Latency question

[famous beagle]

This may be a dumb question, but oh well. I'm still learning about CPU recording.

Does a 512 sample latency setting sound the same, delay-wise, on every interface?

In other words, if I have a PCI-based interface (M-audio Delta 1010LT) set at 512 and then replace that with a USB interface (Tascam US-122) set at 512 on the same exact computer system, will the delay be the same in both instances?

Thanks

 

[Mo Facta]

Not necessarily. I have noticed quite a variance in latency between differing interfaces set to the same buffer. Some are lower, some are higher. It all depends on the driver implementation and how it interacts with the hardware.

Cheers

 

[DM60]

To add to that, it depends on the hardware as well. Should be close, but I found (I was on USB, Presonus, went to PCI M-Audio 192) PCI interfaces can handle the lower settings better than the Presonus I had. If the hardware is struggling with the lower settings, you will get more noise. But usually you will not hear it if rendered.

If you are experiencing this, try non-recording at a higher rate, lower for recording monitoring. Also, if your DAW has the feature, freeze the tracks that are completed (Plugins are pretty much set), you can still mix. This takes load off of the CPU as it creates a processed audio track in the background.

 

[PDP]

I've had several interfaces, my experience was: USB the highest, then Firewire, and PCI has the lowest latency.

 

[famous beagle]

Ok thanks for the replies y'all. That's what I suspected, because I used to have an M-Audio Delta, which was PCI. And I'm almost positive I had the buffer set to 512 on that too (as I do now on my Tascam-US 122), but I don't remember the delay in the phones (while recording vocals, for instance) being this noticeable.

 

[DM60]

Ok thanks for the replies y'all. That's what I suspected, because I used to have an M-Audio Delta, which was PCI. And I'm almost positive I had the buffer set to 512 on that too (as I do now on my Tascam-US 122), but I don't remember the delay in the phones (while recording vocals, for instance) being this noticeable.

Try taking it down to 256 see if it can handle it. You might have to toggle back and forth depending on the load. Even with PCI I sometimes jump up to 1024 when I am just mixing because of CPU load on lower buffer settings.

 

[famous beagle]

Try taking it down to 256 see if it can handle it. You might have to toggle back and forth depending on the load. Even with PCI I sometimes jump up to 1024 when I am just mixing because of CPU load on lower buffer settings.

That's what's weird. I tried that, and it's almost stable. It glitched about 3 or 4 times during a song, whether playing back or recording.

And this was with only two audio tracks and one verb track. The CPU wasn't taxed at all; it was running around 5% to 7%.

This is what I don't understand. I know my CPU is old by today's standards (Dual Core Pent 4, 2.6GHz, 3Gb RAM), but compared to the "recommend" system for the Tascam US-122 (Pentium 2 300Mhz, 128Mb RAM), it's a dream machine.

What gives?

 

[DM60]

That's what's weird. I tried that, and it's almost stable. It glitched about 3 or 4 times during a song, whether playing back or recording.

And this was with only two audio tracks and one verb track. The CPU wasn't taxed at all; it was running around 5% to 7%.

This is what I don't understand. I know my CPU is old by today's standards (Dual Core Pent 4, 2.6GHz, 3Gb RAM), but compared to the "recommend" system for the Tascam US-122 (Pentium 2 300Mhz, 128Mb RAM), it's a dream machine.

What gives?

Check to make sure you have the latest drivers. If it was the one delivered with the unit, it is almost 100% sure they are outdated. I usually go to the website and check the version. If they are newer, there is always a reason

 

[famous beagle]

Check to make sure you have the latest drivers. If it was the one delivered with the unit, it is almost 100% sure they are outdated. I usually go to the website and check the version. If they are newer, there is always a reason

Yeah, thanks. I did that.

I can work around it when recording vocals by using the direct monitoring on the US-122 and can even monitor some verb without recording it. But when recording guitar direct with AmpliTube or something, which I use quite a bit for my work, there's just no way around the 512 buffer -- unless I just want to monitor my completely dry guitar signal, which I don't want to do.

The latency at 512 samples is certainly more tolerable on guitar than it is on vocals (or any miked source), so it's workable. But it still feels "soft" when playing, and it would be great to be able to get a better feel.

 

[DM60]

Yeah, thanks. I did that.

I can work around it when recording vocals by using the direct monitoring on the US-122 and can even monitor some verb without recording it. But when recording guitar direct with AmpliTube or something, which I use quite a bit for my work, there's just no way around the 512 buffer -- unless I just want to monitor my completely dry guitar signal, which I don't want to do.

The latency at 512 samples is certainly more tolerable on guitar than it is on vocals (or any miked source), so it's workable. But it still feels "soft" when playing, and it would be great to be able to get a better feel.

I agree, the amp sound has a direct influence on how one plays. Take it down to 256, it could make noise (Is shouldn't but it might, and you latency should be nearly 0. Freeze as many of the tracks as you can, also, make sure you are using the ASIO driver in your DAW (I hope that I didn't insult there) because really you shouldn't be having this issue. But if you didn't specifically select the ASIO input in the DAW, then you are most likely not not using the ASIO driver.

 

[bouldersoundguy]

The latency caused by a 512 sample buffer is about the same as being 12 feet from your amp. That doesn't account for whatever extra latency there might be from other parts of the signal path through the computer, which would add more delay. That's definitely a noticeable amount of time, plenty to start smearing your time reference.

 

[famous beagle]

I agree, the amp sound has a direct influence on how one plays. Take it down to 256, it could make noise (Is shouldn't but it might, and you latency should be nearly 0. Freeze as many of the tracks as you can, also, make sure you are using the ASIO driver in your DAW (I hope that I didn't insult there) because really you shouldn't be having this issue. But if you didn't specifically select the ASIO input in the DAW, then you are most likely not not using the ASIO driver.

No insult at all. I've been recording for a long time, but my time in the CPU-recording world is only a fraction of that. I am definitely using the ASIO driver. I'm using the one that came with the Tascam. However, I tried using ASIO4ALL as well, just to see if it made a difference, but it didn't.

I've tried recording with just one track. I mean .... starting from a new project and just opening one track with AmpliTube on it. So there's nothing to freeze or render. It doesn't glitch often---maybe once or twice every 30 seconds or so--- but it's just enough to make it unreliable.

I'm at a loss.

 

[famous beagle]

The latency caused by a 512 sample buffer is about the same as being 12 feet from your amp. That doesn't account for whatever extra latency there might be from other parts of the signal path through the computer, which would add more delay. That's definitely a noticeable amount of time, plenty to start smearing your time reference.

Yeah it's definitely noticeable.

 

[DM60]

No insult at all. I've been recording for a long time, but my time in the CPU-recording world is only a fraction of that. I am definitely using the ASIO driver. I'm using the one that came with the Tascam. However, I tried using ASIO4ALL as well, just to see if it made a difference, but it didn't.

I've tried recording with just one track. I mean .... starting from a new project and just opening one track with AmpliTube on it. So there's nothing to freeze or render. It doesn't glitch often---maybe once or twice every 30 seconds or so--- but it's just enough to make it unreliable.

I'm at a loss.

Even though it glitches, it might not capture the glitch. So, you could try recording on 256, ignoring as much as you can the glitch. Then turn the buffers back up to 1024 and see if the glitch is there. It probably isn't. I know that is a good answer, but it at leasts gives you and idea what you're hearing and what you are recording.

Also, depending on your DAW, there should be a direct monitoring setting for the track. This (I think) gives that channel priority with the resources.

 

[bouldersoundguy]

Yeah it's definitely noticeable.

What I'm getting at is that maybe a different approach would improve the process. Although the band I'm working with is moving back to real amps and drums we have been through a few years of playing with hardware and software sims and drums built in Acid. We found that the best way to use the software sims was after the fact, and to have something like a SansAmp GT2 to play through live. I would run the guitar through a DI to capture the dry signal, and loop out of the to the GT2 to get some amp like tone live with no latency. I would record both and have the option of keeping the GT2 tone as is, replace it with a software sim or mix them together. We did this because we had to record all the guitar parts on a very tight schedule and we didn't want to spend too much time fiddling around with amp settings or be stuck with a sound we didn't like.

Honestly, the idea of using a computer to get your guitar tone live seems completely whacked to me.

 

[famous beagle]

Even though it glitches, it might not capture the glitch. So, you could try recording on 256, ignoring as much as you can the glitch. Then turn the buffers back up to 1024 and see if the glitch is there. It probably isn't. I know that is a good answer, but it at leasts gives you and idea what you're hearing and what you are recording.

Also, depending on your DAW, there should be a direct monitoring setting for the track. This (I think) gives that channel priority with the resources.

I'm using Reaper. I wasn't aware of the direct monitoring setting for the track inside the DAW --- only on the interface itself (or obviously you could just do it with an external pre or mixer before hitting the interface). I'll have to check that out.

Thanks

 

[famous beagle]

What I'm getting at is that maybe a different approach would improve the process. Although the band I'm working with is moving back to real amps and drums we have been through a few years of playing with hardware and software sims and drums built in Acid. We found that the best way to use the software sims was after the fact, and to have something like a SansAmp GT2 to play through live. I would run the guitar through a DI to capture the dry signal, and loop out of the to the GT2 to get some amp like tone live with no latency. I would record both and have the option of keeping the GT2 tone as is, replace it with a software sim or mix them together. We did this because we had to record all the guitar parts on a very tight schedule and we didn't want to spend too much time fiddling around with amp settings or be stuck with a sound we didn't like.

Honestly, the idea of using a computer to get your guitar tone live seems completely whacked to me.

Oh I see. I didn't catch your drift at first.

Yes I've thought about that idea too. I have a Roland Micro Cube that sounds perfectly decent as a practice amp, and I thought about maybe just splitting my signal from the guitar and running into the Roland as well. Then I could go line out from the Roland and record it as well, only monitoring it but also recording a dry signal which could later have AmpliTube added.

Is that what you're getting at?

I agree that I wouldn't ever want to use sims for a live setup. Everything I'm discussing has to do with recording and at a time where cranking up an amp is not an option (either sleeping kids or really loud kids that would bleed into a mic).

 

[bouldersoundguy]

Oh I see. I didn't catch your drift at first.

Yes I've thought about that idea too. I have a Roland Micro Cube that sounds perfectly decent as a practice amp, and I thought about maybe just splitting my signal from the guitar and running into the Roland as well. Then I could go line out from the Roland and record it as well, only monitoring it but also recording a dry signal which could later have AmpliTube added.

Is that what you're getting at?

If using the Roland's Rec Out/Phones silences the speaker and the tone is good enough as a reference, then yes.

I agree that I wouldn't ever want to use sims for a live setup. Everything I'm discussing has to do with recording and at a time where cranking up an amp is not an option (either sleeping kids or really loud kids that would bleed into a mic).

I don't mean on stage in front of an audience, I mean getting your guitar tone from a computer in real time, while you're playing. There's just too much of a disconnect. If a real amp isn't an option then I'd at least want to use a dedicated piece of hardware.

 

[famous beagle]

If using the Roland's Rec Out/Phones silences the speaker and the tone is good enough as a reference, then yes.



I don't mean on stage in front of an audience, I mean getting your guitar tone from a computer in real time, while you're playing. There's just too much of a disconnect. If a real amp isn't an option then I'd at least want to use a dedicated piece of hardware.

Oh I see. Man, I'm really bad at catching your drift!

Yeah I think I started using AmpliTube because I reviewed their iStomp pedal thingy for Guitar Edge magazine back in 2006 or 2007, and so I got AmpliTube 2 and AmpliTube Fender for free. That was the first time I'd ever used a plug for guitar, and so there was definitely a novelty factor there. But mainly I was surprised at how good they could sound.

Then a year or so later, I had to do a bunch of recordings for work. At the time, our boy was only 6 months old, so he slept a lot, and I couldn't crank amps. I didn't have a dedicated hardware DI solution and couldn't afford one either, so I used AmpliTube. It worked well, and I got kind of used to working with it.

I always use real amps (and usually record analog) when I record my own stuff (kind of indie/alternative/etc.), but for work (I write and edit guitar instructional books for Hal Leonard and record guitars for those books and others too), I need something quick, easy, and versatile that sounds good and that can be done at any time of the day (with sleeping or screaming kids). While it's true that a piece of hardware would do that as well, I'd just gotten used to using AmpliTube, and the latency wasn't a noticeable issue until I got this new interface.

But now that you gave me the idea about splitting the signal and monitoring the Roland while recording the DI signal, I think I'll try that. The Roland is more than satisfactory as a reference tone. In fact, I even used it to record some parts a book one time in a pinch when my normal rig was on the fritz.