Re: [Dvdauthor-users] SCR moves backwards, remultiplex input

[Cyrus P. <cy...@fb...>]

On 29 Nov 2005 at 13:52, Bob Forsman wrote:

> The way I read it, MPEG Audio Layer 1 audio frames are 384/samplerate
> seconds long, layer 2 = 1152/samplerate, and layer 3 is
> 1152/samplerate (unless you get into the low bitrates from 13818).

The the only permitted form of MPEG Audio for DVD-Video is MPEG-1/-2,
Layer 2, 48KHz.

> This interpretation depends on the definition of "sample" (do you need
> one sample per channel, or does one sample include the data for all
> channels).

neither :) They mean samples that the fft is fed with. Three temporal 
frames ("before", "current", "next") of 384 samples each (12 samples 
in each of the 32 subbands). 3 * 12 * 32 = 1152 samples.
Layer 1 has only one frame (no temporal redundancy removal), 
1 * 12 * 32 = 384 samples.

>   I just examined the PTSs on the 48KHz 192Kbit AC3 PID of a VOD
> transport stream.  It appears that the PTS advances by 2880/90KHz for
> each frame, which corresponds to an audio frame rate 31.25Hz: almost,
> but not exactly, 32Hz.

Correct. Its 32 msec/frame, not 32 frames/sec.

Back to topic:

> > The biggest problem with rewriting SCRs, PCRs, and PTS/DTS is
> > that video and audio give you frames in different sizes.

Doesn't matter, does it? :)

---------------------------------------------
Thought for the day:
  "Many a man's reputation would not know his character if they met 
on the
 street." -- Elbert Hubbard