The chance of zero latency

[lonbluster]

Without latency, I believe there would be no music. No depth. The optimal latency is what makes audio music to our ears IMHO.

Interesting !.. but I still prefer 128 to 256 samples. I don't have much sound degradation with relatively cheap usb hardware, if wasn't for those sporadic asio4all hiccups.

[Da hand]

Nothing is impossible... Apple figured it out.

No they didn't. They don't claim zero latency on their website. Where did you see this?

It is basically latency free

And therefore it is not zero latency

[Fanu]

Another very useful important bit to reduce the amount of added latency as much as possible is to use (only) Live's native devices.

This. Often overlooked, and also not understood by some.
I've seen ppl complain that Live is slow to react and they blame the program, while the reason is 3rd party plugins causing latency.
By using either Live's own devices or 3rd party plugins that introduce the least amount of latency you'll keep it responsive.
E.g., if you use Pro-Q3, use its zero latency mode: perfectly fine for mixing.

It sure is annoying playing beats with pads for example when you get too much latency, so best to do that before you stack a lot of latency-inducing plugins on your project.

[Ast3rix]

Nothing is impossible... Apple figured it out.

No they didn't. They don't claim zero latency on their website. Where did you see this?

It is basically latency free

And therefore it is not zero latency

True they don't directly claim that it is, but from my testing and trying to get it to perform poorly I am able to do things in realtime without the assistance of my UAD Apollo's features. I am experiencing the same performance in Studio One 5. After I reset Ableton back to factory I noticed a big performance boost and my latency is better, but not under 12ms. I still have to use the features of my UAD Apollo to get under 10ms. My plugins are loading much faster!!! I was able to resolve many issues by resetting. I will keep at it with my research, but I'm hoping that Ableton Dev's will address the issues with third party plugin usage and the core audio section.

[Mark Williams]

I'm hoping that Ableton Dev's will address the issues with third party plugin usage

Thts is down to the third party plugin devs.....

And as others have mentioned, Zero Latency is impossible, that is a simple fact.

[Ast3rix]

Again nothing is impossible that's limited thinking. I think it will require someone to see outside the problem and come up with a solution that will resolve it. It's just curious because computer hardware is ahead of software. The things that existed 20 years ago as hardware limitations no longer exist. Many of the work arounds are focused on buffer sizing and cpu which are strange because of all the advancements that has occurred over the years. It feels like we are dealing with a problem on an early 90's pc when I see some of the language.

Also in using DAW's like Logic and Studio One it does seem to be focused on the audio sub system. I realize that Apple has complete control of the plugins as well as the audio subsystem, but Studio One is working in the same realm as Ableton using vst's and their audio sub system. So to say it's impossible I don't think so. Also the plugin developers are working with vst code standard, they have to adhere to the format so that what they are making works. It's easy to blame someone else for a problem, but it's much easier to step back and look at how what you're doing interfaces with the standard and create a way that doesn't allow bad behavior to crash your platform. I haven't experienced any total software crashes in Studio One 5, but I consistently experience it regularly in Ableton.

I recently reset Ableton back to factory by deleting my template and the preference.cfg files. I noticed a major performance improvement in the loading of vst's as well as a reduction in some latency. It's still above 10ms which is noticeable and makes for a hard time when live recording. I will keep working with my setup to see if I can further reduce the amount of items I have in my template to see if I can get it down under 10ms.

[Mark Williams]

Simple fact, anything electronic will have latency of some description wether you think that or not, its a plain and already proven fact. . Plain and simple. Get used to that and learn to deal with it. Very straightfoward. If you dont like latency, play an acoustic instrument, nothing more or less.

[ecuk]

If you dont like latency, play an acoustic instrument, nothing more or less.

Of course, playing an acoustic instrument also has latency of a sort. For the sound wave to travel from a guitar to your ears takes roughly 2 to 3 milliseconds. (This is based of a speed of sound in air, at room temperature and at sea level pressure, of 343.3 m/s or, equivalently, 343.3 mm/ms. YMMV.)

[Mark Williams]

Not acoustic Drums but our friend here thinks zero latency is possbile lmao

[TLW]

Again nothing is impossible that's limited thinking. I think it will require someone to see outside the problem and come up with a solution that will resolve it.

Some things are impossible with current technology, and due to the nature of the time-space continuum of our universe are very likely to remain so indefinitely.

One such example being a time-travel machine that can know in advance what you are going to play and feed the data into a DAW before you actually play it.

Which is what is required to get a “true” zero latency. Because the speed at which the computer hardware operates is governed by the various hardware clock speeds and things can’t be calculated faster than that allows. Therefore you need to find a way to input the data to the DAW the required number of clock cycles ahead of when you play it.

It's just curious because computer hardware is ahead of software. The things that existed 20 years ago as hardware limitations no longer exist. Many of the work arounds are focused on buffer sizing and cpu which are strange because of all the advancements that has occurred over the years. It feels like we are dealing with a problem on an early 90's pc when I see some of the language.

The basic way computers operate hasn’t changed that much in decades - the hardware has got faster and more efficient and the software often better optimised, but until and unless quantum computing becomes a usable thing we’re stuck with the technological approach we have.

Also in using DAW's like Logic and Studio One it does seem to be focused on the audio sub system. I realize that Apple has complete control of the plugins as well as the audio subsystem,

Apple defines what an audio unit plugin is and how it interfaces with the DAW etc. Logic is able to use countless third-party AU plugins, AU being the Mac alternative to Steinberg’s VST system. Live can be a bit temperamental about whether a specific plugin is an AU or a VST but generally it’s not much of a problem.

but Studio One is working in the same realm as Ableton using vst's and their audio sub system. So to say it's impossible I don't think so.

I look forward to seeing the details of your time machine patent - or your patent for a completely new universe with different physical properties to the one we inhabit, whichever is easiest to implement. A guitar plugged into an analogue effect and then an analogue amplifier effectively has zero latency - though the speed of light and how fast an electric field propagates through the circuitry imposes some latency it’s so tiny as to be ignorable even if your guitar leads are miles long.

I haven't experienced any total software crashes in Studio One 5, but I consistently experience it regularly in Ableton.

I’ve not had Live 10 crash on me in ages. And I can get the same latency times out of Live as I can Logic Pro. Often a bit better.

It's still above 10ms which is noticeable and makes for a hard time when live recording. I will keep working with my setup to see if I can further reduce the amount of items I have in my template to see if I can get it down under 10ms.

Live is good at telling you how much “built in” latency a plugin has, which is something many plugin manufacturers are a bit shy about revealing (Waves being a noticeable exception). Personally I find a round trip under 12 milliseconds with nearfields about 4ms away from me to work OK.

[[jur]]

A guitar plugged into an analogue effect and then an analogue amplifier effectively has zero latency - though the speed of light and how fast an electric field propagates through the circuitry imposes some latency it’s so tiny as to be ignorable even if your guitar leads are miles long.

Well, you're still getting latency from the distance between you playing the guitar and your amp. I'm a shit at maths, but it easily means above 15-20ms a few meters away from the amp.

[subparuser]

While the comments invoking the laws of nature / physics are all correct, ditto the statements explaining that any processing will take CPU cycles (and hence time, therefore latency) - I think there is a way to reduce latency below what we consider ultra-low latency under the current paradigm.

Process audio in smaller chunks. I believe this can result in lower latency for processing within plugins / DAWs etc. Data I/O such as PCIe, USB3 and TB3 should certainly support very small buffer sizes.

Of course, your CPU will be working *very* hard to achieve this!

If any one entity were to accomplish a move in this direction and have a chance of success, it would almost certainly be Apple. Not because they have any magic technology but simply because they control the whole stack from hardware to the OS right down to having an in-house DAW.

I welcome any criticism or further info that might support or invalidate my hypothesis.

[ecuk]

Process audio in smaller chunks.

Yes and no. Although you are correct that some sources of latency are based on buffer size, there are many others which are not. And I believe that the real question in this thread (despite my rather tongue in cheek comment earlier about the amount of time it takes for a sound wave to travel from a guitar to your ears) has to do with the kind of latency which will not be affected by the audio buffer size.

For example, an FFT or a DFT (i.e., a Fourier transform), which are commonly used in the signal processing of anything which uses frequency-based analysis, does not at all depend on the audio buffer size. For any given frequency resolution you need a given number of samples, and those samples take a certain amount of time to happen. The time it takes for that many samples to happen is your latency. You can reduce the number of samples in your sample window, but then you also reduce the frequency resolution of the analysis. And in any case, you will still never get to zero latency.

Similar things can be said for any DSP that requires lookahead: compressors and limiters, for example. The latency in these is how far in time you look ahead, which has nothing at all to do with your audio buffer size.

Faster processors will not help either because this has to do with basic discrete-time digital signal processing. Each sample represents a unit of time, and you need to ‘collect’ some number of these samples, representing some period of time, before you can do the processing.

And just to head off the idea that you can use a higher sampling rate and therefore collect the desired number of samples quicker, this also won’t help in either of the cases above. For an FFT or DFT, if you double the sample rate you also need to double the number of samples in your window in order to maintain the same frequency resolution. And as for lookahead in compressors and limiters, a 5 millisecond lookahead requires 5 milliseconds of samples, regardless of your sample rate. Again, this is all fairly basic, undergraduate DSP stuff.

That said, I too, like you, welcome any criticism or fresh insight on this. Even after all these years (decades?) I am always happy to learn something new.

[subparuser]

Thanks ecuk, great reply - very informative.

I would say then that my proposal to process smaller buffers is only a partial solution. I think it might still be useful to reduce buffer-based latencies wherever possible even if there are cases where latency is intrinsic and unavoidable. After all, serial processing latencies are cumulative so I'd think it makes sense to reduce buffer-based no?

Maybe we're into seriously diminshing returns territory and I imagine the majority of folk using Live will only care about single roundtrip or just internal processing (obvs don't have data to back up this assumption) but some folk using lots of outboard can find themelves making several trips out and back again.

All that said, I don't actually expect any of this to happen and I'm not agitating for it!

Thanks again for your response, it brought much needed technical grounding to the thread.

[subparuser]

RME stuff is supposed to have very low latency although I haven't personally tried them.

I can vouch for RME low-latency performance. Rock solid 64 samples over USB2 on my (admittedly fast) modern machine but also stable @ 96 on an older (2014) laptop. Machine specs really not necessary for the purpose of this post, suffice it to say . . . RME are the masters at this (though some other vendors have caught up over Thunderbolt, I think RME are still the king of the USB hill).

My intention is to upgrade to one of their PCIe cards at great expense, the upshot of which is I will drop I/O latency by a whole millisecond or two! ^_^ Since my MIDI outs are driven by an Expert Sleepers setup hanging off the ADAT ports this will help out there (very slightly).


Very fine margins, more of a feel thing than any real barrier to working.