Audio will be resampled from 48000Hz to 47952Hz: terminology?

[Kewl]

Just to be sure what is actually done to the audio stream: is the audio really "resampled" (i.e. creating new samples) or is the audio just played faster?

As the frame rate is going from 23.976 to 24 fps, is the same thing done to the audio, i.e. just being played faster? Because, if it's just played faster, it's not resampled per se. The original audio is considered as being "47952 Hz" and then played at 48000 Hz with the image going at 24 fps.

Or course, that would mean that the audio pitch is slightly going up.

[carl]

If we have some audio at 48kHz and want to run it faster, so that in 1 second we play 48048 samples from the source, there's no other way to do it (as far as I can see) than resampling the source so that we turn 48000 source samples into 47952. Then these are put into the DCP to run at 48kHz, so that things run fast.

So yes, the audio is pitch-shifted. But because we have to deliver samples at 48kHz we have to resample, I believe.

[Kewl]

Once a 23.976 fps source is DCP-fied and runs at 24 fps, is the DCP a bit shorter, duration-wise, than the 23.976 source?

[carl]

Right, we have the same number of video frames in the DCP as the source, but they just go past a little more often.

[Kewl]

I'm just trying to wrap my head around this... But if resampling is done, then pitch could be preserved.

Edit: head is now wrapped. Pitch should be preserved.

[Carsten]

No, technically speaking, it is not preserved end-to-end. It IS preserved during the resampling to 47952Hz - but during playback, it is actually played back at 48KHz, thus, faster, thus, pitched up.
However, this is REALLY negligible for the 23.976->24fps case. It may be more of a concern for 24<->25fps conversions.

Of course you can always use a separate tool for adjusting the pitch on the source. They, however, are not acoustically transparent as well.

[carl]

Perhaps what you are calling "resampling" I would call "time stretching".