MP3 decoding using
libmad. This is against release ffmpeg-5.0.1, and works for libmad
0.15.1b (the most recent available).

I hope that this is useful. Some context - I found the FFmpeg inbuild
fixed point MP3 decoder produced very distorted audio for low bitrate
streams (56k/s and lower). This was on an ARMv4 CPU for which I needed
to make some adaptation to the fixed point maths routines which
currently target more recent CPUs. I was not able to resolve the
distortion issue and it was easier to integrate libmad decoding which is
already optimised for very fast decoding on the ARMv4 CPU. More info
about this application here:
http://www.megapico.co.uk/sharpfin/mediaserver.html.

Best regards, David.

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: libmad.patch --]
[-- Type: text/x-patch; name="libmad.patch", Size: 9107 bytes --]

diff -Nur ./ffmpeg-5.0.1/configure ./ffmpeg-5.0.1-mad/configure
--- ./ffmpeg-5.0.1/configure	2022-04-04 15:40:22.000000000 +0100
+++ ./ffmpeg-5.0.1-mad/configure	2022-05-01 22:50:01.435432431 +0100
@@ -244,6 +244,7 @@
   --enable-libklvanc       enable Kernel Labs VANC processing [no]
   --enable-libkvazaar      enable HEVC encoding via libkvazaar [no]
   --enable-liblensfun      enable lensfun lens correction [no]
+  --enable-libmad          enable MP3 decoding via libmad [no]
   --enable-libmodplug      enable ModPlug via libmodplug [no]
   --enable-libmp3lame      enable MP3 encoding via libmp3lame [no]
   --enable-libopencore-amrnb enable AMR-NB de/encoding via libopencore-amrnb [no]
@@ -1772,6 +1773,7 @@
     frei0r
     libcdio
     libdavs2
+    libmad
     librubberband
     libvidstab
     libx264
@@ -3334,6 +3336,7 @@
 libilbc_encoder_deps="libilbc"
 libkvazaar_encoder_deps="libkvazaar"
 libmodplug_demuxer_deps="libmodplug"
+libmad_decoder_deps="libmad"
 libmp3lame_encoder_deps="libmp3lame"
 libmp3lame_encoder_select="audio_frame_queue mpegaudioheader"
 libopencore_amrnb_decoder_deps="libopencore_amrnb"
@@ -6569,6 +6572,7 @@
 fi
 
 enabled libmodplug        && require_pkg_config libmodplug libmodplug libmodplug/modplug.h ModPlug_Load
+enabled libmad            && require libmad "mad.h" mad_decoder_init -lmad
 enabled libmp3lame        && require "libmp3lame >= 3.98.3" lame/lame.h lame_set_VBR_quality -lmp3lame $libm_extralibs
 enabled libmysofa         && { check_pkg_config libmysofa libmysofa mysofa.h mysofa_neighborhood_init_withstepdefine ||
                                require libmysofa mysofa.h mysofa_neighborhood_init_withstepdefine -lmysofa $zlib_extralibs; }
diff -Nur ./ffmpeg-5.0.1/libavcodec/Makefile ./ffmpeg-5.0.1-mad/libavcodec/Makefile
--- ./ffmpeg-5.0.1/libavcodec/Makefile	2022-01-14 18:45:39.000000000 +0000
+++ ./ffmpeg-5.0.1-mad/libavcodec/Makefile	2022-04-06 22:56:06.000000000 +0100
@@ -1054,6 +1054,7 @@
 OBJS-$(CONFIG_LIBILBC_DECODER)            += libilbc.o
 OBJS-$(CONFIG_LIBILBC_ENCODER)            += libilbc.o
 OBJS-$(CONFIG_LIBKVAZAAR_ENCODER)         += libkvazaar.o
+OBJS-$(CONFIG_LIBMAD_DECODER)             += libmaddec.o
 OBJS-$(CONFIG_LIBMP3LAME_ENCODER)         += libmp3lame.o
 OBJS-$(CONFIG_LIBOPENCORE_AMRNB_DECODER)  += libopencore-amr.o
 OBJS-$(CONFIG_LIBOPENCORE_AMRNB_ENCODER)  += libopencore-amr.o
diff -Nur ./ffmpeg-5.0.1/libavcodec/allcodecs.c ./ffmpeg-5.0.1-mad/libavcodec/allcodecs.c
--- ./ffmpeg-5.0.1/libavcodec/allcodecs.c	2022-01-14 18:45:39.000000000 +0000
+++ ./ffmpeg-5.0.1-mad/libavcodec/allcodecs.c	2022-04-06 22:54:45.000000000 +0100
@@ -744,6 +744,7 @@
 extern const AVCodec ff_libgsm_ms_decoder;
 extern const AVCodec ff_libilbc_encoder;
 extern const AVCodec ff_libilbc_decoder;
+extern const AVCodec ff_libmad_decoder;
 extern const AVCodec ff_libmp3lame_encoder;
 extern const AVCodec ff_libopencore_amrnb_encoder;
 extern const AVCodec ff_libopencore_amrnb_decoder;
diff -Nur ./ffmpeg-5.0.1/libavcodec/codec_id.h ./ffmpeg-5.0.1-mad/libavcodec/codec_id.h
--- ./ffmpeg-5.0.1/libavcodec/codec_id.h	2022-01-14 18:45:39.000000000 +0000
+++ ./ffmpeg-5.0.1-mad/libavcodec/codec_id.h	2022-04-06 22:52:18.000000000 +0100
@@ -516,6 +516,7 @@
     AV_CODEC_ID_HCA,
     AV_CODEC_ID_FASTAUDIO,
     AV_CODEC_ID_MSNSIREN,
+    AV_CODEC_ID_LIBMAD,
 
     /* subtitle codecs */
     AV_CODEC_ID_FIRST_SUBTITLE = 0x17000,          ///< A dummy ID pointing at the start of subtitle codecs.
diff -Nur ./ffmpeg-5.0.1/libavcodec/libmaddec.c ./ffmpeg-5.0.1-mad/libavcodec/libmaddec.c
--- ./ffmpeg-5.0.1/libavcodec/libmaddec.c	1970-01-01 01:00:00.000000000 +0100
+++ ./ffmpeg-5.0.1-mad/libavcodec/libmaddec.c	2022-04-12 19:00:21.000000000 +0100
@@ -0,0 +1,181 @@
+/*
+ * MP3 decoder using libmad
+ * Copyright (c) 2022 David Fletcher
+ *
+ * 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
+ */
+
+#include <mad.h>
+
+#include "libavutil/channel_layout.h"
+#include "libavutil/common.h"
+#include "avcodec.h"
+#include "internal.h"
+#include "decode.h"
+
+#define MAD_BUFSIZE (32 * 1024)
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+
+typedef struct libmad_context {
+  uint8_t input_buffer[MAD_BUFSIZE+MAD_BUFFER_GUARD];
+  struct mad_synth  synth; 
+  struct mad_stream stream;
+  struct mad_frame  frame;
+  struct mad_header header;
+  int got_header;
+}libmad_context;		
+
+/* utility to scale and round samples to 16 bits */
+static inline signed int mad_scale(mad_fixed_t sample)
+{
+  /* round */
+  sample += (1L << (MAD_F_FRACBITS - 16));
+  
+  /* clip */
+  if (sample >= MAD_F_ONE)
+    sample = MAD_F_ONE - 1;
+  else if (sample < -MAD_F_ONE)
+    sample = -MAD_F_ONE;
+  
+  /* quantize */
+  return sample >> (MAD_F_FRACBITS + 1 - 16);
+}
+
+static av_cold int libmad_decode_init(AVCodecContext *avc)
+{
+    libmad_context *mad = avc->priv_data;
+
+    mad_synth_init  (&mad->synth);
+    mad_stream_init (&mad->stream);
+    mad_frame_init  (&mad->frame);
+    mad->got_header = 0;
+
+    return 0;
+}
+
+static av_cold int libmad_decode_close(AVCodecContext *avc)
+{
+    libmad_context *mad = avc->priv_data;
+
+    mad_synth_finish(&mad->synth);
+    mad_frame_finish(&mad->frame);
+    mad_stream_finish(&mad->stream);
+
+    mad = NULL;
+    
+    return 0;
+}
+
+static int libmad_decode_frame(AVCodecContext *avc, void *data,
+                          int *got_frame_ptr, AVPacket *pkt)
+{
+  AVFrame *frame = data;
+  libmad_context *mad = avc->priv_data;
+  struct mad_pcm *pcm;
+  mad_fixed_t const *left_ch;
+  mad_fixed_t const *right_ch;
+  int16_t *output;
+  int nsamples;
+  int nchannels;
+  size_t bytes_read = 0;
+  size_t remaining = 0;
+  
+  if (!avc)
+    return 0;
+    
+  if (!mad)
+    return 0;
+      
+  remaining = mad->stream.bufend - mad->stream.next_frame;
+  memmove(mad->input_buffer, mad->stream.next_frame, remaining);
+  bytes_read = MIN(pkt->size, MAD_BUFSIZE - remaining);
+  memcpy(mad->input_buffer+remaining, pkt->data, bytes_read);
+  
+  if (bytes_read == 0){
+    *got_frame_ptr = 0;
+    return 0;
+  }
+  
+  mad_stream_buffer(&mad->stream, mad->input_buffer, remaining + bytes_read);
+  mad->stream.error = 0;
+  
+  if(!mad->got_header){
+    mad_header_decode(&mad->header, &mad->stream);
+    mad->got_header = 1;
+    avc->frame_size = 32 * (mad->header.layer == MAD_LAYER_I ? 12 :	\
+			    ((mad->header.layer == MAD_LAYER_III &&	\
+			      (mad->header.flags & MAD_FLAG_LSF_EXT)) ? 18 : 36));
+    avc->sample_fmt = AV_SAMPLE_FMT_S16;
+    if(mad->header.mode == MAD_MODE_SINGLE_CHANNEL){
+      avc->channel_layout = AV_CH_LAYOUT_MONO;
+      avc->channels = 1;
+    }else{
+      avc->channel_layout = AV_CH_LAYOUT_STEREO;
+      avc->channels = 2;
+    }
+  }
+
+  frame->channel_layout = avc->channel_layout;
+  frame->format = avc->sample_fmt;
+  frame->channels = avc->channels;
+  frame->nb_samples = avc->frame_size; 
+    
+  if ((ff_get_buffer(avc, frame, 0)) < 0)
+    return 0;
+ 
+  if (mad_frame_decode(&mad->frame, &mad->stream) == -1) {
+    *got_frame_ptr = 0;
+    return mad->stream.bufend - mad->stream.next_frame;
+  }
+
+  mad_synth_frame (&mad->synth, &mad->frame);
+  
+  pcm = &mad->synth.pcm;
+  output = (int16_t *)frame->data[0];
+  nsamples = pcm->length;
+  nchannels = pcm->channels;
+  left_ch = pcm->samples[0];
+  right_ch = pcm->samples[1];
+  while (nsamples--) {
+    *output++ = mad_scale(*(left_ch++));
+    if (nchannels == 2) {
+      *output++ = mad_scale(*(right_ch++));
+    }
+    //Players should recognise mono and play through both channels
+    //Writing the same thing to both left and right channels here causes
+    //memory issues as it creates double the number of samples allocated.
+  }
+  
+  *got_frame_ptr = 1;
+ 
+  return mad->stream.bufend - mad->stream.next_frame;
+}
+
+AVCodec ff_libmad_decoder = {
+    .name           = "libmad",
+    .long_name      = NULL_IF_CONFIG_SMALL("libmad MP3 decoder"),
+    .wrapper_name   = "libmad",
+    .type           = AVMEDIA_TYPE_AUDIO,
+    .id             = AV_CODEC_ID_MP3,
+    .sample_fmts    = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE },
+    .capabilities   = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_CHANNEL_CONF,
+    .priv_data_size = sizeof(libmad_context),
+    .init           = libmad_decode_init,
+    .close          = libmad_decode_close,
+    .decode         = libmad_decode_frame
+};
+

[-- Attachment #3: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 14+ messages in thread
* Re: [FFmpeg-devel] PATCH - libmad MP3 decoding support
@ 2022-05-04 23:47 David Fletcher
  2022-05-06 18:29 ` Lynne
  0 siblings, 1 reply; 14+ messages in thread
From: David Fletcher @ 2022-05-04 23:47 UTC (permalink / raw)
  To: FFmpeg development discussions and patches

[-- Attachment #1: Type: text/plain, Size: 10150 bytes --]

> Andreas Rheinhardt wrote:
> 
> David Fletcher:
> > Following today's posts about help with submitting patches I realised I
> > sent the libmad patch yesterday in the wrong format. Apologies, I was
> > not familiar with the git format patches.
> > 
> > Hopefully the attached version is now in the correct format against the
> > current master branch.
> > 
> > The bug report about why this exists is at the following link, including
> > a link to sample distorted audio from decoding an mp3 stream on ARMv4
> > hardware: https://trac.ffmpeg.org/ticket/9764
> > 
> > Best regards, David.
> > 
> 
> > 
> > diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c
> > index c47133aa18..e3df6178c8 100644
> > --- a/libavcodec/allcodecs.c
> > +++ b/libavcodec/allcodecs.c
> > @@ -744,6 +744,7 @@ extern const FFCodec ff_libcodec2_decoder;
> >  extern const FFCodec ff_libdav1d_decoder;
> >  extern const FFCodec ff_libdavs2_decoder;
> >  extern const FFCodec ff_libfdk_aac_encoder;
> > +extern const AVCodec ff_libmad_decoder;
> >  extern const FFCodec ff_libfdk_aac_decoder;
> >  extern const FFCodec ff_libgsm_encoder;
> >  extern const FFCodec ff_libgsm_decoder;
> 
> This should look weird to you.

Now you've pointed it out - yes, it does! This error was matched by a similar
use of AVCodec in place of FFCodec in the libmaddec.c file. It seems these
structures have enough in common that I'd got away with it running without
noticing this. 

> 
> > 
> > diff --git a/libavcodec/codec_id.h b/libavcodec/codec_id.h
> > index 8b317fa121..be70f4a71c 100644
> > --- a/libavcodec/codec_id.h
> > +++ b/libavcodec/codec_id.h
> > @@ -519,6 +519,7 @@ enum AVCodecID {
> >      AV_CODEC_ID_FASTAUDIO,
> >      AV_CODEC_ID_MSNSIREN,
> >      AV_CODEC_ID_DFPWM,
> > +    AV_CODEC_ID_LIBMAD,
> >  
> >      /* subtitle codecs */
> >      AV_CODEC_ID_FIRST_SUBTITLE = 0x17000,          ///< A dummy ID
> > pointing at the start of subtitle codecs.
> 
> This makes no sense: Your decoder is still expected to decode MP3 and
> not a new, previously unsupported format.

This was left over from some development experiments while I was working out
how to integrate libmad. It was not used and needed to be removed. The
ff_libmad_decoder structure in libmaddec.c already used AV_CODEC_ID_MP3 and
never mentioned AV_CODEC_ID_LIBMAD.


> 
> > diff --git a/libavcodec/libmaddec.c b/libavcodec/libmaddec.c
> > new file mode 100644
> > index 0000000000..7082c53f4d
> > --- /dev/null
> > +++ b/libavcodec/libmaddec.c
> > @@ -0,0 +1,181 @@
> > +/*
> > + * MP3 decoder using libmad
> > + * Copyright (c) 2022 David Fletcher
> > + *
> > + * 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
> > + */
> > +
> > +#include <mad.h>
> > +
> > +#include "libavutil/channel_layout.h"
> > +#include "libavutil/common.h"
> > +#include "avcodec.h"
> > +#include "internal.h"
> > +#include "decode.h"
> > +
> > +#define MAD_BUFSIZE (32 * 1024)
> > +#define MIN(a, b) ((a) < (b) ? (a) : (b))
> > +
> > +typedef struct libmad_context {
> > +    uint8_t input_buffer[MAD_BUFSIZE+MAD_BUFFER_GUARD];
> > +    struct mad_synth  synth; 
> > +    struct mad_stream stream;
> > +    struct mad_frame  frame;
> > +    struct mad_header header;
> > +    int got_header;
> > +}libmad_context;		
> > +
> > +/* utility to scale and round samples to 16 bits */
> > +static inline signed int mad_scale(mad_fixed_t sample)
> > +{
> > +     /* round */
> > +     sample += (1L << (MAD_F_FRACBITS - 16));
> > + 
> > +     /* clip */
> > +     if (sample >= MAD_F_ONE)
> > +         sample = MAD_F_ONE - 1;
> > +     else if (sample < -MAD_F_ONE)
> > +         sample = -MAD_F_ONE;
> > +    
> > +     /* quantize */
> > +     return sample >> (MAD_F_FRACBITS + 1 - 16);
> > +}
> > +
> > +static av_cold int libmad_decode_init(AVCodecContext *avc)
> > +{
> > +     libmad_context *mad = avc->priv_data;
> > +
> > +     mad_synth_init  (&mad->synth);
> > +     mad_stream_init (&mad->stream);
> > +     mad_frame_init  (&mad->frame);
> > +     mad->got_header = 0;
> > +
> > +     return 0;
> > +}
> > +
> > +static av_cold int libmad_decode_close(AVCodecContext *avc)
> > +{
> > +     libmad_context *mad = avc->priv_data;
> > +
> > +     mad_synth_finish(&mad->synth);
> > +     mad_frame_finish(&mad->frame);
> > +     mad_stream_finish(&mad->stream);
> > +
> > +     mad = NULL;
> 
> This is pointless as the lifetime of this pointer ends with returning
> from this function anyway.

Thanks for pointing this out - removed now. 

> 
> > +    
> > +     return 0;
> > +}
> > +
> > +static int libmad_decode_frame(AVCodecContext *avc, void *data,
> > +                          int *got_frame_ptr, AVPacket *pkt)
> > +{
> > +     AVFrame *frame = data;
> > +     libmad_context *mad = avc->priv_data;
> > +     struct mad_pcm *pcm;
> > +     mad_fixed_t const *left_ch;
> > +     mad_fixed_t const *right_ch;
> > +     int16_t *output;
> > +     int nsamples;
> > +     int nchannels;
> > +     size_t bytes_read = 0;
> > +     size_t remaining = 0;
> > +     
> > +     if (!avc)
> > +	 return 0;
> 
> A codec can presume the AVCodecContext to not be NULL.
> 
> > +     
> > +     if (!mad)
> > +	 return 0;
> > +     
> 
> Similarly, every codec can presume the Codec's private data to be
> allocated (and be retained between calls to the same AVCodecContext
> instance).
> (Furthermore, the initialization of mad presumes avc to be not NULL,
> which makes the above check pointless.)

I'd played it safe by including checks on things - but since you've confirmed
that AVCodecContext will not be NULL and private data will always be
allocated these can be removed. 


> 
> > +     remaining = mad->stream.bufend - mad->stream.next_frame;
> > +     memmove(mad->input_buffer, mad->stream.next_frame, remaining);
> > +     bytes_read = MIN(pkt->size, MAD_BUFSIZE - remaining);
> > +     memcpy(mad->input_buffer+remaining, pkt->data, bytes_read);
> > +     
> > +     if (bytes_read == 0){
> > +	 *got_frame_ptr = 0;
> > +	 return 0;
> > +     }
> > +     
> > +     mad_stream_buffer(&mad->stream, mad->input_buffer, remaining +
> > bytes_read);
> > +     mad->stream.error = 0;
> > +     
> > +     if(!mad->got_header){
> > +	 mad_header_decode(&mad->header, &mad->stream);
> > +	 mad->got_header = 1;
> > +	 avc->frame_size = 32 * (mad->header.layer == MAD_LAYER_I ? 12 :
> > \
> > +				 ((mad->header.layer == MAD_LAYER_III &&
> > \
> > +				   (mad->header.flags &
> > MAD_FLAG_LSF_EXT)) ? 18 : 36));
> > +	 avc->sample_fmt = AV_SAMPLE_FMT_S16;
> > +	 if(mad->header.mode == MAD_MODE_SINGLE_CHANNEL){
> > +	     avc->channel_layout = AV_CH_LAYOUT_MONO;
> > +	     avc->channels = 1;
> > +	 }else{
> > +	     avc->channel_layout = AV_CH_LAYOUT_STEREO;
> > +	     avc->channels = 2;
> > +	 }
> > +     }
> > +     
> > +     frame->channel_layout = avc->channel_layout;
> > +     frame->format = avc->sample_fmt;
> > +     frame->channels = avc->channels;
> > +     frame->nb_samples = avc->frame_size; 
> > +     
> > +     if ((ff_get_buffer(avc, frame, 0)) < 0)
> > +	 return 0;
> 
> Return the error.

Modified to return the error now

> 
> > +     
> > +     if (mad_frame_decode(&mad->frame, &mad->stream) == -1) {
> > +	 *got_frame_ptr = 0;
> > +	 return mad->stream.bufend - mad->stream.next_frame;
> > +     }
> > +     
> > +     mad_synth_frame (&mad->synth, &mad->frame);
> > +     
> > +     pcm = &mad->synth.pcm;
> > +     output = (int16_t *)frame->data[0];
> > +     nsamples = pcm->length;
> > +     nchannels = pcm->channels;
> > +     left_ch = pcm->samples[0];
> > +     right_ch = pcm->samples[1];
> > +     while (nsamples--) {
> > +	 *output++ = mad_scale(*(left_ch++));
> > +	 if (nchannels == 2) {
> > +	     *output++ = mad_scale(*(right_ch++));
> > +	 }
> > +	 //Players should recognise mono and play through both channels
> > +	 //Writing the same thing to both left and right channels here
> > causes
> > +	 //memory issues as it creates double the number of samples
> > allocated.
> > +     }
> > +     
> > +     *got_frame_ptr = 1;
> > +     
> > +     return mad->stream.bufend - mad->stream.next_frame;
> > +}
> > +
> > +AVCodec ff_libmad_decoder = {
> > +    .name           = "libmad",
> > +    .long_name      = NULL_IF_CONFIG_SMALL("libmad MP3 decoder"),
> > +    .wrapper_name   = "libmad",
> > +    .type           = AVMEDIA_TYPE_AUDIO,
> > +    .id             = AV_CODEC_ID_MP3,
> > +    .sample_fmts    = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16,
> > AV_SAMPLE_FMT_NONE },
> > +    .capabilities   = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_CHANNEL_CONF,
> > +    .priv_data_size = sizeof(libmad_context),
> > +    .init           = libmad_decode_init,
> > +    .close          = libmad_decode_close,
> > +    .decode         = libmad_decode_frame
> > +};
> 
> This should not even compile with latest master.
> 
> - Andreas

Hi Andreas,

Many thanks for your review and help to improve this. I've inserted
comments/replies above, and an updated version of the patch file is attached.
It compiles and runs, and I've just converted a range of mp3 files with it. 

I've realised I was suppressing some warnings when I compiled previously,
including about deprecated code in use of avc->channel_layout and
avc->channels. I've made updates to use the avc->ch_layout and
avc->ch_layout.nb_channels in the updated patch. Could you take another look?

Best regards, David.










[-- Attachment #2: libmad_v2-git-master.patch --]
[-- Type: application/octet-stream, Size: 9250 bytes --]

[-- Attachment #3: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 14+ messages in thread
* Re: [FFmpeg-devel] PATCH - libmad MP3 decoding support
@ 2022-05-07 11:43 David Fletcher
  0 siblings, 0 replies; 14+ messages in thread
From: David Fletcher @ 2022-05-07 11:43 UTC (permalink / raw)
  To: FFmpeg development discussions and patches

On Fri May 6 21:29:51 EEST 2022 Lynne dev at lynne.ee wrote:

> I disagree with this patch. The native decoder should be fixed instead,
> or if it's not able to be fixed, users should use specialized libraries.
> We're not adding a yet another decoding library to fix an issue on
> very old ARM chips.

> I suspect the problem is with the MDCT. We already have a better and
> more precise fixed-point MDCT in libavutil/tx. The plan was to replace
> the MDCT in mp3 once lavu/tx has more assembly, but if it helps,
> should be fine to replace it now.

Hi Lynne,

I don't have a strong feeling either way. The patch (an earlier version)
is already doing what I needed in the application with the Reciva
radios. I posted it here to contribute back to FFmpeg, but if there's a
better way to fix the problem that's great.

If there's a better version of MDCT integrated into the native mp3 fixed
point decoder I can test that on the ARMv4 hardware to see if the
decoding bug is fixed. Certainly these are old chips but still lots of
them around - a big part of them being useful is the assembly
optimisations otherwise everything just slows down too much.

Best regards, David.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 14+ messages in thread

end of thread, other threads:[~2022-05-07 11:44 UTC | newest]

Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-05-01 23:16 [FFmpeg-devel] PATCH - libmad MP3 decoding support David Fletcher
2022-05-02  7:19 ` Paul B Mahol
2022-05-02 17:41   ` David Fletcher
2022-05-02 18:46     ` Neal Gompa
2022-05-02 22:50   ` David Fletcher
2022-05-03 18:34     ` David Fletcher
2022-05-04  2:16       ` Andreas Rheinhardt
2022-05-02  9:47 ` Nicolas George
2022-05-02 17:45   ` David Fletcher
2022-05-02 18:36     ` Timo Rothenpieler
2022-05-02 19:24     ` Martin Storsjö
2022-05-04 23:47 David Fletcher
2022-05-06 18:29 ` Lynne
2022-05-07 11:43 David Fletcher

Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \
		ffmpegdev@gitmailbox.com
	public-inbox-index ffmpegdev

Example config snippet for mirrors.


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git