You are here

how noticeable is 4ms of latency?????

For anything relating to music-making on Windows computers, with lots of FAQs. Moderated by Martin Walker.

how noticeable is 4ms of latency?????

Postby Sonny M's » Mon Dec 06, 2004 1:27 am

hello all,

i bought an m-audio 88 key recently am using it with reason and cubase 2.0. I have an m-audio delta 66, and the current reading is telling me i have 4ms of latency. since ive been using it regularly i dont notice the latency at too much: im in time with the bpm, but there is some massaging of the track that needs to be done.

however, is this 4ms noticable by someone who plays on a "real" piano? A friend of mine says it really messes with his timing. does he just need more practice, or is this 4ms noticeable? any opinions? :?

Thanks
Sonny M's
Posts: 2
Joined: Thu Nov 25, 2004 1:00 am

Re: how noticeable is 4ms of latency?????

Postby Jim Y » Mon Dec 06, 2004 8:48 am

Your friend just needs to get used to it.
A real acoustic piano isn't "instant on" from the keys either, compared to an electric organ.

Actual latency isn't only that which is reported when setting a soundcard buffer. Martin Walker did some tests on "real world latency" some time back, and it might help to search for the article.
Jim Y
Regular
Posts: 211
Joined: Tue Mar 30, 2004 12:00 am

Re: how noticeable is 4ms of latency?????

Postby Wurlitzer » Mon Dec 06, 2004 9:55 am

Your friend is full of sh1t. I'd defy anyone to consciously notice 4ms of latency.

As an example, it takes approximately 3ms for sound to travel 1 metre through the air. Adding 4ms latency is equivalent to moving about 1.3 metres further away from the speakers. It's nothing.
Wurlitzer
Regular
Posts: 403
Joined: Wed Dec 11, 2002 1:00 am

Re: how noticeable is 4ms of latency?????

Postby Jez Corbett » Mon Dec 06, 2004 10:13 am

Jim Y wrote:Actual latency isn't only that which is reported when setting a soundcard buffer. Martin Walker did some tests on "real world latency" some time back, and it might help to search for the article.

This is the real point. Just because 4ms is what the software reports doesn't necessarily mean thats what you are getting. If someone is complaining they can feel 4ms latency, its because they are actually getting a lot more than 4ms latency. The drivers could be reporting the figure incorrectly, or other factors such as the midi device, software, OS etc could be adding to it.

Are you using the proper ASIO drivers for the soundcard? Just that I noticed the ASIO DirectX drivers in cubase/nuendo are grossly innacurate.
User avatar
Jez Corbett
Frequent Poster
Posts: 526
Joined: Fri Mar 21, 2003 1:00 am
Location: Helsinki, Finland
http://www.jeremycorbett.co.uk

Re: how noticeable is 4ms of latency?????

Postby husker du b doo » Mon Dec 06, 2004 1:48 pm

4ms of latency - then add 4ms of midi latency and maybe 1 ms from the soundcard

9 ms or 3 meters away

thats fairly noticeable if youa re used to a piano - which lets face it is never more than a meter away

but its nothing compared to the experience of a big church organ player - and they manage

i think he just needs practice
husker du b doo
Posts: 1
Joined: Fri Sep 03, 2004 12:00 am

Re: how noticeable is 4ms of latency?????

Postby Arse Bandit » Mon Dec 06, 2004 2:33 pm

Wurlitzer wrote:Your friend is full of sh1t. I'd defy anyone to consciously notice 4ms of latency.

As an example, it takes approximately 3ms for sound to travel 1 metre through the air. Adding 4ms latency is equivalent to moving about 1.3 metres further away from the speakers. It's nothing.

If it's really 4mS then I'd pretty much agree. Thing is, it could easily double if you have MIDI latency to add on, and other rubbish. I've played on setups that report as little as 4mS and felt there was maybe 8mS or more - a figure which definitely is noticeable.

Fair comment also about a real piano not being 'instant-on', but on the other hand, if you want to go for some real low-latency, try playing a proper analogue synth with CV (not MIDI!!) keyboard, e.g. Minimoog etc. Now that's fast...

Disclaimer - I might be able to notice latency, but I am also full of sh1t.
Arse Bandit
New here
Posts: 8
Joined: Mon Sep 17, 2001 12:00 am

Re: how noticeable is 4ms of latency?????

Postby thefruitfarmer » Mon Dec 06, 2004 2:40 pm

Wurlitzer wrote:

Your friend is full of sh1t. I'd defy anyone to consciously notice 4ms of latency.


If I shift the snare even 1ms forward or back in time on the Atari Cubase then I notice a difference to the feel of the track.
User avatar
thefruitfarmer
Frequent Poster
Posts: 623
Joined: Wed Sep 01, 2004 12:00 am
Location: Kent UK

Re: how noticeable is 4ms of latency?????

Postby Martin Walker » Mon Dec 06, 2004 3:35 pm

OK. I've just a search through this thread, and no-one has yet mentioned latency jitter. It's this uncertainty in timing that is far more perceptable by most musicians , and not the absolute latency value. After all, as Wurlitzer pointed out, 4mS latency is simply equivalent to moving 1.3 metres away. However, if the timing has up to 4mS of jitter, that sound will be darting backwards and forwards by up to 1.3 metres.

In my two 2-part Real World Latency features I found that my PCI MIDI interface and serial port interface both offered latencies of the same order as playing a hardware synth (three to five milliseconds), while my particular USB one offered five to seven millisecond latency under Windows XP. In all three cases the jitter level was no larger than an imperceptible 2mS.

Playing softsynths in 'real time' is the worst case scenario, since you also incorporate the jitter of their software buffers. With a 128-sample buffer running at 44.1kHz the host application would typically report a 3mS latency, but once the MIDI and softsynth buffer uncertainties have been added, I measured 6.7mS to 9.4mS with a PCI MIDI input, and 7.1mS to 9.0mS with 64-sample GSIF drivers. The maximum jitter level here is 2.7mS for ASIO drivers, and an even better 1.9mS for GSIF drivers.

Now I don't personally think that many musicians will be able to discern a fixed 9mS latency, AS LONG AS THE JITTER STAYS AT UNDER 3mS.

Shifting drum beats backwards and forwards by tiny amounts to alter the feel of a track is a slightly different issue, since when playing back softsynths their timing will always be locked to sample accuracy (they are locked to the crystal oscillator of the audio interface that is also used for audio purposes). It's only when playing in 'real time' that these MIDI and interface buffer latencies enter the equation.


Martin
User avatar
Martin Walker
Moderator
Posts: 14975
Joined: Wed Jan 13, 2010 9:44 am
Location: Cornwall, UK

Re: how noticeable is 4ms of latency?????

Postby Wurlitzer » Mon Dec 06, 2004 3:45 pm

thefruitfarmer wrote:
Wurlitzer wrote:

Your friend is full of sh1t. I'd defy anyone to consciously notice 4ms of latency.


If I shift the snare even 1ms forward or back in time on the Atari Cubase then I notice a difference to the feel of the track.

That's different. You're talking about shifting one part, while keeping everything else as it is, so of course you'll notice the changed relationship between the parts.

Somebody playing a keyboard and noticing latency (or not) is a different thing. They'll still be hearing all the parts (ie, any audio tracks being played back, plus the keyboard part they're playing live) in the same relationship to each other. The question is whether hearing all this 4ms later seriously affects the relationship between their fingers and their ears.

I take the previous points about the added MIDI latency, and the fact that a real piano is never more than a metre from the player's ears. I suppose I've gotten used to the fact of hearing what I play coming out of speakers, just from gigging as well as studio work, and I don't expect it to feel like playing a piano. It's a different experience.

Can't honestly say that it makes it difficult to play in time, though.
Wurlitzer
Regular
Posts: 403
Joined: Wed Dec 11, 2002 1:00 am

Re: how noticeable is 4ms of latency?????

Postby Sonny M's » Mon Dec 06, 2004 11:55 pm

I see.

Another question arises then. How can I minimize the latency. I have the ASIO drivers on, and the settings are right within reason and Cuabse, but what about the settings within XP?

If this topic has already been addressed can someone lead me to the thread (im about to do a search as well).

THANKS
Sonny M's
Posts: 2
Joined: Thu Nov 25, 2004 1:00 am

Re: how noticeable is 4ms of latency?????

Postby Martin Walker » Tue Dec 07, 2004 10:39 am

There are no settings in XP - the latency value is set solely by the soundcard buffer size.

If you want to read more about setting latency, watch out for my next PC Musician feature in the January 2005 issue of SOS, out very shortly.


Martin
User avatar
Martin Walker
Moderator
Posts: 14975
Joined: Wed Jan 13, 2010 9:44 am
Location: Cornwall, UK