Re: [libdv-dev] Re: [Kino-dev] libdv alters audio playback by forcing audio emphasis

[Charles Y. <cha...@pa...>]

On Tue, 2003-01-14 at 21:39, Dan Dennedy wrote:
> Roman, I hope Charlie will provide more info, but I know his
> calculations are based upon the spec for "locked audio" if not fully
> compliant ( I don't have the spec to validate it.) However, consumer
> cameras do not usually produce locked audio. Locked audio means there is
> a pre-determined sequence of sample counts for a consistent bitrate.

Hi Roman,

Yep - I base the calcs on locked samples per frame. 

This was reverse engineered from the very little NTSC footage that I
have as I too lack the DV spec (and some calcs [ie: 44100hz] are based
on a 'best fit' approach, rather than on actual footage or the spec).

As you pointed out, the distribution will be slightly different when you
reapply previously captured audio from an unlocked source, but the
difference is minor and, I would imagine, indiscernible.

The only scenario where this would not be the case which springs to mind
is if you dub two or more separate scenes and then subsequently
cut/paste one of the scenes. You *may* find that a sample or two may be
displaced.

Ultimately, the dependence on dubbing in kino for audio fx is something
which will reduce over time so I don't see too much gain in convoluting
the current work-around. My feeling is that if audio fx are applied in
place on the original footage, the need to calculate the number of
samples is removed (ie: only required when the 'effect' changes the
sampling rate or when the input has no audio at all).

Hope that helps,

Charlie