The Importance of the "Original-Speed" Audio

[jerryshadoe]

Those of us that work on various projects are all very familiar with having to deal with material that is sped-up or slowed-down due to NTSC/PAL conversions. Granted, most of the time we are slowing PAL audio back down to IVTCfilm, but there are those times where is has to be done the other way around.

In most cases, the audio has to be speed corrected and sometimes pitch has to be addressed too. This is where a rather large amount of degradation occurs that most people are not aware of. If possible, it's always best to use the "original-speed" audio, even if it has to be re-synced to match the project that your working on.

I've been trying to figure out the best way to demonstrate this to people visually, so that the importance of this becomes apparent. After ripping my CD collection recently, I've been having to go through the audio and label some of the unique stuff I have (personal recordings, etc) While doing this, I discovered almost 80 CDs that were copies made on professional equipment of my old audio cassettes. While going through all of it, I was curious about what the spectral analysis of the audio will show me. Even though in the higher frequencies, technically, it's all "noise" - there is a frequency response all the way up to 44.1Khz.

This made me realize that I can easily show how the audio is affected by slowing down or speeding up the audio WITH a pitch correction (In cases where ONLY the SPEED is being manipulated, this does NOT apply)
So, I opened up one of the tracks and slowed it down by 4% applying a pitch correction, saved it. Re-opened the original track, sped it up by 4% applying a pitch correction and saved that. Opened all three in Spek and took some screenshots...



Here is the original, unaltered audio track:



Here is the same audio track, sped up: (notice the frequency cut-off that occurs)



Here is the same audio track, slowed down: (notice the even larger difference in cut-off - THIS happens every time PAL audio is slowed down AND pitch correction has to be applied)






In most cases, the difference will not be this glaringly obvious, which is why I intentionally used this as an example since the audio fills 44.1Khz all the way through - unlike anything we'd see in any project for a film/tv show/etc. Regardless, even though the difference might not be this obvious, it STILL occurs which is why it's important to always try to use the "original-speed" audio.

Hope this helps

[Stamper]

Thanks, the problem though is you can't tell if the "original speed audio" is not a pal convert, this happens from time to time.
Studio Canal, especially, are culprits to this, they mess the audio left and right.

[spoRv]

Thanks for the insight, Jerry.

Changing perspective, we always manipulate video, from compressed sources, upscale enhance denoise etc. and back to compressed final file, and usually result is very satisfing... and for us, humans, sight is the sense we most rely on, about 90%... and, if we, video experts, could not spot a problem in the former video, I wonder how this audio problem could be noticeable by the average Joe, using both TV speakers and a cheap surround setup - I'm quite sure that an audio expert with high-end amp and speakers could spot the eventual problems...

In a perfect world, we will have lossless audio and video, and so no quality loss in any conversion... but it that world, all the movies would be perfect, hence we would not be needed, so... let's stay here, for the moment!

[FrankT]

Does that mean we have to rely on NTSC DVDs if there is no Blu-ray?

[jerryshadoe]

IF there is an NTSC version, then it would be the preferable one for a project (in most, but NOT all cases)

There are occasions, where even after correcting the speed and pitch of PAL speed-up, the PAL-sourced audio will sound better (and have better dynamics/frequency response) IF the PAL-sourced audio had a higher bitrate (example: PAL AC3 is encoded at 448kbps vs NTSC AC3 being encoded at 384kbps) OR if the original (NTSC speed) audio is severely clipped. I have worked on numerous projects (actually am having this problem with the audio on one of the projects I'm trying to finish up) where the original audio is too hot and clipped, making it instantly inferior to the non-clipped PAL audio even after the speed/pitch correction. There are examples that I can show of this being true if anyone is interested

[Moshrom]

Yes, I've come across PAL tracks that sound better than their NTSC counterparts, pitch differences aside.

A good example is Arrow's new Dekalog Blu-ray set. It's been touted as definitive, but it's actually incorrectly pitched - 0.7 semitones too high, or the equivalent of PAL speed-up without the speed-up. The Criterion blu-rays are pitched 0.7 semitones too low, but this is understandable since the Criterion discs are 23.976 fps for North American compatibility. The Polish blu-rays, however, are 25 fps like the Arrow, but they're correctly pitched -- however, the audio is presented as 384 kbps AC-3. In theory, the best Blu-ray tracks in terms of fidelity alone would be the Arrow tracks with their pitch corrected and tempo preserved. (Also, I wouldn't use anything but iZotope Radius to adjust pitch/tempo, especially when changing just one of the two.) I wrote more about this here: http://blah-ray.blogspot.com/2017/01/dekalog-1989.html

But in practice, noise reduction must also be taken into account, as it's far more significant than any differences in compression (even, say, lossless versus 192 kbps AC-3). So, the best sounding audio for the Dekalog actually comes from the old Warner DVDs, which, despite being encoded at 192 kbps AC-3, is considerably more detailed. I sync stuff like this all the time - tracks from DVDs that sound better than their Blu-ray counterparts due to differences in mastering/processing. If it's mono or stereo, I typically just transcode from lossy to lossless to preserve whatever I can. If there were any interest, I'd happily share them, but folks here seem to reduce anything but lossless as automatically inferior, so...

[Chewtobacca]

If there were any interest, I'd happily share them, but folks here seem to reduce anything but lossless as automatically inferior, so...

I'm not sure where you get that impression from.  Folks here show plenty of interest in lossy DTS and AC-3 from a variety of sources.  As long as there is an explanation of why lossy tracks are superior (or simply different), people are usually very receptive of them.

[PDB]

Yeah agree with Chew. We just like anything rare even if lossy.

[Moshrom]

Well, they wouldn't necessarily be rare... Most are readily available on old DVDs. I've just synced them.

[PDB]

Well rare is too much how about different.

Do any of your syncs stand out as different mixes?