|
Post by Blackdawg on Jan 3, 2021 18:39:54 GMT -6
Something else to keep in mind here..
64samples at 88.2k is the same as 32samples at 44.1k.
Since you are doubling the sample rate the latency samples double so (assuming the clocking 100% the same which it probably isn't 100% but super close) the amount of actual time that passes in 32 samples at 44.1k is the same as 64 samples at 88.2k
What you're asking to do would be the same as 16 samples at 44.1k aka 32 samples at 88.2k. And you're interface just isn't going to do that.
If you want better latency then you need a different DSP work flow. So UAD synths or HDX DSP stuff.
otherwise I'm pretty sure you'll always be dealing with around 32 sample latency unless it's DSP driven.
|
|
|
Post by Guitar on Jan 3, 2021 18:55:07 GMT -6
HDX doesn't help with virtual instruments, I'm told. But I just skimmed a Carbon review on ProToolsExpert.
HDX CAN do audio at zero latency, I am told.
UAD also has "some latency" when you do the measurements. It might be inaudible, an ms or two. It's also dependent completely on UAD plugins, which I don't rate for instruments, so you're sort of up the creek with a tiny paddle.
The fast, native thing has been my personal solution kind of like LesC is going for with the RME's. I like it better than Apollo.
|
|
|
Post by Blackdawg on Jan 3, 2021 19:06:26 GMT -6
HDX doesn't help with virtual instruments, I'm told. But I just skimmed a Carbon review on ProToolsExpert. HDX CAN do audio at zero latency, I am told. UAD also has "some latency" when you do the measurements. It might be inaudible, an ms or two. It's also dependent completely on UAD plugins, which I don't rate for instruments, so you're sort of up the creek with a tiny paddle. The fast, native thing has been my personal solution kind of like LesC is going for with the RME's. I like it better than Apollo. Of course. But surely ( I don't know this because I don't use VI's really) some VI's out there are DSP by either HDX or UAD..or something. Or I suppose using an analog system and just running MIDI out to it might work..not sure though. Again, I don't do VI's very much. Pretty much just acoustic stuff and mixing/mastering. So not super familiar with those things anymore, been years.
|
|
|
Post by indiehouse on Jan 3, 2021 19:08:07 GMT -6
Something else to keep in mind here.. 64samples at 88.2k is the same as 32samples at 44.1k. Since you are doubling the sample rate the latency samples double so (assuming the clocking 100% the same which it probably isn't 100% but super close) the amount of actual time that passes in 32 samples at 44.1k is the same as 64 samples at 88.2k What you're asking to do would be the same as 16 samples at 44.1k aka 32 samples at 88.2k. And you're interface just isn't going to do that. If you want better latency then you need a different DSP work flow. So UAD synths or HDX DSP stuff. otherwise I'm pretty sure you'll always be dealing with around 32 sample latency unless it's DSP driven. Ah. I was just reading about this. Thanks for that info! So what about people that say they have the option to set their buffer to 32 samples when using higher session sample rates?
|
|
|
Post by Guitar on Jan 3, 2021 19:16:06 GMT -6
Of course. But surely ( I don't know this because I don't use VI's really) some VI's out there are DSP by either HDX or UAD..or something. Or I suppose using an analog system and just running MIDI out to it might work..not sure though. Again, I don't do VI's very much. Pretty much just acoustic stuff and mixing/mastering. So not super familiar with those things anymore, been years. I'm not sure how the UAD ones work. They have a Moog, and a piano, (I think.) I'm not sure if there are any AAX virtual instrument plugins that specifically can use the HDX feature of "Sub 1ms Latency," but I don't believe there are. For acoustic performance, I'm assuming that sub 1ms round trip could be wonderful for headphone mixes, or tracking a guitar amp plugin instead of a real amp, or bass. Or monitoring a DI instrument through some hardware chain that takes a trip through the interface, like maybe drbill could do. For mixdown, mastering, I'm not really sure if any of it matters.
|
|
|
Post by tkaitkai on Jan 3, 2021 19:44:26 GMT -6
So what about people that say they have the option to set their buffer to 32 samples when using higher session sample rates? I mean, it's definitely possible to do this... Here's a screenshot of SSD5 running in Logic at 88.2/32: Very taxing on the CPU of course, but it still works. I can even record MIDI if I stick to extremely basic/lightweight VIs.
|
|
|
Post by Blackdawg on Jan 4, 2021 0:53:52 GMT -6
Something else to keep in mind here.. 64samples at 88.2k is the same as 32samples at 44.1k. Since you are doubling the sample rate the latency samples double so (assuming the clocking 100% the same which it probably isn't 100% but super close) the amount of actual time that passes in 32 samples at 44.1k is the same as 64 samples at 88.2k What you're asking to do would be the same as 16 samples at 44.1k aka 32 samples at 88.2k. And you're interface just isn't going to do that. If you want better latency then you need a different DSP work flow. So UAD synths or HDX DSP stuff. otherwise I'm pretty sure you'll always be dealing with around 32 sample latency unless it's DSP driven. Ah. I was just reading about this. Thanks for that info! So what about people that say they have the option to set their buffer to 32 samples when using higher session sample rates? It probably has more to do with the interface they are running would be my guess. And computer. I mean hell I just was working on a track in 24/96k and the lowest I could do on my HDX system was 128 samples. Why? no clue. Im not using any Vi's either. But if I had to guess it would be because Im working on a slowish portable hard drive.
|
|
|
Post by indiehouse on Jan 4, 2021 4:36:25 GMT -6
Ah. I was just reading about this. Thanks for that info! So what about people that say they have the option to set their buffer to 32 samples when using higher session sample rates? It probably has more to do with the interface they are running would be my guess. And computer. I mean hell I just was working on a track in 24/96k and the lowest I could do on my HDX system was 128 samples. Why? no clue. Im not using any Vi's either. But if I had to guess it would be because Im working on a slowish portable hard drive. Hmmm...It’s def not my computer or SSD. Maybe the Motu 828ES? I’ll continue to investigate.
|
|
|
Post by indiehouse on Jan 4, 2021 4:44:07 GMT -6
So what about people that say they have the option to set their buffer to 32 samples when using higher session sample rates? I mean, it's definitely possible to do this... Here's a screenshot of SSD5 running in Logic at 88.2/32: Very taxing on the CPU of course, but it still works. I can even record MIDI if I stick to extremely basic/lightweight VIs. Thanks for that info. That’s helpful to know a comparison.
|
|
|
Post by the other mark williams on Jan 4, 2021 8:38:28 GMT -6
I mean, it's definitely possible to do this... Here's a screenshot of SSD5 running in Logic at 88.2/32: Very taxing on the CPU of course, but it still works. I can even record MIDI if I stick to extremely basic/lightweight VIs. Thanks for that info. That’s helpful to know a comparison. I wouldn’t necessarily *expect* to get lower than 64 at 88.2 or 96. I forget: are you using Pro Tools, or something else? And what is your RTL with 64 at 88.2? Surely it’s low enough to play VIs? Oftentimes VIs are playable at 128 or even 256 at 44.1 IME...
|
|
|
Post by kcatthedog on Jan 4, 2021 9:05:45 GMT -6
If really midi latency stuck, one thing I do is bounce my 2 bus mix and add to new session with nothing else in it, no plugs ins, create midi track, same instrument etc. as first session and do performance in 2nd session, export midi file and import into first session.
|
|
|
Post by ragan on Jan 4, 2021 10:33:48 GMT -6
FWIW, I’m pretty sensitive to latency and I regularly track keys VIs (Diva, Keyscape, MTron, etc) at 256 buffer (88.2kHz) and have no issue with the feel at all.
|
|
|
Post by Guitar on Jan 4, 2021 10:40:27 GMT -6
FWIW, I’m pretty sensitive to latency and I regularly track keys VIs (Diva, Keyscape, MTron, etc) at 256 buffer (88.2kHz) and have no issue with the feel at all. Ragan do you have any new music streaming since Day And Night And The Lakeside Triumph? It took me on a trip the other day. We were talking about timing a little bit and you definitely are precise with that. I'm also interested in the lyrics there seems to be a bit of eastern wisdom in there, but I'm not sure if that's deliberate or not.
|
|
|
Post by ragan on Jan 4, 2021 10:52:21 GMT -6
FWIW, I’m pretty sensitive to latency and I regularly track keys VIs (Diva, Keyscape, MTron, etc) at 256 buffer (88.2kHz) and have no issue with the feel at all. Ragan do you have any new music streaming since Day And Night And The Lakeside Triumph? It took me on a trip the other day. We were talking about timing a little bit and you definitely are precise with that. I'm also interested in the lyrics there seems to be a bit of eastern wisdom in there, but I'm not sure if that's deliberate or not. Well thanks! I just completed my next record (after DANATLT) and as soon as I have artwork it’ll be streamable. I posted a few tracks off of it in the ‘What Are You Working On?” sub-forum recently.
|
|
|
Post by indiehouse on Jan 4, 2021 11:32:53 GMT -6
FWIW, I’m pretty sensitive to latency and I regularly track keys VIs (Diva, Keyscape, MTron, etc) at 256 buffer (88.2kHz) and have no issue with the feel at all. Thanks man!
Still working through understanding it all. I'm probably overthinking it, I just know next to nothing about midi stuff. Just trying to sort out if I have a setup problem with my rig or if it's working as intended.
|
|
|
Post by ragan on Jan 4, 2021 11:52:45 GMT -6
FWIW, I’m pretty sensitive to latency and I regularly track keys VIs (Diva, Keyscape, MTron, etc) at 256 buffer (88.2kHz) and have no issue with the feel at all. Thanks man!
Still working through understanding it all. I'm probably overthinking it, I just know next to nothing about midi stuff. Just trying to sort out if I have a setup problem with my rig or if it's working as intended.
Most of the time, for me, it's just create an instrument track, choose what MIDI input you want (just like you'd choose any other kind of input), throw the VI on an insert and arm the track. I used to be intimidated by dealing with MIDI too and now it's bread and butter workflow for me.
|
|
|
Post by drbill on Jan 4, 2021 13:20:48 GMT -6
Something else to keep in mind here.. 64samples at 88.2k is the same as 32samples at 44.1k. I was wondering why no one mentioned this. Thx Blackdawg! Yes, this is true. MIDI latency is generally a fixed time, and changing the session sample rate does not change that.
|
|
|
Post by ragan on Jan 4, 2021 13:23:17 GMT -6
Something else to keep in mind here.. 64samples at 88.2k is the same as 32samples at 44.1k. I was wondering why no one mentioned this. Thx Blackdawg! Yes, this is true. MIDI latency is generally a fixed time, and changing the session sample rate does not change that. But they were talking about DAW buffer latency, which does change (on a per sample basis) with sample rate.
|
|
|
Post by drbill on Jan 4, 2021 20:55:59 GMT -6
Yes I understood that - unless I'm misunderstanding.... All samples lengths are not created equal. Samples at 48k are twice as long as samples at 96k -- as there are 48,00 samples in a second at 48k sampling rate and 96,000 samples in a second at 96k sampling rate. The fixed constant is "the second". Therefore, a 128 sample buffer at 48k is the same fixed amount of actual time (latency) as a 256 sample buffer at 96k. There are double the samples in one second at 96k, thus necessitating a larger (although ultimately the same) buffer setting for a computer with fixed computing power. i.e. Double your sample rate, double the buffer size - lower your sample rate in half, and you can shrink the buffer size in half - but in reality - they are the same time constant. Hope that makes sense. I'm tired... I'm fairly sure this is what you meant - and I'm also fairly sure this is what Blackdawg meant.
|
|
|
Post by ragan on Jan 4, 2021 21:08:53 GMT -6
Something else to keep in mind here.. 64samples at 88.2k is the same as 32samples at 44.1k. I was wondering why no one mentioned this. Thx Blackdawg! Yes, this is true. MIDI latency is generally a fixed time, and changing the session sample rate does not change that. Yes I understood that - unless I'm misunderstanding.... All samples lengths are not created equal. Samples at 48k are twice as long as samples at 96k -- as there are 48,00 samples in a second at 48k sampling rate and 96,000 samples in a second at 96k sampling rate. The fixed constant is "the second". Therefore, a 128 sample buffer at 48k is the same fixed amount of actual time (latency) as a 256 sample buffer at 96k. There are double the samples in one second at 96k, thus necessitating a larger (although ultimately the same) buffer setting for a computer with fixed computing power. i.e. Double your sample rate, double the buffer size - lower your sample rate in half, and you can shrink the buffer size in half - but in reality - they are the same time constant. Hope that makes sense. I'm tired... I'm fairly sure this is what you meant - and I'm also fairly sure this is what Blackdawg meant. Yes, totally. I was just responding to you making a distinction between “MIDI latency” and DAW latency and saying DAW buffer won’t affect the former. I was just pointing out that it was the latter actually being discussed and that it, just like you say, is a function of both sample rate and buffer.
|
|
|
Post by drbill on Jan 4, 2021 23:15:42 GMT -6
I was wondering why no one mentioned this. Thx Blackdawg! Yes, this is true. MIDI latency is generally a fixed time, and changing the session sample rate does not change that. Yes I understood that - unless I'm misunderstanding.... All samples lengths are not created equal. Samples at 48k are twice as long as samples at 96k -- as there are 48,00 samples in a second at 48k sampling rate and 96,000 samples in a second at 96k sampling rate. The fixed constant is "the second". Therefore, a 128 sample buffer at 48k is the same fixed amount of actual time (latency) as a 256 sample buffer at 96k. There are double the samples in one second at 96k, thus necessitating a larger (although ultimately the same) buffer setting for a computer with fixed computing power. i.e. Double your sample rate, double the buffer size - lower your sample rate in half, and you can shrink the buffer size in half - but in reality - they are the same time constant. Hope that makes sense. I'm tired... I'm fairly sure this is what you meant - and I'm also fairly sure this is what Blackdawg meant. Yes, totally. I was just responding to you making a distinction between “MIDI latency” and DAW latency and saying DAW buffer won’t affect the former. I was just pointing out that it was the latter actually being discussed and that it, just like you say, is a function of both sample rate and buffer. There is a degree of midi latency added to the soup as well. Depends on your drivers, your controller, whether or not you're doing a midi interface and midi out of your controller or USB output, the drivers in your computer / daw, etc.. But IME, the DAW latency is far more critical in getting right than worrying about any midi latencies. (which tend to float around far more than DAW latency which seems (at least to me) pretty rock solid.
|
|
|
Post by indiehouse on Jan 5, 2021 8:26:15 GMT -6
I just had a Motu guy tell me that the master fader track in PT's will cause latency.
Me: "I always have a master fader track set up in my sessions. Does this cause latency? I realize that having plugins on that master fader track would cause latency, but does just the master fader track itself cause latency?"
Motu: "Sure. It's the bottleneck for the entire project. It has extra buffers to accommodate this."
Is that true? Should I not set up a master fader track until mix time?
|
|
|
Post by Guitar on Jan 5, 2021 8:50:16 GMT -6
If you want to get nitty gritty with it, set up a loopback test of some sort with this utility application: oblique-audio.com/rtl-utility.phpThen you can have some actual data, some numbers, to reference. I think the tech help person was talking about plugins on the master track, not the master track itself, but I am just guessing. I'm not sure how much latency the summing engine has in any given DAW. Obviously you have to have an output of some sort to listen to your track.
|
|
|
Post by indiehouse on Jan 5, 2021 9:07:55 GMT -6
I think the tech help person was talking about plugins on the master track, not the master track itself, but I am just guessing. Yes, but if you read my question, I clearly ask if just having a master fader WITHOUT plugins will add latency.
I've never heard of this before. I feel like a newbie asking all these fundamental questions, but I am on a mission to learn something I didn't know before.
|
|
|
Post by Guitar on Jan 5, 2021 9:47:28 GMT -6
I would start measuring if I were you. Try different configurations, with some standards, only one variable. Try with a master bus, without, put a number on it. That tech person didn't give you any numbers it could be some tiny buffer that you won't hear.
I'm learning too so I'm enjoying your journey with this.
|
|