Good buffer size for my computer?

  • Thread starter Thread starter marc32123
  • Start date Start date
M

marc32123

New member
Right now I am watching a vid about buffer size and how it relates to latency. I am learning that for recording, you want a really low buffer size, around 128. I also know that you can't go with to low of a buffer size or it will give you popping and clicking noises on playback. I heard that you should record with a low buffer size and mix/edit with a higher buffer size, around 1024.


I am really trying to get the lowest latency I can right now though. I have been running at 10 milliseconds. I want to get around 2 or 3. I know that being able to run with a low buffer size to get low latency is only possible though if you have a good computer. I don't know much about how computer specs relate to this though, so I am wondering if someone can review my computer specs and tell me what they think would be a good buffer size for me? And also, if I am lacking in anything that could be preventing me from running at a lower latency...


Computer Specs- I bought my computer about 2 years ago already preassembled and everything new from dell. It is version 6.1.7601 Service Pack One Build 7601. The system model is XPS 8500. The system type is x64-based pc. My processor is Intel (R) Core(TM) i5-3450 CPU @ 3.10 GHz, 3101 Mhz, 4 core(s), 4 logical pro. The installed physical memory(RAM) is 8.00 GB. My available physical memory is 5.64 GB. My total virtual memory is 15.9 GB and my available virtual memory is 13.2 GB. I am running on Microsoft Windows 7 Home Premium.
 
Hi,
Not really a direct answer to your question but why do you want to get below 10ms? 10 shouldn't be noticeable.
 
I think that the answer is "the smallest buffer size that doesn't produce popping, clicking, or dropouts". Set your buffer to 256 samples. Does it produce noticeable latency? If so, turn the buffer size down. Does it pop and click when your projects grow large? Can you add audio tracks, VST effects and instruments without dropouts or crackles? If not, then turn your buffer size up. If latency is too much to bear, investigate the direct monitoring capabilities of your interface.

You're just gonna have to experiment a little. Every system is different, based not only on the specs that you provided, but also upon background processes, scheduled tasks, project complexity, and your own expectations of latency. 3ms in and out is a tall order for a USB interface. Not so much for a PCI interface.

Make sure that your interface's drivers are up-to-date, and uninstall or disable any non-essential programs/processes, and check out some Windows 7 optimization guides. Scan for malware. Make sure that you're using your interface's ASIO drivers, and not the default Windows WDM or WSAPI sound drivers. Try changing your sample rate and bit depth settings...shoot for 44.1 kHz and a 16-bit depth, just as a baseline. If your interface excels at that, try bumping it up. If it falters at those baseline settngs, then you've got issues.

A lot of it is dependent upon your interface's drivers. Some manufacturers write excellent drivers, and some don't. Some manufacturers devote their research, development, and money into developing Mac drivers, and Windows drivers are an afterthought...some invest more heavily in PC drivers and give the finger to Mac users. If you're not getting satisfactory results out of your interface, then it may be time to find a better interface. Or a more powerful computer (although your computer's specs are pretty solid).

I'm using Win7 x64 on an Intel i7 machine, with an RME interface (USB) and I'm getting 3ms in, 3.4ms out with a 256 sample ASIO buffer, recording at 96 kHz and a 24-bit depth. That's pretty speedy. RME write freaking excellent Windows drivers. With my previous M-Audio PCI card, I had to record at 44.1 kHz, 16-bit depth and a 256 sample buffer to get similar latency. And 44.1 kHz doesn't sound bad by any means. I'm only doing 96 kHz now because I can...no real audible difference.
 
Last edited:
I think my record is 7ms and playback is like huge maybe 35.
 
Tadpui (and others), I have a Scarlett 2i2. This is basically where I'm at...

I forgot to mention I have an audio interface, a Focusrite Scarlett 2i2. Basically this is where I'm at...

In my opinion, I think the first thing anyone should be taught that wants to set a home studio up is latency. If you have latency when you are recording, you are never going to get very far when trying to make professional songs. As I said before, at 10-20 milliseconds, our ears might not be perceiving very much latency if any at all, but that doesn't mean that it isn't there, even if in a very minuscule way. It also, in my opinion, can effect the feel and rythm of a song, even if its just a little teensy weensy bit off. Two things I am dead set on learning right now are how to get better vocals in my songs, and how to get the lowest latency possible. I am still confused as to one thing though. I don't understand how on my audio interface there is a button for direct monitoring. In the manual, it said with this button on, there will be zero latency. I heard that it is actually not really zero though. I am wondering, with this button on, what do you think it really puts the latency at? -1, -2, -3? Also, with this button on, does it effect only my midi? Or also vocals, guitar etc.???

Oh and one other thing. If I have this button on, do I even have to worry about what I have for sound device settings in my DAW's preferences? Like being ASIO or Wave RT or whatever...Or does having the direct monitoring switch on on my audio interface override the need to mess with the buffer size and sample rate etc. in Mixcraft?
 
The direct monitoring function simply routes a signal directly from the inputs to the outputs. It won't affect playback at all, it only affects monitoring of your inputs. So if you're recording vocals, you would hear yourself with zero (or near zero) latency directly through the interface. In this case, you'd disable monitoring of your vocal track in your DAW or else you'd hear an echo or weird phasey version of your voice.

If you're monitoring a MIDI instrument, it won't make any difference since it's only travelling through the output path of the interface.

Since you'd still be using your DAW for playback, you'd still want to use ASIO drivers, since they'll be the lowest latency of all of the drivers available. Head to Focusrite's website and grab their most recent drivers, just to make sure that you're all up to date.

The reason that they say that direct monitoring isn't exactly zero-latency is because the signal still has to go through ADC on the way in, and DAC on the way out. That takes a minuscule amount of time to do those two conversions...short enough that we don't notice it. But technically, it does introduce a tiny bit of latency. Not enough to worry about though.
 
The direct monitoring function simply routes a signal directly from the inputs to the outputs. It won't affect playback at all, it only affects monitoring of your inputs. So if you're recording vocals, you would hear yourself with zero (or near zero) latency directly through the interface. In this case, you'd disable monitoring of your vocal track in your DAW or else you'd hear an echo or weird phasey version of your voice.

Two questions from what you said I have... (Yoda is my father by the way):wtf:

You said that "the direct monitoring function simply routes a signal directly from the inputs to the outputs. It won't affect playback at all, it only affects monitoring of your inputs."

So I am confused about this. Basically you are saying that if I have the direct monitoring function on on my audio interface, that this will make it so I hear (or monitor) myself directly without any latency, because the signal is taking a shortcut. Is this correct?

And second of all, you said it "won't affect playback at all". Does this mean that even though I can hear myself recording in real time (basically) with direct monitoring on, that actually what is being recorded by my DAW isn't the real time that I am hearing?
 
Two questions from what you said I have... (Yoda is my father by the way):wtf:

You said that "the direct monitoring function simply routes a signal directly from the inputs to the outputs. It won't affect playback at all, it only affects monitoring of your inputs."

So I am confused about this. Basically you are saying that if I have the direct monitoring function on on my audio interface, that this will make it so I hear (or monitor) myself directly without any latency, because the signal is taking a shortcut. Is this correct?

And second of all, you said it "won't affect playback at all". Does this mean that even though I can hear myself recording in real time (basically) with direct monitoring on, that actually what is being recorded by my DAW isn't the real time that I am hearing?

On point 1, yes that's it. The signal takes a shortcut and therefore can be monitored with almost no latency.

On point 2, DAW software has some latency compensation built-in. It asks your interface what latency it's reporting, and the DAW knows to offset its recordings accordingly. It's confusing and I'm ignorant on most of the specifics, but there is some hocus-pocus going on behind the scenes that makes an attempt to keep things lined up in the timeline.
 
So I am confused about this. Basically you are saying that if I have the direct monitoring function on on my audio interface, that this will make it so I hear (or monitor) myself directly without any latency, because the signal is taking a shortcut. Is this correct?

Yes, but keep in mind that means you won't hear any ITB effects like reverb or delay.
 
I don't even know if 2 or 3 ms is possible, could be very wrong though. At any rate, I wouldn't think you'd notice more than 8 or 9ms latency. not to the extent that it'd affect playing back to it.
 
Yes, but keep in mind that means you won't hear any ITB effects like reverb or delay.

If you set the channel effect send to pre-fader, activate input monitoring on the DAW and mute the channel or pull the fader down you may be able to get those effects added to the direct monitored inputs. There may be a little extra "pre-delay" caused by the latency, but that's probably okay with reverb.
 
I don't even know if 2 or 3 ms is possible, could be very wrong though. At any rate, I wouldn't think you'd notice more than 8 or 9ms latency. not to the extent that it'd affect playing back to it.

I want -30ms latency...that way I can hear the music before I even begin playing, and if I fuck up something, I can quick play it over before it's even been recorded.

Make sense...?


:)
 
Back
Top