Posts: 1,225
Threads: 51
Joined: 2019 Oct
Thanks: 943
Given 654 thank(s) in 384 post(s)
Country:
Currently having a bit of a fight with one particular LD and I'm not sure what's causing it, but maybe somebody who understands more about the LD format itself may be in a better position to follow what's happening.
My usual method for syncing the digital audio to the video is as follows:
1. Extract the in-sync 48 kHz analogue audio from raw LD video/audio capture (which I'll have made some effort to ensure actually is in sync, with AviSynth and VirtualDub2)
2. Resample that to 44.1 kHz and dither to 16-bit in RX to use as sync target for digital audio
3. Delete samples from start of digital capture to precisely sync with analogue audio, therefore exactly sync with video as well:
4. Verify no other changes to digital audio contents by checking in hex editor that it matches exactly apart from the missing zero bytes from start
Then of course I'd go on with the resync of the digital audio.
However, I've just hit a wall because I've found an LD this does not work on. It's NTSC like all the others I've captured thus far so it isn't a PAL thing. I'm using the exact same hardware for capture as well. But basically, I can align a point near the start fine, but by later on they've desynchronised again. With every other LD I've captured, sync remains rock solid to the sample from start to finish using this method, but with this disc, the analogue track drifts and gradually falls earlier than the digital one over time. It's admittedly a very small discrepancy at least, but it's consistent across sides 1 and 2 (which is the whole thing). By the end of side 2 for example, it's something like 86 samples out of sync (analogue ahead and/or digital behind).
What might be going on here?
One thing I'm wondering: if an LD has digital audio, is that just converted through an ADC and chucked out both the TOSLINK (as digital) and RCAs (as analogue), or is there a completely separate "legacy" analogue audio track also stored on the disc which is just played directly out of the RCA output? If so, I guess it could be that this disc has a slight discrepancy between the analogue and digital tracks, and it's just the first time I've found that on a disc. If the digital track is converted in real time and thrown out the analogue output though, something's going wrong somewhere in the chain and I have no idea why it's suddenly different. I guess my next move should be capturing another disc (ideally one I already captured, but that feels like potential time wasted) to see if the problem resurfaces without changing anything else. I already captured this one repeatedly with identical results on both side 1 and side 2.
deleted user
Unregistered
Thanks:
Given thank(s) in post(s)
The digital track is separate from the analogue tracks afaik, so that'd be my guess, the analogue on this disc having slight differences to the digital. The digital track is an actual digital track and doesn't go through an ADC. Well, technically I guess the analogue representation of it does before it is decoded. But it is still a lossless digital track.
Posts: 1,225
Threads: 51
Joined: 2019 Oct
Thanks: 943
Given 654 thank(s) in 384 post(s)
Country:
(2020-10-18, 01:12 PM)TomArrow Wrote: The digital track is separate from the analogue tracks afaik, so that'd be my guess, the analogue on this disc having slight differences to the digital.
Thanks, I suspected that was probably the case.
(2020-10-18, 01:12 PM)TomArrow Wrote: The digital track is an actual digital track and doesn't go through an ADC. Well, technically I guess the analogue representation of it does before it is decoded. But it is still a lossless digital track.
Yeah, that bit I knew; after all, it'd be odd if you could get bit perfect representations of the digital audio from the disc from different players otherwise, right?
Either way, this is extremely annoying because it's the first case of a disc I actually do want to keep the video from for viewing purposes and I am going to have to make a subjective decision re. sync on the digital audio, which isn't great considering I'm trying to be as reasonably objective as possible with this stuff. Sadly, when it comes to sync, there is no true objectivity but I at least want to avoid introducing any *further* error compared to the contents of the disc. As in, I found that sync was actually pretty bad on RESERVOIR DOGS (on both the fullscreen and widescreen discs), which I would have reproduced if in doubt but I was pretty sure the later French DVD had more accurate sync so I instead targeted its waveform rather than just following the original LD sync 100%. Here, there *is* no DVD, only LaserDiscs (one from the US, one from Japan). I do have both, so I guess I could compare them to see if sync differs by a few samples.
Frankly, this is probably a lot of fuss over nothing important, considering it's only drifting by 86 *samples*, which at 44.1 kHz is a little under 2 milliseconds. But y'know... it's me.
Posts: 2,048
Threads: 56
Joined: 2016 Dec
Thanks: 160
Given 1007 thank(s) in 611 post(s)
If the raw 44.1/48 captures sync up together then it's probably the result of IVTC, I've had this before where there will be a little loss of sync a few seconds into a side and then it will be solid for the rest of the side. I put it down to a slightly irregular pull down, it's no biggy
Posts: 1,225
Threads: 51
Joined: 2019 Oct
Thanks: 943
Given 654 thank(s) in 384 post(s)
Country:
2020-10-18, 03:20 PM
(This post was last modified: 2020-10-18, 03:26 PM by pipefan413.)
(2020-10-18, 03:07 PM)zoidberg Wrote: If the raw 44.1/48 captures sync up together then it's probably the result of IVTC, I've had this before where there will be a little loss of sync a few seconds into a side and then it will be solid for the rest of the side. I put it down to a slightly irregular pull down, it's no biggy
Maybe I'm not following your meaning correctly but there's been no IVTC applied to this one (yet, at least) because the source isn't film: it's a Genesis documentary so will almost certainly have been shot on PAL tape, but sadly the only PAL release is VHS afaik. The LDs are US & JP. (I have both, plus the tape.)
I'm not aligning the 48 kHz as is, I'm looking for accuracy to the sample so I'm resampling to 44.1 kHz first (not going to actually be using that for listening anyway so even if it isn't transparent it doesn't matter, I just need it chopped into slightly bigger increments). That way I can lock the 44.1 kHz digital in alongside it with some degree of confidence and check in throughout the track to see if it is still sample accurate. So far it always has been but not on this disc!
I guess I'm just going to sync the beginning to the sample and accept that it'll drift after that, but hope the drift is in the analogue track and it's actually accurate on the digital. We're only talking 1-2 ms here anyway. It's just bugging my happiness a wee bit.
Note: My understanding of pulldown is most likely a little ropey, being that I'm pretty new to dealing with NTSC video and it's basically been a huge headache from my POV. One thing that worries me a little bit is whether I can cut out the side change cleanly without affecting the accuracy of the pulldown during playback. Still a lot of experimental stuff happening over here and I'm probably making a tit of some things on the video side.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1465 thank(s) in 963 post(s)
Country:
(2020-10-18, 01:28 PM)pipefan413 Wrote: Frankly, this is probably a lot of fuss over nothing important, considering it's only drifting by 86 *samples*, which at 44.1 kHz is a little under 2 milliseconds. But y'know... it's me.
...and I thought I was crazy!
2ms... well, if you are a dolphin or Superman *MAYBE* you could hear it... if you are a (more or less) normal human being, you will NEVER notice it - IIRC the acceptable lip sync lost would be around half frame, so 2ms is ten times less... I dare you to detect it!!!
Trust me, for your peace of mind: do not care about 2ms sync difference - even more if spread around what, 20m? 30m?
Posts: 1,225
Threads: 51
Joined: 2019 Oct
Thanks: 943
Given 654 thank(s) in 384 post(s)
Country:
(2020-10-18, 03:34 PM)spoRv Wrote: I dare you to detect it!!!
Trust me, for your peace of mind: do not care about 2ms sync difference - even more if spread around what, 20m? 30m?
Again: I am fully aware this is a non-issue from a perception perspective. However, I am interested to understand *how* it has occurred, from a technological perspective. That's why I was asking about the analogue vs digital audio on the disc itself. I may run into another case like this where it matters more, as well, and it'd be good to know whether I should more closely follow the analogue or digital track for sync in that case.
In my experience the most noticeable sync problems by far are when the audio falls too early, so if in doubt I push it back so it lands a little later. I did that on RESERVOIR DOGS in a couple of places where I was not convinced that sync was correct on the LD in the first place. So here, the digital track is later and therefore I'm tempted to assume it's more "correct". But... it really isn't important given that we're talking about 86 samples at 44.1 kHz. It'll just irritate me knowing there's a tiny discrepancy.
If I'd never seen the waveforms I 100% guarantee I'd know nothing of the difference.
Posts: 7,153
Threads: 601
Joined: 2015 Jan
Thanks: 1081
Given 1465 thank(s) in 963 post(s)
Country:
Well, if both digital and analog tracks use the same mix, possible explanation is that digital is from a digital master, while analog is from analog master, where audio could "drift" a bit due to tape... I guess.
Posts: 1,225
Threads: 51
Joined: 2019 Oct
Thanks: 943
Given 654 thank(s) in 384 post(s)
Country:
(2020-10-18, 04:20 PM)spoRv Wrote: Well, if both digital and analog tracks use the same mix, possible explanation is that digital is from a digital master, while analog is from analog master, where audio could "drift" a bit due to tape... I guess.
Yeah, maybe. It's *so* close though. I would have guessed it would drift significantly more if so. Ach well, I'll just sync the beginning and move onto more pressing matters...
Thanks all @TomArrow @ zoidberg @ spoRv
Posts: 2,048
Threads: 56
Joined: 2016 Dec
Thanks: 160
Given 1007 thank(s) in 611 post(s)
I must have got my wires crossed when I read the first post. Yeah I'd most likely say it was mastered that way. If it's showing up the same way on repeated captures it's probably the disc.
IVTC/decimation really isn't that big of a deal, occasionally you may have to tweak settings and sometimes you'll get a disc with borked pulldown cadence but by and large it's pretty solid, 3:2 60i in, 23.976p out. Things like video black at side breaks won't cause problems.
(2020-10-18, 03:20 PM)pipefan413 Wrote: We're only talking 1-2 ms here anyway
C'mon man
|