Space Designer Inconsistency

[ski]

Backround: I regularly double-check the recording delay setting by playing back a mix through my (digital) board and recording it right back into Logic. Then I put the re-record out of phase with the original and usually the two cancel perfectly (silence).

 

Now, to the situation at hand... Today I recorded a mix that used an SD reverb using the default setting. In other words, I loaded up a SD plug and didn't select a preset. It had the right sound, so I used it as is. I printed the mix through my board as usual. Then I double-checked the recording delay setting by putting that mix out of phase and playing the track down. Perfect cancellation. I saved the song in that state.

 

I closed that song, worked on something else, then re-opened the first song. When I hit PLAY I expected to hear nothing because the song was saved with the mix phase cancelling with the track. But... this was not the case!

 

Instead, I heard only reverb. It was clear that the dry signal was cancelling tho. So... I removed the SD and then loaded it up again. Sure enough, perfect cancellation again.

 

I thought Logic was supposed to remember the state of plugs without having to actually save the settings... I wonder what this says for the performance of other plugs that give you the sound right immediately after they're loaded up, or, without saving a preset...

 

Thoughts? Comments?

 

-=sKi=-

[Threlly]

Why are you doing this in the first place ?

Is the delay that large ?

[ski]

Why are you doing this in the first place ? Is the delay that large ?

 

The delay is "only" 28 samples (according to Logic's record delay setting), but for me that's 28 samples too much.

[Threlly]

Not sure what sample rate you are using, what time period does 28 samples equate to on your system ?

[Jope]

Not sure what sample rate you are using, what time period does 28 samples equate to on your system ?

 

Although I don't really understand what ski is trying to achieve I know he usually knows what he does. I assume it's something about phasing and phase cancellation effects that were not intended... 28 samples is a lot for phasing effects... It's in the lower kHz range.

[David Robinson]

hi ski, there's a related topic that you started on the apple forum in the past coupla hours.

please read what i, and others have said.

as you have found, and myself as well, these "little delays" add up.

i can repost this same stuff here if users would like to read it.

thanks to all, david r.

[fader8]

Thoughts? Comments?

 

Only that I've experienced this also. Usually with a compressor's settings. I've gotten into the habit of always saving a tweak as a preset, or updating one of my existing ones. I don't know if this has solved it or not as it is intermittent behaviour.

 

Actually, I'm surprised that you achieved complete silence as I assumed convolution reverbs responded a bit differently every time you played back.

[Jope]

Actually, I'm surprised that you achieved complete silence as I assumed convolution reverbs responded a bit differently every time you played back.

 

I thought If any kind of reverb were 100% predictable, it would be the convolution reverb as it scans along a fixed impulse response for every sample... Which can be a disadvantage of convolution reverbs. No random variation.

[Steven_C]

Hi,

 

I think the default preset in SD uses a randomly generated response which is not saved to disk, so everytime it is regenerated (which must be everytime you reopen the session) you get a different IR which sounds very similar but won't cancel out completely. If I am right, this effect shouldn't occur with IRs loaded from disk.

[David Robinson]

Steven, i've observed this also.

best, david r.

[ski]

Hi Guys, thanks for the replies.

 

What do I achieve by setting my recording delay (and testing that it's the right value)? All manner of advantages but it would require a bit of explanation which I don't have time to get into ATM.

 

I only check the recording delay value occasionally because recently I discovered that my system sometimes wants a recording delay of -28, other times -29. That one sample difference can be significant.

 

If I have time one day I'll explain why setting the recording delay to the proper value is, I find, a crucial part of setting up Logic/audio interface.

 

BTW, 28 samples = .583 milliseconds @ 48K

 

I'm glad that I'm not the only one to experience this behavior with a plug, so thanks for confirming David/Steven/Fader8. Gonna add one more item to my workflow.

 

I too was surprised to find that (within the same session) that the reverb would cancel out doing my phase reversal stunt. But what Jope wrote makes sense and explains the predictability of it.

[ski]

I thought I'd be at it for a while solving a technical problem, but fortunately I fixed it, so here goes on the cancellation thing...

 

The purpose of properly and precisely setting the recording delay parameter is to ensure that audio you record into Logic from the outside world plays back precisely in time against any other tracks that Logic was playing back while you made that recording. Your recorded audio should not be one sample early, not one sample late... not 50 samples early or late, not 500 or 5000, but exactly where it was played, to the sample. And this can mean only one thing: zero samples early or late.

 

The idea behind this is that you need to compensate for latency introduced by your audio drivers as well as any outboard digital signal processing that you might do (in my case, routing digital audio through my board and back into Logic). If you don't set the recording delay parameter appropriately, chances are that audio you record into Logic will play back late against your existing track. (Some FW drivers actually introduce a negative delay, where recorded tracks play early).

 

How to set the recording delay parameter is another subject for another thread tho.

 

Now, aside from the (not so minor) issue of ensuring that your recorded tracks play back in time, here's another aspect of this having to do with sample accuracy and phase coherence...

 

When I deliver a cue for a film, I print my reference 2-mix plus unity gain stems of all the major instrument groups (strings, brass, percussion, etc.). The sound mixer will start with my 2-mix and hope that it will carry for the cue. But let's say that he wants to bring up the percussion a little bit. How does he do it?

 

1) he can always use some EQ on the 2-mix, but this might adversely affect the overall sound of that mix. So there are two other choices...

 

2) mute my 2-mix and instead put up all of my stems. If set to play back at unity they will re-create precisely the sound of my 2-mix with zero variation in any regard. And this can be double-checked after the printing of the stems by the phase cancellation test.

 

3) leave the 2-mix where it is and simply mix in a bit of my percussion stem. If that stem is a precise, sample-for-sample copy of the percussion within my 2-mix, and properly in phase, the net result is that the level of the percussion increases in the track. No comb filtering, no nuthin', just louder percussion.

 

There are other reasons to ensure phase coherence and recording delay parameters, but there's some food for thought for now.

 

-=sKi=-

[Bluespro123]

Wow.

No wonder all my recordings sound like @@@@.

No Phase cancellation. Thats my problem.

[Eric Wikman]

Hi Guys, thanks for the replies.

 

What do I achieve by setting my recording delay (and testing that it's the right value)? All manner of advantages but it would require a bit of explanation which I don't have time to get into ATM.

 

I only check the recording delay value occasionally because recently I discovered that my system sometimes wants a recording delay of -28, other times -29. That one sample difference can be significant.

 

If I have time one day I'll explain why setting the recording delay to the proper value is, I find, a crucial part of setting up Logic/audio interface.

 

BTW, 28 samples = .583 milliseconds @ 48K

 

I'm glad that I'm not the only one to experience this behavior with a plug, so thanks for confirming David/Steven/Fader8. Gonna add one more item to my workflow.

 

I too was surprised to find that (within the same session) that the reverb would cancel out doing my phase reversal stunt. But what Jope wrote makes sense and explains the predictability of it.

 

Hey Ski,

I am quite interseted in how to conduct your test. I'd like to find out the latency of my I/O (Meteric Halo). Can you post it, or send me a link to a topic so I can get a grip on this too?

[xs4is]

Does this thread explain things about the recording delay ski?

 

http://discussions.apple.com/thread.jspa?messageID=4144338

 

(scroll down to the long thread from iSchwartz listing the steps....)

 

X

[ski]

Hi X-Man,

 

Yes, that's the thread and the full explanation.

 

[EDITED]

[Eric Wikman]

Thanks for the link... I need to do that test this weekend.

 

The OT conversation about the Monster cables made me remember reading an article in Tape Op.. about David Gilmour's House Boat Studio in the UK... and all the insane wiring & jacks tests his engineer did to make it the best they could. Even tested wiring north vs. south!

lol

I don't think I could hear such nuances.

Thanks