Posts: 271
Threads: 8
Joined: 2015 Feb
Thanks: 155
Given 118 thank(s) in 78 post(s)
Country:
Anyone ever encounter a Laserdisc which was sped up to exactly 60.00Hz instead of 59.94? I think that’s the case with an anime series I’m currently trying to sync (details below if you’re curious)*. The Blu-ray audio constantly lags behind the Laserdisc by about 1 millisecond for every 1 second of audio, so I’d have to stretch the LD audio by 60ms or so every minute to keep the audio from drifting out of sync. Sure enough, when I slow the LD audio down by setting the sample rate to 44056Hz (=44100Hz divided by 1.001, just like 60 divided by 1.001 is 59.94) in Sound Forge, the timings stopped drifting.
The Blu-ray is definitely 1080i59.94 and not exactly 1080i60 (according to MediaInfo at least), so I don’t think it’s the one that was mastered at the wrong speed. I’ve never had a problem like this with Laserdisc audio before and I haven’t changed anything in my capture setup, so I don’t think it was a problem with my capture either.
So to sync this audio, I’m first going to set the sample rate at 44056Hz (without actually resampling to this rate) and then resample to 48kHz instead of 44.1, since I’ll need to get 48Hz for BD compatibility anyway, and I don’t want to resample the audio twice. This will mean no 44.1kHz for this audio, but oh well. (Is it even possible to make an mkv with 1080i video?)
If anyone has any suggestions for a better way to go about this, please LMK, thanks!
*Details: Well, I’m pretty sure that the Laserdisc is still outputting NTSC at 59.94Hz since I don’t think a LD player can do exactly 60.00Hz, but for some reason the LD master must have been transferred to disc at 60.00Hz (there might be frame skips in the video, but I’ve only been paying attention to the audio recently).
The anime series is called Devilman Lady and I’m attempting to sync it to the recent Japanese Blu-ray release, which has a host of video problems (1080i, all 26 episodes compressed to hell on 2 BD-25s (which makes each one 5+ hours), poor black levels) but is still in HD and not an upscale, so it still looks better than the DVDs and Laserdiscs. But the BD audio also appears to be sourced from the 16mm film master’s optical track rather than the actual audio masters, so it’s in mono and sounds rather poor in comparison with the stereo Laserdisc audio.
Posts: 758
Threads: 34
Joined: 2018 Feb
Thanks: 648
Given 1064 thank(s) in 404 post(s)
Country:
2019-04-15, 07:07 PM
(This post was last modified: 2019-04-15, 07:15 PM by bronan.)
That's interesting, I've definitely synced a few with crazy drift like you describe. CRT TVs have quite a bit of flexibility when it comes to timing, so it wouldn't surprise me if that was the case. You could maybe do what I do when syncing 23.976 audio to a 24fps - change your reference source speed to match it, then sync based off that. At least that way you can keep the original quality of the audio. (assuming you can set the video to exactly 60hz in an MKV)
Posts: 1,554
Threads: 60
Joined: 2015 Jan
Thanks: 229
Given 626 thank(s) in 372 post(s)
Country:
2019-04-15, 08:45 PM
(This post was last modified: 2019-04-16, 12:49 AM by Chewtobacca.)
(2019-04-15, 06:34 PM)BusterD Wrote: Sure enough, when I slow the LD audio down by setting the sample rate to 44056Hz (=44100Hz divided by 1.001, just like 60 divided by 1.001 is 59.94) in Sound Forge, the timings stopped drifting.
That's because 44056Hz is the sampling rate for PCM on NTSC LDs. We just round it up to 44100Hz.
From Wikipedia:
Quote:Sound could be stored in either analog or digital format and in a variety of surround sound formats; NTSC discs could carry two analog audio tracks, plus two uncompressed PCM digital audio tracks, which were (EFM, CIRC, 16-bit and 44.056 kHz sample rate).[21] PAL discs could carry one pair of audio tracks, either analog or digital and the digital tracks on a PAL disc were 16-bit 44.1 kHz as on a CD; in the UK, the term "LaserVision" is used to refer to discs with analog sound, while "LaserDisc" is used for those with digital audio. The digital sound signal in both formats are EFM-encoded as in CD.[21]
(2019-04-15, 06:34 PM)BusterD Wrote: Is it even possible to make an mkv with 1080i video?
Yes, of course.
Posts: 758
Threads: 34
Joined: 2018 Feb
Thanks: 648
Given 1064 thank(s) in 404 post(s)
Country:
2019-04-15, 09:23 PM
(This post was last modified: 2019-04-15, 09:36 PM by bronan.)
That could possibly be misinformation Chew. I was googling around and have found people arguing both ways. Over on LDDB though I found a post with this quote:
Quote:Furthermore, since a lot of wrong information seems to be floating around the net including the allegedly 44056 Hz for PCM of NTSC LaserDiscs, the correct sample rates are:
AC3: 48000 Hz
PCM (and thus DTS as well of course): 44100 Hz
One of the arguments I saw basically said that its a confusion of video tapes that had an early form of PCM for NTSC. Here is some more info on that from this page:
Quote:BTW on to 44.1kHz sampling rates - it predates Redbook CD Audio and was used because the format developed FROM certain video recording research. That research by Sony and Phillips lead to the joint Redbook standard. Sony in the mid-1970s, developed from the Laserdisc development, audio only Audio discs that used 44056Hz. 44056Hz came from PCM Adaptors used by NTSC Color Video Tapes, and so Sony continued the usage of the 44056Hz that they were already using for PCM, since it was good enough to cover 20kHz auido. Phillips unlike Sony had to worry a lot more about covering PAL and NTSC, so they developed a sampling rate for their PCM research that could be derived from both PAL and NTSC, this lead them to 44100kHz. While they were combining technology to create the Compact Disc, they chose 44.1kHz because it covered NTSC and PAL.
Posts: 1,554
Threads: 60
Joined: 2015 Jan
Thanks: 229
Given 626 thank(s) in 372 post(s)
Country:
2019-04-15, 10:18 PM
(This post was last modified: 2019-04-15, 10:18 PM by Chewtobacca.)
It's possible that 44056Hz could be wrong, I suppose, but it is the exact figure upon which BusterD hit. The academic paper to which Wikipedia links is behind a paywall, so I can't check the source. I don't have any doubt that CDs are 44.1kHz; I've just always understood that 44.1kHz for NTSC LD was rounded up, albeit not by very much.
Posts: 271
Threads: 8
Joined: 2015 Feb
Thanks: 155
Given 118 thank(s) in 78 post(s)
Country:
Yeah, I remember seeing the 44056Hz thing before, I've mostly been on the side of it being misinformation as I've never had sync drift like this before(and I'm pretty sure that there would be sync drift in such cases as my audio card has to use the external sync of the digital input when capturing and can't be set for 44056Hz, so an actual 44056Hz signal would be recorded as 44100Hz with slightly sped up audio). I do seem to remember PAL LDs being reported as 44056Hz as well, this would seem to make sense as PAL is exactly 50.00Hz I believe, but I don't have any PAL LDs.
Anyway, thank you both for the help. I think I'll continue with my current plan of resampling the audio, as I believe that an exact 1080i60.00 video stream might cause problems in some setups, and Sound Forge is reportedly very good at resampling since it has very accurate interpolation.
Posts: 1,554
Threads: 60
Joined: 2015 Jan
Thanks: 229
Given 626 thank(s) in 372 post(s)
Country:
Yeah, 44056Hz sped up to 41000Hz on capture was something that crossed my mind. I also wondered if the engineers made the audio on 44056Hz equipment and sped it up to 41000Hz, dropping (video) frames to do so. As this is anime, which is notorious for having sections made at different frame-rates anyway, dropped frames would be hard to detect (and probably go unnoticed).
deleted user
Unregistered
Thanks:
Given thank(s) in post(s)
44056 is what you get when you multiply 44100 with 1000/1001. It's not uncommon to encounter these numbers because of this 1001 divisor. To illustrate, the common Blu Ray Framerate 23.976 is actually 24 * 1000/1001 or 24000/1001, rounded to 3 digits behind the comma. 29.97 is actually 30000/1001. 59.94 is actually 60000/1001.
I have no clue where this originates from (it's probably some trivial answer, but I just don't know). Some Blu Rays come at real 24fps, others at 24000/1001. The actual speed at which films run in cinema is 24 fps. PAL, for some reason, is always exactly 25 fps, at least I've never seen any kind of 25000/1001 thing anywhere. So it must be a US thing, heh.
As such, when you are for example syncing Cinema DTS audio to a 24000/1001 (or ~23.976) Blu Ray, I do a similar thing where I change the sample rate to 44100 * 1000/1001 = 44055.944.... = ~44056 Hz. Then I resample with a high quality resampler to 48000 for compatibility purposes.
So what this tells me is that the Laserdisc actually has the correct timing (as it would be in cinema) while the Blu Ray is too slow as per the 1000/1001 convention.
Of course if it's an anime series with interlacing, there is certainly a debate to be had about whether 60 or ~59,94 (respectively 30 vs. 29,97) is the "authentic" version, as television in the US always runs (or used to run) at an interlaced ~59,94 and there is no "clean" number as if it had run in cinema.
It's also worth mentioning that I think at least some Laserdiscs encapsulate 24/23,976 fps in an interlaced 30/29,97 fps stream, like it's done on TV (see many HDTV releases) and require an inverse telecine to restore the original 24 fps.
I can also imagine that if a Laserdisc is thus run on an older American TV that requires this framerate, the most intuitive way of doing that would be to play the Laserdisc at a slower speed so that no frames get lost, which would mean that ~44056 Hz is the effective sample rate at which some LD players connected with some types of TVs may play it. Though I'm no electrician or anything, so no idea how this plays out in reality.
Logically, this would also mean that a 48kHz AC-3 track on a US Laserdisc would be played back at ~ 47952 Hz.
Afterthought ... what if Laserdiscs (or at least some of them) are simply mastered at the "clean" speeds (without the 1000/1001 multiplier) to make their lives easier, and the actual playback speed is left to the hardware? Because we do know that the bitstreams of bit-perfectly captured AC-3/PCM tracks are 48000/44100 and not some weird other number, if only because AC-3 doesn't support any arbitrary samplerates (I think?).
Posts: 1,554
Threads: 60
Joined: 2015 Jan
Thanks: 229
Given 626 thank(s) in 372 post(s)
Country:
^ Yeah, we know the basics of video. It's not what the discussion is about.
(2019-04-16, 02:57 PM)TomArrow Wrote: So what this tells me is that the Laserdisc actually has the correct timing (as it would be in cinema) while the Blu Ray is too slow as per the 1000/1001 convention.
This is what the discussion is about, whether it's possible for an LD to run at the "correct timing". Such an LD would be unexpected because before true 24fps appeared as an optional frame-rate on BD, everything in the NTSC world had customarily run a little too slow.
Posts: 271
Threads: 8
Joined: 2015 Feb
Thanks: 155
Given 118 thank(s) in 78 post(s)
Country:
Heh, it's cool guys, as the OP I don't mind a brief recap on the 1000/1001 thing.
Don't have the info in front of me but I seem to remember it being due to the electrical frequency of power outlets in North America, while PAL developed its 50Hz rate for similar reasons. Somewhat surprising that Japan never developed two separate television systems considering it uses 50Hz electricity in the eastern half while the western half uses 60Hz. It's always been an interesting topic to me, but I've never read too much into it.
Kind of amazing that historical transfer of film to video wasn't more complicated than it already was. (I'm no fan of 3:2 pulldown judder but then again I'd take it over PAL speedup.)
|