Tracking with Buffersize 32?

[Bob Olhsson]

Just an A/D and D/A wired back to back with no computer in between is a latency problem. Musicians don't identify it but it takes them a lot longer to get a good take.

Interestingly, they do lots better singing or playing to a loudspeaker. I've wondered if a large increase in digital latency might be a solution.

[LDS]

Originally Posted by RobertDorn

Yes, silicon Macs are beasts! But as stated above, tracking without DSP / and a dedicated DSP path, requires to keep an eye on the native playback buffer and on the number of samples that plugins used in the recording session add. For those who don't want to be bothered with that and like to punch in something on a fully loaded session with 3000+ sample delay plugins and a native playback buffer of 2048 samples, luckily also in 2023, we still have DSP options :)

Except that every AAX-DSP plug-in has higher inherent latency than their native counterparts…

[JFreak]

Originally Posted by LDS

Except that every AAX-DSP plug-in has higher inherent latency than their native counterparts…

Yes. Here is an example (posted in 2016).

Originally Posted by propower

All are 96K/24bit session values (though mostly these have stayed the same when I look at 44.1kHz)....

AVID:
Channel Strip = 10
EQ3 = 10
Pultec = 10
All BF = 10
Dyn 3 = 10
Fairchild = 10
Impact = 10
Pro Comp = 10
Pro Exp = 10
Mod Delay = 10
Real Tape delay = 10
RectiFi = 10
Real Tape Sat = 10
SciFi = 10
purple 77 = 13
Smack = 11
Pro Multiband = 34
Pro Limiter = 73
Gray Comp = 34
Black Spring = 66
Dverb = 34
Reverb One = 66
Revibe = 66
Space = 66
Studio Reverb = 66
Tape Echo = 66
SansAmp = 63
Black shiny Wah = 34
C1, Flanger, Orange Phasor, Vibe Phasor = 66
Real Tape Flanger = 4811 !!!!!
Roto Speaker = 34
Distortion Pedals (all) = 66
Pro Subharmonic = 394


METRIC HALO
Channel Strip = 10

SOFTUBE
Summit EQ = 34
Summit Grand Ch = 38
Tube Tech Classic Ch = 38
Tube tech pultecs = 42
FET Comp = 34

From a fellow source I have talked with about this...
PLUGIN-ALLIANCE
XLA-3 = 76
BX Limiter = 348
Refinement = 76
Saturation V2 = 56
All Elysias = 76
SPL PassEq = 60
SPL Vitalizer = 60
VSM-3 = 76
All others are 34.

From memory of recent demo of McDsp all were at least 34....

On a side note - amazing how the OLD Avid ones are all a mere 10 samples (were even less in the TDM versions). Now - no one does better than 34 samples. Three of these takes a stellar 0.52ms basic mixer to the same as HD Native at 96kHz 64buffer...

[JFreak]

Be prepared for more popcorn.

This just in.

It says:

Soothe Live was built from scratch to suit the live environment and offers a latency of only 64 samples, resulting in a roundtrip latency of less than 1 ms or 95 samples on the S6L.

Poor performers, they suffer from minimum of 95 sample roundtrip latency... must be hard to sing in tune.

This also means S6L has roundtrip latency of 31 samples with zero plugins. Must be terrible comb filtering in the artists' in-ears.

[685]

Sorry to revamp this but how are you guys actually measuring your native Protools system's RTL(round trip latency) in 2023?
I do feel as if I'm asking a rhetorical question but I'm asking to clarify a few things. With auto delay comp turned off, I normally send the audio out a physical output, re-record it on another input, than measure the space in between the start of each audio file by samples before converting to milliseconds.




.

[JFreak]

I stopped counting when I realised from countless live (FOH + monitoring) gigs that the problems begin at about 350 samples (@48kHz).

[685]

Originally Posted by JFreak

I stopped counting when I realised from countless live (FOH + monitoring) gigs that the problems begin at about 350 samples (@48kHz).

But if you had to measure it today(Sept 8th, 2023) what steps would you use?

[JFreak]

Only way to reliably measure roundtrip latency is an analog cable from output to input. And it needs to be with delay compensation turned off.

[685]

Originally Posted by JFreak

Only way to reliably measure roundtrip latency is an analog cable from output to input. And it needs to be with delay compensation turned off.

So you're pretty sure that turning off Protools auto delay compensation, sending audio out and then back in using an analog cable, and recording to a separate audio track will show some type of latency between the original audio and the new audio?

[JFreak]

it should not if you do not use hardware insert for it, but better play safe if you are measuring.