How many samples delay can most people track with

[Raoul23]

In my Pro Tools HD system I always track with a lot of plugs on my vocal change sometimes it they add up to 84 sample dly other times if I only have a few plugins 22 sample dly. Just wondering at what "dly" do most people start to hear the latency, how high could I go

[mesaone]

Depends on sample rate. Remember, X number of samples at 48 kHz represents twice as much actual time as the same number of samples at 96 kHz.

[Raoul23]

Ok I track at 44.1 all the time, but if I was at 88.2 with 16 dly it would be less latency than 16 dly at 44.1

[albee1952]

Originally Posted by Raoul23

Ok I track at 44.1 all the time, but if I was at 88.2 with 16 dly it would be less latency than 16 dly at 44.1

Generally(based on my reading), doubling the sample rate cuts the latency in half(roughly). But, its still also dependent on the buffer and the interface/connection(USB vs. firewire, vs PCIe/TB). For me(HD/Native card and HD IO boxes); I work at 48K with the buffer set to 64 and, during tracking, any plugin with more than 10 samples is left inactive. None of my players ever mention latency, and I have heard that this scenario with a 128 buffer is still going to feel "good" to almost any player. Having said all this, I recently had a client decide he needed one more overdub on a fully mixed session, with 60+ tracks and a ton of UAD plugins(all over 1100 samples of latency). I set the buffer to 32 and rolled and he never mentioned latency. Does that mean he didn't feel any? Tough to say(math is not my strong suit)

[Raoul23]

But in HD (not NativeHD) the playback buffer doesn't effect plugin latency so I'm told. If a plugin causes 6 samples at 128 buffer it should still be 6 samples at 1024 buffer I understand. So is a chain of plugins with say 80 samples dly in total noticeable on a HD system ??

[albee1952]

Given that HD has lower inherent latency, its probably going to be fine. Its pretty easy to test it out yourself(and I highly recommend everyone check it out). Slap on some phones, put up a mic with that same chain of plugins and give a listen. For most of us, this will reveal any "ugliness" in the time it takes to say "test one two"(I would recommend moving to 48K over 44.1K though)

[Raoul23]

Why 48k I've heard lots of fors and against due to dithering maths at bounce down.
I wanted to track with Waves c4 on and this introduces quite a large amount of dly, so I'm wondering what's the lowest dly people have had and what's the highest before clients have noticed the latency

[Michael Zull]

In the past, when tracking at 44.1/24, I could run the buffer as high as 128 without a single client noticing any latency in the cans. Once you bump it up to 256 it's incredibly noticeable...down at 128, if I zero in and listen I can hear the delay, but it's so small that the vast majority of people won't hear it. These days I'm recording at 96/24 with a buffer of 64 and, again, no noticeable latency, but I'm not normally running any inserts on the record tracks or printing any inserts routed through an Aux in.

In some of the pop & urban sessions, I used to run a comp and EQ insert on the vocal track, IIRC they were Waves C1 and Waves Q6, and again, no complaints of latency. This was probably 4 years ago that I was doing alot of this type of vocal tracking, so I can't remember my session setups exactly. Often times I was printing Autotune (can't remember which version, maybe 7?) during the vocal tracking as well, and again, no latency complaints.

Oh, and all of this has been on PT Vanilla, versions 9 through 11, which of course are running RTAS or AAX64 plugs. Prior to V9 I was doing almost exclusively rock/metal/punk or post production work, so I wasn't using inserts during tracking, at least not very often. Back then I was running a 002, now I'm running an RME Babyface.

[mesaone]

Dither shouldn't be a consideration when choosing a sample rate, since it is only applied when reducing bit depth.

I wish I could give a useful answer to your question, but I don't track with plug-ins so my answer would be "zero".

[albee1952]

Originally Posted by Raoul23

Why 48k I've heard lots of fors and against due to dithering maths at bounce down.
I wanted to track with Waves c4 on and this introduces quite a large amount of dly, so I'm wondering what's the lowest dly people have had and what's the highest before clients have noticed the latency

As said, dither only enters when lowering bit depth(like from a 24bit session to a 16 bit final product). The stories of poor SRC are from the early days when the process was not so good. These days, conversion is done so well, I doubt you would ever be able to hear it under normal circumstances. As for 48 vs 44.1; #1-it pushes the upper frequency a tad higher. Now you say we can't hear 20K, never mind 23K, so why bother? Because the higher you push the Nyquist filter, the higher the phase shift starts(and it starts well below the top frequency). #2-video almost always wants 48K(if your product will end up in video world). #3-the performance hit is negligible(if even measurable). Just my 2 cents