[FFmpeg-devel] [PATCH 5/7] avformat/mpegts: add s337m support

[Kieran Kunhya]

On Fri, 13 Dec 2024, 13:41 Niklas Haas, <ffmpeg at haasn.xyz> wrote:

> On Fri, 13 Dec 2024 01:37:34 +0000 Kieran Kunhya via ffmpeg-devel <
> ffmpeg-devel at ffmpeg.org> wrote:
> > On Fri, 13 Dec 2024, 00:39 Michael Niedermayer, <michael at niedermayer.cc>
> > wrote:
> >
> > > Hi
> > >
> > > On Thu, Dec 12, 2024 at 12:54:29PM +0000, Nicolas Gaullier wrote:
> > > [...]
> > > > This is a significant work with some weird things like dealing with a
> > > truncated sample rate
> > > > and droping one sample from time to time to keep sync. And with no
> active
> > > > maintainer of the dolby_e decoder, so that won't help.
> > >
> > > why would the decoder need to drop a sample from time to time ?
> > > the timestamps on audio should be enough.
> > >
> > > Yes it would mean the samplerate and the timestamps and the number of
> > > samples
> > > would mismatch but that can be corrected by a resampler following
> after the
> > > decoder or by other means, none of which really are specific to this.
> > > Recording audio with cheap hardware is unlikely to have its sample rate
> > > synchronized to an atomic clock either.
> > >
> > > but maybe i misunderstand something ?
> > >
> > > thx
> > >
> >
> > One of the features of Dolby E is output is resampled to a 1602, 1601,
> > 1602, 1601, 1602 cadence ( 48000*5*1001/30000 = 8008)  number of samples
> so
> > that it doesn't drift relative to 29.97 video.
>
> Does the spec mandate a specific resampling algorithm?
>

There is no specification.

Kieran

>