Does nobody understand PDC on here?

Discussion of music production, audio, equipment and any related topics, either with or without Ableton Live
Post Reply
re:dream
Posts: 4559
Joined: Fri Dec 28, 2007 9:42 am
Location: Hoerikwaggo's sunset side...
Contact:

Re: Does nobody understand PDC on here?

Post by re:dream » Tue Dec 17, 2013 12:51 pm

This is great and really useful.

I am a complete n00b so I can only comment on clarity.\

There are a couple of places where it seems repetitive or confusing - when I get an hour or two I will PM you more detailed comments.

Angstrom
Posts: 14686
Joined: Mon Oct 04, 2004 2:22 pm
Contact:

Re: Does nobody understand PDC on here?

Post by Angstrom » Tue Dec 17, 2013 2:51 pm

the part where you try to explain around this topic is not very accurate
You should significantly alter this portion.
The reason some do not phase cancel is because the compensation is a tiny bit off in certain devices at certain sampling rates. Live can only calculate to whole numbers of samples, e.g. 1 or 2, but not 1.5 samples. Some Live devices use oversampling algorithms which can introduce these 'float value' numbers. This can occur in devices which have a Hi-Quality mode in which internal oversampling takes place. During downsampling these float values are introduced. The Saturator for example can give a slight inaccuracy this way. EQ Eight used to have this problem in Live 8. However in Live 9, EQ Eight does not introduce any latency, and thus phase cancels correctly, in oversampling mode. Remember of course that the default mode of the new EQ Eight is equivalent in sound quality to the old Hi-Quality mode.
That section is not accurate, it is vague and misleading.
It seems you are unfamiliar with the terms and concepts here, so you might be better to cut it right down to what you understand rather than spread bad information.

I won't pretend to know much more, but I know that this reads like a blind man talking about the colour of his dog. Personally I am squinting at the problems through thick glasses, so I wouldn't want to pretend to be the provider of authoritative correct details.
---
My suggestion to describe this topic

PDC/latency correction generally operates in cases where a process causes the audio frequencies in a stream to be delayed equally (by the same amount), this is a linear offset which you can rectify by offseting the stream by that number of milliseconds to put it back into sync.
Some processes however affect the audio frequencies non-linearly, where different frequencies are delayed by different amounts. The bass frequencies slower than the high frequencies. When the delay is not equal across the audio spectrum it becomes very difficult to mover the audio back into perfect alignment, because the offset is non-linear. Oversampled filters are notorious for being non-linear processes and the Live 8 oversampled EQ8 produced some non-linearity and therefore produced slight phase issues. In Live 9 the EQ8 oversampling was upgraded, partly in order to avoid this effect.

---

As I said, my maths is terrible, I am not an authority, and I have little real knowledge of the true workings here.
If you do want to comment publicly on the topic, you might want to read up on non-linear gain in oversampling filters, and find out a bit more about how floating point numbers operate, and how upsampling and downsampling and integration work. Good luck to you. Alternatively you could ask Andy Cytomic for a layman's explanation, as he created the linear gain version of EQ8.

Just as a final point, AFAIK the basic EQ8 does not have any latency issues in either L8 or L9. It's the old oversampling version which had a problem. The other oversampled devices still exhibit non-linear phase issues in L9.

8O
Posts: 5502
Joined: Fri Mar 28, 2008 9:29 am
Location: Berlin

Re: Does nobody understand PDC on here?

Post by 8O » Tue Dec 17, 2013 5:14 pm

Non-technical comment: you've used "it's" a few times when it should be "its".
Image

H20nly
Posts: 15844
Joined: Sat Oct 27, 2007 9:15 pm
Location: The Wild West

Re: Does nobody understand PDC on here?

Post by H20nly » Tue Dec 17, 2013 5:42 pm

PDC, killed r crops n took r jawbs!


Image

Angstrom
Posts: 14686
Joined: Mon Oct 04, 2004 2:22 pm
Contact:

Re: Does nobody understand PDC on here?

Post by Angstrom » Tue Dec 17, 2013 6:01 pm

my problem is not with the underlying facts, its the application of those facts which is strangely distorted.

The article is for explaining to a confused general user why oversampling can cause a strange sort of latency, which is more tricky to deal with. Talking about samplerates and floating point issues in vague terms just leads people off deeper into the woods. For example: That intro sentence implies this is a problem only at "certain" sample rates, whereas it's a problem at all sample rates - but is more evident at lower samplerates. This could equally be said of basic plugin latency, its also less of a problem at high sample rares. So as an explanation of the crucial difference this is not a good point of entry. It may well be true, but it is not phrased well and not pertinent to the task of explaining.

Then it states that " Live can only calculate whole numbers of samples", very true, but as stated it reads as Live-specific deficit which is causing this particular issue. Consider : could a user come away with the wrong ideas from this :"I heard Live has bad pdc bcause it cant calculate the upsampled data correctly" . Not correct.

The core concept for an end user is that this sort of offset is not equally distrubuted across the frequency range, as is the case with a more normal latency issue. And that is why its hard to compensate for.

theophilus
Posts: 520
Joined: Fri Mar 06, 2009 3:54 pm

Re: Does nobody understand PDC on here?

Post by theophilus » Wed Dec 18, 2013 9:48 pm

afaik, there is no way to shift midi clock timing. a lot of plugins that let you do these things, you can shift the lfo curve in the plugin itself though.

Post Reply