• 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Latency Improvement?
#11
(04-14-2020, 04:40 PM)Patrice Brousseau Wrote: Translation: (yes, they talk about 2nd gen but I think the original article mentioned also Clarett and 3rd gen Scarletts). There is an uninstaller in the package if it doesn't work. Should work I think.)

How can I improve the latency of my Scarlett 2nd generation interface on Mac?

    vor 2 Jahren Updated

This article applies to: Scarlett 2nd Generation

We have developed a small additional app that enables Mac users to improve the roundtrip latency of their Scarlett 2nd generation interface to 1ms (this specification depends on the respective system). The app cannot be installed on Windows PCs.

The Scarlett (2nd) Low Latency Installer installs an additional file that reduces the latency of the Apple Core Audio driver. After running the installer as well as after deleting it is necessary to restart the system.

Please note that small buffer sizes can increase the processing power of your processor and this can lead to problems with audio performance.

If you want to delete the Low Latency Installer from your Mac, please do the following:

Mac HD -> System -> Library -> Extensions -> Scarlett2ndGen.kext

Then empty the trash and restart the system.

Please note that the Low Latency Installer does not work with the Scarlett interfaces of the first generation and that proper operation is therefore not possible under certain circumstances.

As is now, you are running with the generic USB audio stack of MacOS, thus higher latencies.

Thank you very much Patrice.  I will give this a try.
  Reply
#12
Pleasure, please report back, I'm curious to know if 3rd gen works with it.

Patrice
  Reply
#13
(04-15-2020, 10:48 PM)Patrice Brousseau Wrote: Pleasure, please report back, I'm curious to know if 3rd gen works with it.

Patrice

Thanks, haven't tried it yet.  I'm doing a few test sessions with the frame size adjustment to 1.0 - it has reduced my interface latency to 8.3ms.  I'll let you know if/when I try the installer.

- Jim
  Reply
#14
Hi,

I'm Kelana, from Jakarta, Indonesia. I'm using a Mac Pro (Late 2013 .. trashcan model) running on OS X Catalina and using Avid HD Native Thunderbolt & HD Omni audio interface. I previously had some successful connections with my friends in Jakarta but since the recent update, the app sometimes doesn't even load. It has loaded successfully right now but I'm receiving no audio from my friends. Any way I can fix this somehow?

- My audio interface latency has always been at around 12.7 ms (How can I improve this?)
- Internet speed is 93.4 Mbps download / 22.2 Mbps upload
- Upnp is enabled
- i haven't tried switching to a static IP address yet. It's still a dynamic IP address. I will try to contact my ISP tomorrow.

We seriously need help, me & my friends are trying to achieve real-time jamsessions in Indonesia Smile)
  Reply
#15
Here are a few things from a list I posted on another thread (most of these actually refer to your AUDIO INTERFACE LATENCY, but a couple will address INTERNET LATENCY - together they make up you TOTAL LATENCY).  Also, I use Mac and some of this is specific to Macs, but most is not.  A quick note that apparently some PC users are seeing an interface latency of 0.0 which I don't believe is actually possible, as even the very best pro gear still will give you around 1ms each way (depending on sample rate) and the system still needs a buffer.  Not sure what is going on there, but those numbers are fantastic either way!  Possibly JK accounts for some buffers and system latency differently for PC vs MAC?  Anyway, here goes:

1) Audio Frame Rate (main menu at top: Manage --> Audio Settings --> Audio Booster  *also found in network settings):
The higher the audio frame size, the higher the latency you will have.  Because we are dealing with round trip audio, the added latency is essentially double the audio frame size (that's my theory anyway - and you can confirm the numbers if you look at your interface latency before and after a change).  So an audio frame size of 1 = 2ms added latency to your interface.  Audio frame size of 2 = 4ms of added latency.  2.5 = 5ms of additional latency, and so on.  If you have a newer powerful computer you should be able to run at frame size of 1, but test it out.  If you hear clicks, pops, etc., you will need to change to a higher frame size.  I am using a very powerful computer but it is 12 years old and I suspect the bus speed may be limiting this, but it could have something to do with the os (I'm stuck on 10.10.5 until I buy a new computer).  So the lowest setting I can use is 2.5.  If I upgrade and can get to an audio frame size of 1 I would shave off 3ms of latency.  FWIW, everyone I know that is using a newer Macbook Pro has no issue running a frame size of 1.
* Note that you must be in a session (even a test session will work) to adjust these settings.

2) Bit rate (main menu at top: Manage --> Audio Settings --> Audio Booster  *also found in network settings):
Grateful Dead Jams claims that changing this number to 128 made a noticeable improvement on his session.  I haven't seen a measurable change when switching to lower a bit rate from the default 320 (I was the only one in my session that changed it - while the others looked at my numbers), but it makes sense that if everyone did it things could be better.  I will have everyone try it on my next session and see what happens.  There is definitely a noticeable difference in quality as you change this setting.  Low and high end will suffer digital artifacts more noticeably (especially for bass players and other instruments heavy in those frequencies).  You will also notice a change in how forward the mids are.  I actually prefer the 320 setting to the 512, all else being equal (that is for vocals and electric guitar anyway).  Ultimately, if you can't play with someone else due to sync issues, it doesn't matter how good it sounds.  So if lower quality helps you be able to play, that is what you need to do.  I can deal with lower quality audio, but not big delay and time $fts.
* Note that you must be in a session (even a test session will work) to adjust these settings.

3) You interface's sample rate:
One person I play with was able to make a considerable savings in his interface latency by changing his interface's sample rate from 44.1 to 48.  And JK allows for each player to be at a different sample rate.  Not sure if this creates other issues though.  I could see how it may increase JK's buffer needs or possibly jitter.  I think that is because it is converting your audio to the bitrate you select (see above) and that is what is being sent.  Not sure how it deals with mixed bit rates on the other end though.  I have a feeling that the more settings everyone in a session has the same, the smother everything will work.  Although, I have no proof of this as I haven't even tested it and don't know how the software is handling many things.  But ultimately, if someone can shave 10 or more ms off their latency, that may be a better return then slightly reduced jitter or something.  After all, I think the system has to cater to the worst setup in the group of players.  That means that everyone's experience will significantly improve if one person can shave that much off of their latency.  Good luck out there!
* Note that I see a help article is in the works about the need for sample rates to match, so we should soon have information about how I am actually wrong about using mixed sample rates - although it does work, it must not be ideal and causes issues.  I will do my best to update when that gets posted.  But it should still be true that the higher the sample rate, the lower your latency will be.  Curious to see what they say about the need to have everyone on the same Audio Frame Size.

4) System Buffer (not normally adjustable on Mac's Core Audio - I think ASIO on windows can be adjusted, maybe someone on windows can chime in?):
*BigStu posted this for anyone using a Focusrite interface on Mac:
"I just was searching the Focusrite website support area, and found this article:
https://support.focusrite.com/hc/en-gb/a...ces-on-Mac
In it, you will find a link to a driver you can install to lower latency. I just installed it, and my latency went from 10.4ms down to 7.8ms, which puts me in the green zone on JamKazam. Not sure if this will help enough if your latency is up around 20ms, but it's worth a try.
Also, I am using Ableton Live 10 Lite, and I have lowered my buffer size setting to 64 samples to help speed things up.
Hope this all helps."
* People are saying that a lot depends on which ASIO driver you are using.  I also think I saw that this may be adjustable in the Audio Gear (from the dropdown menu next to your profile name)

5) Interface / drivers:
Your interface and it's drivers can make a huge difference.  Bad ones can be over 30 or 40ms.  Great ones will be under 9m or even lower.  Some exceptional ones with high bandwidth cable may be under 5 with the right settings but this would be very difficult considering the rest of what is going on with the software and computer (see other items here).  Many manufactures have specs for latency on their websites or in their manuals.  Some test results can be found on forums such as Gearslutz, or elsewhere on the web.  I can tell you that RME, UA, and Slate all have very low latency interfaces but higher costs as well.  Those numbers will vary depending on the unit, cable type, if you need a conversion cable to get to your computer and how good that is, the design of that particular interface (there is variation between units from a give manufacturer... usually), and the drivers for that unit.  So make sure you have the most recent drivers for your interface! 
* Keep in mind that your "interface latency" as shown in JK is your actual interface's latency (round trip) as well as some stuff happening in the computer / JK software.  So even if your interface specs a latency of 1ms in and 1ms out, your "interface latency" as displayed in JK will be higher than 2ms.  Adding audio frame size to that for starters puts you at at least 3ms (with an audio frame size of 1).  I think there are other things in the software contributing to the number, but I don't fully understand them yet.  Core Audio (Mac) is definitely buffering and that is adding on as well. Not sure what those numbers are.  Does anyone know?

6) Ethernet / wifi:
If you don't already know this you MUST be on a wired ethernet connection and NOT wifi for this to work.  This really should be number 1 on the list, but I'll assume everyone already knows this if you made it to the forums. 

7) Modem:
Make sure you have a good modem with the most up to date software / firmware. 

8) Internet:
The better your internet service, the better your experience will be.  Get fiber if you can.  You need good download AND upload speeds.  You can run tests to check your speeds both ways, as well as ping and jitter which are both very important.  Also look at packet loss if you can see that.  There are many sites on the web to use for this.  A quick Google search will bring up a bunch of them.  I recommend trying a few different ones. 

On a Mac you can even go into Terminal and type "ping 8.8.8.8" and press enter.  Press "Control" and "c" together to stop.  It will then show you info on packet loss and ping stats.  These may differ from what you actually experience in a session depending on providers and where everyone is.  There are other number / sites to try as well (8.8.8.8 is one of Googles).  Check this out or search for yourself:
https://etherealmind.com/what-is-the-bes...onnection/

I'm sure this can be done on PC as well, but I don't know how.  Any PC people want to chime in?


Those are the things I know of at the moment.  I was hoping to come up with 10, but oh well.  I hope this helps some people.  Good luck and let me know if you find anything else or have additional info on what I wrote (or corrections).
  Reply
#16
Fiber is NOT better than "old" copper - for JamKazam

I firmly believe.

From search on the WEB:

"Is fiber better than copper?
The Difference Between Fiber and Copper
Underlying technologies cause the bandwidth gap between fiber and copper. ...
Copper uses electrons for data transmission, while fiber uses photons.
Light is faster than electrical pulses, so fiber can transmit more bits of data per second and offer higher bandwidth. 23. jan. 2019""

"... more bits of data per second and offer higher bandwidth ..."!?

Nothing about PING - PING is very crucial.

PING for fiber is NOT better than for copper. In short

Of course - one need to find better documentation for that claim.

Maybe I can do that

Now there are at least two threads about excactly the same topic.

It is confusing - not good

Now one had to post the same message different places.

In fact - it is a total waste and crazy

  Reply
#17
New version of the Focusrite low latency installer (djs, your link is dead): https://support.focusrite.com/hc/en-gb/a...ces-on-Mac

It's almost the same kext I posted before (1.1.7) but this one supports 3rd gen interfaces.
  Reply
#18
Thank you for the very useful post! This is generally consistent with what I am finding out as I try to reduce latency with a cheap USB audio adapter (driven by tape out from a PA head) with the ASIO4ALL driver on a Windows 10 system. Not the way to go, but almost adequate until I get my hands on a real audio interface. I've been able to beat my hardware latency down to 15 ms as reported by solo session diagnostic, I measure 14.1 ms with a scope looking at analog in/out to the USB audio adapter.

Frame size seems to have the largest impact on latency of any of the knobs I can turn. With the ASIO4ALL driver I was expecting to see a difference in latency as I vary buffer size in the ASIO4ALL settings, but I do not. The Buffer In/Out counts also do not seem to affect latency. I've experimented with power of 2 buffer sizes from 64 to 512, and In/Out counts from 0 to 3.

Changing the USB audio adpater sample rate from 44.1 to 48.0 kHz made a difference. This is about a 9% increase in sample rate, and produced about a 6% reduction in latency. I haven't attempted to change any of the other settings you mention, such as the bitrate settings.

As far as fiber vs. copper, there are so many factors affecting latency that it is difficult to make a broad and correct generalization. More bandwidth does not necessarily mean less latency. It is entirely possible for a 20 Mbps copper internet connection to have lower latency than a 200 Mbps fiber connection. It really depends on your in-home network interface and the equipment your internet service provider deploys in the path between your network interface and the external internet. A large system that aggregates many high bandwidth fiber channels and does a lot of buffering can introduce delay.
  Reply
#19
Interesting information Matt.  I am starting to wonder if there is a difference between how (where) the buffer if accounted for on Mac vs PC.  You saying that you did not see a change in your latency when changing buffer settings makes me think that it is possibly being factored into the internet latency rather than the audio interface latency (which is where I believe it should be, but who cares). 

If you can join a session with someone else, have them look at the different latency numbers as you change the buffer and see what happens.  This would also explain how some PC users are seeing 0.0 interface latency.  I can't change the buffer on the Mac (core audio) so I can't test this, but I have a feeling that the buffer is factored into my interface latency (because it is higher than it should be with just my interface).  Oh, and also that some Focusrite users have reported that their interface latency went down with the use of the Focusrite buffer app.  I could be wrong though.  It would be great if someone from JK could clarify for us so we don't have to keep guessing.
  Reply
#20
LATENCY, mostly TOTAL LATENCY

> https://forum.jamkazam.com/showthread.php?tid=171

In some cases Jamkazam is not able to figure out the audio interface latency

That happend for me - with a Scarlett 2i2 2th Gen.

And for others too.

Dont believe a second on that info about that ZERO latency. Zero is impossible.

In my case it showed up to be - 7.3 milliseconds
  Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)