[FFmpeg-devel] [PATCH] fix clicks in ADPCM IMA AMV decoder

[Aurelien Jacobs]

Vladimir Voroshilov wrote:

> 2007/10/4, Aurelien Jacobs <aurel at gnuage.org>:
> > Vladimir Voroshilov wrote:
> >
> > > 2007/10/3, Vitor Sessak <vitor1001 at gmail.com>:
> > > > Hi
> > > >
> > > > Vladimir Voroshilov wrote:
> > > > > Hi, All
> > > > >
> > > > > Attached simle patches fixes clicks in audio when playing AMV files.
> > > >
> > > > Cool. Is it still not loud enough?
> > >
> > > Hm. It is enough loud for me now.
> > >
> > >
> > > > >
> > > > > Index: libavcodec/adpcm.c
> > > > > ===================================================================
> > > > > --- libavcodec/adpcm.c        (revision 10652)
> > > > > +++ libavcodec/adpcm.c        (working copy)
> > > > > @@ -1184,7 +1184,7 @@
> > > > >          break;
> > > > >      case CODEC_ID_ADPCM_IMA_AMV:
> > > > >      case CODEC_ID_ADPCM_IMA_SMJPEG:
> > > > > -        c->status[0].predictor = *src;
> > > > > +        c->status[0].predictor = (int16_t)AV_RL16(src);
> > > > >          src += 2;
> > > >
> > > > c->status[0].predictor = bytestream_get_le16(&src);
> > >
> > > I've made simple (imho) solution which also comply with other code style.
> >
> > This new patch is not good. It could segfault on arch which requires strict
> > alignment if src is not properly aligned.
> > The solution proposed by Vitor is the right one, and it's simpler (it saves
> > you the src += 2).
> 
> Thanks. Didn't know that.
> 
> > BTW: I wonder if step_index shouldn't be read as an int16_t as well ?
> 
> According to multimedia wiki it should.
> This also also shows why fourth byte is always zero (index<=88)
> Included in patch.
> 
> Index: libavcodec/adpcm.c
> ===================================================================
> --- libavcodec/adpcm.c	(revision 10652)
> +++ libavcodec/adpcm.c	(working copy)
> @@ -1184,10 +1184,8 @@
>          break;
>      case CODEC_ID_ADPCM_IMA_AMV:
>      case CODEC_ID_ADPCM_IMA_SMJPEG:
> -        c->status[0].predictor = *src;
> -        src += 2;
> -        c->status[0].step_index = *src++;
> -        src++;  /* skip another byte before getting to the meat */
> +        c->status[0].predictor = (signed short)bytestream_get_le16(&src);
> +        c->status[0].step_index = bytestream_get_le16(&src);

I suppose the signed short cast is useless. If so, please remove it.
If you remove it, you could also align the = sign.
Except those remarks, patch is OK. I think you can apply it yourself.

Aurel