Plug-in Delay Compensation

[Nard]

Will Pro Tools LE ever come with Automatic Delay (Latency) Compensation for plug-ins (PDC or ADC?) ?

[nikki-k]

It already does...sort of...

As far as full compensation, so people can use thrid party DSP solutions...well...I would think they would if they developed their OWN brand of DSP solution that is similar. NOT like TDM, but rather just a simple DSP card. But I dont think there would be any $$$ in that, so probably not...

[tele_player]

nikki - sort of?

[DigiTechSupt]

This page has a link to info on how ADC works on LE systems:

http://www.digidesign.com/index.cfm?...54&itemid=5143

[nikki-k]

I was looking for this the other day!
Actually, it was a post made some time ago about delay comp in LE (pre - MP I think?)

Anyway...
There are some things missing from the paper IMO. For instance:
How is the delay computed that is being reported via Command + Click on vol/pk/dly field til displaying the delay amount? Common thinking is that the plugins themselves report the delay. If so, when plug-ins do not report their own delay properly, this method being used would fail, obviously.

I did a test; I created a new session with a 64 HW Buffer, @ 24/44.1k, and created 2 Audio tracks. On the first, I placed some audio with a nice intial transient. I then set the routing so that the Output of that Audio track was a mono Buss. I then set the INput of the second Audio track to that same mono Buss. Rec enable trck 2, and record. The recorded material is sample accurate to where it was on the first track. I then stuck Dverb on track one, and observed a *reported* delay via vol/pk/dly field of 0. Repeat record, and once again, sample accurate line-up of both waveforms. Sigh of relief, and now insert Smack. Now there is a delay of 1 sample showing in the vol/pk/dly field. However, when I record this to track 2, instead of the resulting recorded bit on track 2 being 1 sample late, it is in fact aligned perfectly. Why? (I might have simply pulled a complete brain fart here...)

Finally, just to clarify...
If the vol/pk/dly field is showing ANY value other than 0, this is due to a plugin on that track presenting a latency/delay that exceeds the current HW Buffer, correct? Or is this the actual amount of delay the plugin will incur?

The white paper states it is the amount of exceeding HW Buffer, however...if I use the previous example, with Smack instantiated on a track, and with an HW Buffer of 64, and see a reported 1 sample delay, why does it remain the same irregardless of changing the HW Buffer to a higher value?

Oh! One more! What is the amount of delay...round trip...with the various interfaces (002, Mbox, etc) @ various rates? I think this would be invaluable for those using hardware/outboard as inserts...

Sorry, but I would really love ot be able to relay info properly, and play "mythbuster" when I see inaccuracies regarding certain constant put-downs of PT LE. As such, any help is greatly appreciated!!!

[NickH]

Nikki,

I think (I am not near tools at the moment) that it’s the recording part that has fixed the 1 sample on smack.

PTLE automatically compensates for the delay if you are recording the track apparently.

Try it with an AUX track that is phase inverted, & no recording, the 1 sample should be present…

Cheers



N

[tele_player]

I'd heard of that document - but didn't really believe it could exist. I'm highly dubious of its technical accuracy with regard to plugin delay compensation.

[marty martin]

Why? It makes a lot of sense to me.

[tele_player]

It's specious. The buffer size (which all DAWS have) is what allows time for processing to occur, but the delay introduced by some plugins isn't caused by the time it takes to do the processing.

[marty martin]

It's specious.

I was hoping for a litle more detail there.
Have you tried the delay compensation methods described and still had problems with phasing, sample inaccuracies etc for example.
I have used these methods (ie either time adjuster or nudging) and have never had a problem, although it can be a little fiddly and time consuming.