44.1kHz to 48kHz laserdisc PCM track conversion

[spoRv]

It is common ground that the simplest way to convert 44.1kHz laserdisc PCM track to 48kHz is to capture it via digital cable by a bit-perfect audio card, then convert it to 48kHz

BUT

this introduces a very tiny error in the signal

AND

if we slowdown PAL to NTSC (or speedup 23.976fps to 24fps) we introduce further quality degradation

AND

if the PCM on NTSC laserdisc is not 44100Hz but 44056Hz then any bit-perfect capture is, well, not perfect!

THEN

why don't we capture at 48kHz 24bit both analog and PCM tracks via analog cables?

• no quality loss converting 44.1kHz to 48kHz
  
• less quality loss converting from 25fps to 23.976fps - or 23.976fps to 24fps
  
• if the laserdisc NTSC PCM frequency is really 44056Hz then capture it digitally at 44.1kHz, bit-perfect or not, is futile - unless it's possible to capture at 44056hz bit perfect
  

sure, it depends all from the player's DAC quality but AFAIK mid models had good DACs, up to excellent ones on top and high end models.

We must find out if a laserdisc PCM track taken from a good/excellent player's DAC analog output captured with a good audio card at 48kHz 24bit is better than a digital output captured bit-perfect at 44.1kHz then converted to 48Khz 24bit (24bit to be eventually edited - I'd always release at 48/16); I'd also add to the equation the PAL/NTSC slowdown affaire - even if PAL laserdisc market was much smaller than NTSC, we are still talking about several thousand titles...

Thoughts?

[dwalkerdon]

why not just convert the 44khz wav file to a 48khz wav file?

[spoRv]

No need to quote whole post...

PCM=WAV, so all the reasons I posted stand the same for WAV.

[bronan]

It is common ground that the simplest way to convert 44.1kHz laserdisc PCM track to 48kHz is to capture it via digital cable by a bit-perfect audio card, then convert it to 48kHz

BUT

this introduces a very tiny error in the signal

AND

if we slowdown PAL to NTSC (or speedup 23.976fps to 24fps) we introduce further quality degradation

AND

if the PCM on NTSC laserdisc is not 44100Hz but 44056Hz then any bit-perfect capture is, well, not perfect!

I disagree with 2 points here. The first is that resampling from 44->48 is going to add "error" and going from 44 -> analog -> 48 isn't. You are digitizing it twice then.

The other point is the myth of 44056Hz that you keep repeating. It makes no sense. If that was the case then tracks would go out of sync if you capture the PCM and drop the digital track on top of the analog. Plus the DTS decoding works capping at 441.

[ilovewaterslides]

That's pretty much what I've been doing with all of my laserdisc audio tracks
I'm using a Scarlett 2i2 3rd gen which seems to be more than adequate in term of DAC quality for the job.

[spoRv]

I disagree with 2 points here. The first is that resampling from 44->48 is going to add "error" and going from 44 -> analog -> 48 isn't. You are digitizing it twice then.

The other point is the myth of 44056Hz that you keep repeating. It makes no sense. If that was the case then tracks would go out of sync if you capture the PCM and drop the digital track on top of the analog. Plus the DTS decoding works capping at 441.

Converting from 44.1kHz to 48kHz does introduce an error, do we agree? I know for sure it's tiny - what I don't know if it's possible to hear it or not.

Agree that 44.1kHz -> analog -> 48kHz involves two passages, but are them worse than digital interpolating 44.1kHz to 48kHz? On theory it does seem so, but in practice? I mean, is the sum of both errors of ADC and DAC less than 44.1-48 conversion? Or, better, does the dual passage procedure sounds worse, or more noticeable?

Add to this the whole PAL-to-NTSC conversion and, less important but present, the 23.976-to-24.

About 44056Hz myth: you are right, I sound like a broken record, but this time I wrote "IF" at least!
Still, I can swear that I read this figure on VirtualDub during laserdisc capture; at the same time, you are right, it should not be that, or the DTS should not work... but what if, in some way, capturing with a no bit-perfect capture card, VirtualDub will read that figure?!? If only I can have access to my capture rig right now...

[bronan]

If you had a crappy DAC/ADC, or you use a bad resampling method either method could sound equally bad. But, if you used good hardware and conversion methods, I'm sure either would be acceptable in blind test.

BUT.. with different people capping stuff all the time, bit perfect PCM still makes the most sense in getting a common format without having to wonder about quality. And, as we saw in the other thread, using eac3to for resampling results in a super clean conversion that rivals software that costs hundreds.

[spoRv]

Well, I did imply (but admittedly not written) that the comparison should use the best hardware "mere mortals" could afford - hence very good laserdisc player but not necessarily a MUSE one, a good capture card - is it possible to get great cards below 100 quids (Euro, US/Canadian/Australian $, but not Yen...), a not-five-cent-a-meter analog cable etc. Vs using a free software.

It's undebatable that proper bit-perfect capture converted to 48kHz using free software leads to great final result, but as the hardware rig would be in most case the same - same player, same card, maybe adding a cheap but good analog cable if missing - would be interesting to make a comparison.

So if given the same player (again, high quality but not high end) and same capture card (around 100 quids used or new) we can get better result capturing at 48kHz, anyone could be encouraged to do it... if at the contrary the classic bit-perfect will be better, let's continue to use this - and if any difference could be spotted using only "la creme de la creme" hardware, let the lucky owner of the latter make that.

OR

if someone is crazy enough to capture twice (or once with capable hardware), let's do (and release) both, and let people decide which one they do prefer!

[zoidberg]

Capturing bit-perfect preserves the audio at the source so to speak. That way as resampling software continues to improve one can always go back to the original to make a new conversion vs the 'baked' in conversion done by analogue capture. Obviously if you're not burning to disc then 48kHz may not even be necessary especially if using a server/media player/htpc.
If you are determined to go the analogue route there may be some merit in using a newer DAC from a high end CD player or A/V receiver, something that outperforms even the best LD player, this could be fed by a 44.1kHz bit-perfect capture or the optical out of a LD player, essentially using it as a transport.

[bronan]

One other thing to be aware of is even if your MKV contains 44.1kHz audio, it doesn't mean its not being tampered with before it reaches your ears. I have all my HDMI devices routed into my LG OLED, then use an optical cable to my amp. After recording the output of the TV I discovered that everything is getting resampled to 48kHz regardless. I have a feeling its a worse conversion than an offline one..