What buffer setting do you use most?

[lwilliam]

I can't seem to get any lower than 256 with my 002R without getting crackles and pops. It's rough laying down "in the pocket" grooves with that and vocalists sure can't track to that (I monitor vocals via a mixer).

Are there other interfaces where you can actually use 64 or even 128 buffer samples for tracking?

[David Polich]

512 buffer setting 90% of the time.

Here's where I strongly recommend getting a small good sounding hardware mixer. Yes, you can use CueMix's zero latency monitoring, but not everyone has a MOTU audio interface.

With a hardware mixer, you can have vocalists and musicians monitor themselves thorugh an aux or headphone send. Bus audio from the mixer to an aux send and feed that send to your Mac's audio interface. Take the main outs of your audio interface and feed them into two channels of the mixer, but don't bus those channels anywhere. Result - latency free monitoring.

I also have most of my virtual synths and samplers on a separate PC, which further reduces any latency problems I'd otherwise have. And a second machine will extend the life of your current Mac because you're not putting any extra strain on it by running VI's inside DP.

With my setup (hardware mixer and separate PC for VI's) I've even tracked vocals and instruments at 1024 buffer setting without any trouble.

[Shooshie]

I can't seem to get any lower than 256 with my 002R without getting crackles and pops. It's rough laying down "in the pocket" grooves with that and vocalists sure can't track to that (I monitor vocals via a mixer).

Are there other interfaces where you can actually use 64 or even 128 buffer samples for tracking?

If you use MOTU interfaces, you can monitor with zero latency via CueMix. Outside of that, I can't think of too many reason why you'd need to record with such small buffers. Have you thought of getting a MOTU interface? If you need reverb or effects for the artist, get a cheap Yamaha effects unit or equivalent for your monitoring system.

That's what completes DP as a system, IMHO.

Shooshie

[daniel.sneed]

But real problems appear when trying to play VI on the same computer.

I do need to use buffer 128 to get a playable time response.
On very slow musical pieces with no rythm concern, I sometimes use 256 and even sometimes 512.

But the best way to go is 128 IMHO.
To achieve a correct flow on "big" projects, I disable some tracks and some plugins.

When tracking is done, I always switch back to 1024 and stay there.

[Shooshie]

Many VIs have an economy mode in which you can strip the sound to its barest essentials. If rhythm is the key, then chances are you don't need all 12 layers of that Ivory Bosendorfer. 4 or 6 would do. Then you can get the buffer on down to 128 sometimes. You can go back and get the sound later when you've got a healthier buffer for playback, like 512 or 1024.

Another option is to record with a sound created in one of the MOTU VIs. If you can create one that matches the response of the sound you ultimately plan to use, you can let your imagination do the rest, along with possibly an outboard effects unit, and get it tracked pretty much the way you'd play with very little velocity editing required.

Yet another option is to record with an outboard rack unit, or a Kurzweil, Roland, or other standard keyboard. The idea that it's got to sound like the final product while you play is not one that's going to get you far at the current state of the art. Maybe someday we can all have full orchestras, choirs, rock, jazz and chamber groups that come to life at the flip of a switch, but we're really not there yet. Our final output gives the illusion that we are, but in reality we have to make CPU sacrifices along the way and piece together the final product.

Shooshie

[HCMarkus]

Using a MOTU 828mkII, I run my buffer at 128 without troubles. Last night I was running Ivory, BFD, Ethno and East-West Symphonic and two or more Altiverbs and a few misc other plugs simultaneously on my Quad at 128 buffer without any problem. Performance meters were in the 60%-70% range. I do switch to 1024 when it's time to mix so I can load up on the plug ins, but agree that latency is a problem when tracking VI's unless the buffer is set to 128. On the other hand, you can lay down audio and MIDI tracks in sync regardles of buffer size. If you simply must track a new VI after your system is bogged down with tracks and plugins, use an external MIDI module with a similar sound to monitor your performances, then switch to playback from your VI.

lwilliam, you don't say what size project you are working with in your post, but unless it is really big, I would expect you should be able to work at 128 given your machine configuration. Have you tried the RME interface you list?

[toodamnhip]

Many VIs have an economy mode in which you can strip the sound to its barest essentials. If rhythm is the key, then chances are you don't need all 12 layers of that Ivory Bosendorfer. 4 or 6 would do. Then you can get the buffer on down to 128 sometimes. You can go back and get the sound later when you've got a healthier buffer for playback, like 512 or 1024.

Another option is to record with a sound created in one of the MOTU VIs. If you can create one that matches the response of the sound you ultimately plan to use, you can let your imagination do the rest, along with possibly an outboard effects unit, and get it tracked pretty much the way you'd play with very little velocity editing required.

Yet another option is to record with an outboard rack unit, or a Kurzweil, Roland, or other standard keyboard. The idea that it's got to sound like the final product while you play is not one that's going to get you far at the current state of the art. Maybe someday we can all have full orchestras, choirs, rock, jazz and chamber groups that come to life at the flip of a switch, but we're really not there yet. Our final output gives the illusion that we are, but in reality we have to make CPU sacrifices along the way and piece together the final product.

Shooshie

I understand all of this...But damn it..I really hate piecing together projects when needing to set buffers to 128. It suck..period..but of course, I'd have died yrs ago for what I have now as far as quality and sounds and power...

But the day 128 can be achieved on a full mix, will be a grand day indeed...
Otherwise, buffers 1024.
There is NO reason why not to run 1024..it is dumb toi tax ones computer for NO reason. If you don;t have latency issues, keep your comp at 1024..period.

[Timeline]

With all the instances of VI I use I end up with three x 1024 on average to keep the track sounding great. For low latency overdubs I just import a mix - the overdub item and start a new session. This gets me to 64-128 on the latency setup which is a hell of lot closer to reality and where we should be to do our work.

I think when eight cores are available we will begin to be able to get reasonable latency as plugs used towards the end of a songs production will vanish as non issue anchors on the cpu. Until then, it's buffer shuffle dude!

oh one other tip.

Turn input switching monitor to off & monitor through the IO. This works ok without buffer changes but the previous recorded voice has to be set in the main mixer very low in playback. Treat the delay coming back in input or record as a slap echo. At least it gives the player or singer a chance to figure out where they're at. I leave it as a slap echo when he/she is doing the main vocal. Have not had to many complaints so far. It's defiantly a cheat but it's quick.

Would be nice to se a "PLAY ONLY" in the audio patch thru selections.

[lwilliam]

I'm using the RME, but it's connected to the ADAT i/o on the 002R. There's no direct input via F/W or USB for it that I know of.

I can have a small project with only a couple of VIs (Trilogy, B4) and get pops when I switch to 128.

VI's and live guitar are the biggest areas to work around with the higher buffer settings. I'm primarily a guitar player and secondarily a keyboardist.

I get around the vocalist issue by using an outboard mixer for monitoring. That's not really an issue. However, I have also had to do that with my guitar, and it's not the same as playing it through GtrRig2 at all.

I can work at 128. I can use 128 when I'm running PTLE until the session gets loaded, then I have to switch to 256. But when running DP, I have to use Digi's CAM, which isn't the best. It at least WORKS with DP now, whereas before, DP just choked and crashed with Digi's CAM. It still has to sync from the 002R clock, it won't clock to the RME.

Maybe when I upgrade from 4.61 to 5.11 things will improve; maybe not. I can use it as is, but would sure like to go to at least 128.

There have been a few posts on the Logic boards about Symphony running at 32 and 64 samples, so I'm wondering if anyone can really use 128 or lower with "real world" cards and what they do to get around it if they can't.

Great responses so far...looks like I'm not really experiencing too much that anyone else isn't.

Importing a stereo bounce for overdubs sounds like it might have potential...

[Timeline]

RME definitely has no latency monitoring. I use it daily.
it's in the input setup of total mix faders.

[lwilliam]

RME definitely has no latency monitoring. I use it daily.
it's in the input setup of total mix faders.

Don't you have the fireface, though? That's a different animal from the ADI-8 Pro, which is just an a/d/a converter.

I don't think that no latency (direct) monitoring will help with recording guitar tracks if there are effects you want/need to hear while recording. Only something like one of the Line 6 boxes with built-in guitar effects would be able to record the DI guitar while you listen with effects - and minimal to no latency: kind of like a Line6 POD with a DI output. I believe that's the POD XT for $600 or so...not gonna happen any time soon.

....OR...if there's an interface that works well at less than 128 samples for the buffer.

[Timeline]

Ahh.. Sorry yes. Thought you were talkin 400 or 800.

Wonder if the 400-800 software would eventually be adapted. Talk to George about it at support Synthax. Maybe he could give some ideas.

I use a little Behringer as well for headphone monitors. It's great with digital effects and very clear. For vocals and overdubs I patch the track into two channels of mixer and Y the vocal limiter OP for direct voice. The result for the player is great no latency overdubs with effects. This little mixer is the bomb and the best product they ever made with pretty good digi effects and under $100 bucks.

Best
G

[Frodo]

I think when eight cores are available we will begin to be able to get reasonable latency as plugs used towards the end of a songs production will vanish as non issue anchors on the cpu. Until then, it's buffer shuffle dude!

Indeedy weedy.

I'm still mostly tracking VIs ar 512, occasionally 256, and mixing at 1024 out the yin-yang like many others, but I will try your tips on my next project-- thanks, T.

But your statement quoted above raises a question:

Do you think 8 cores hovering around the 3G brick wall would be better/worse/same than-as 4 cores running at 6Ghz each as far as buffer latency behavior is concerned?

[Timeline]

Not sure yet but you raise a good question.

I think we need a user utility to manage some of what cores process what audio tracks and VI's and plugs if we wish.

Currently there is mass confusion letting the computer select data states and paths to these numerous CPU's and in the AMD's case it does not work at all with tons of errors and miss handling cpu loads from differend applications. We need hard paths with software able to parse the data directions I think.

I asked Apple 7 years ago for the ability to select processor loads for general applications with multiple processors. Then they were llistening to us and had no multi core machines. I asked for a quad machine.

On the data thing my thought was by being able to see the spikes we could eliminate the dynamic nature of miss handling the load and take out some processing sluggishness as the system directs the flow of data. After all, we know what and when our projects are going to get busy.

For instance in this scenario the first 44-plugs go directly through core 2 and second 44 core 3 etc and track throughputs operate similarly up to eight or as we require with NO data path switching.

This would likely be a little bit less efficient but allot more reliable and likely faster to commit path workload. Then again... maybe not.

Just my .02c

[monkey man]

Hey, T

I see your "Monthly Peyote treatments" are back in your sig.
They're working; you're morphing into a hobbit-like, propellorheaded geek.
Gimme sum.