Forums > MaxMSP

usb device latency issue

November 9, 2008 | 6:01 pm

Hi,

When running a guitar through a mackie-xd2 (usb) and max 5 i find there is a very noticeable latency issue. Is it worth getting a firewire device to see if the latency will drop or will it always be the same through max.

My specs are

2ghz intel t7300
2gig memory
vista 32 bit.

Is there anyway to lower latency without getting a new audio device? Would using runtime cut back on CPU usage?

I cant remember if ive posted about this before but if i have I apologise.


November 9, 2008 | 8:18 pm

I am not sure abotu the runtime sissue, but you can definately cut back letancy by using a good firewire interface, namely an rme interface for instance. if you have a pcmcia bus you’ll probably be even better on the latency. check out http://www.rme-audio.de/en_index.php for both options.
besides the pcmcia card bridge you can get high end interface called multiface connecting to the pcmcia via ebay (out of production, but converters sound even better than the new rme interfaces).

if you use fft based max objects you will still have to add the latency caused by the fft; i don;t know figures, someone said approx. 70 ms. before.

jrp


November 9, 2008 | 10:00 pm


November 10, 2008 | 8:51 am


November 10, 2008 | 6:48 pm

Quote: jayrope wrote on Sun, 09 November 2008 15:18
—————————————————-

> if you use fft based max objects you will still have to add the latency caused by the fft; i don;t know figures, someone said approx. 70 ms. before.

I’m pretty sure like vector sizes, the latency introduced by FFT is based on window size and sampling rate.


November 27, 2008 | 3:37 pm

is the EDIROL FA-66 any good for guitar, there is a MOTU 828 mkii available for more money but i dont need 8 ins and outs. Would the quality/latency be much better on the MOTU?


November 27, 2008 | 5:39 pm

i use an 828 mkI, which doesn’t have much of a noticeable latency. i suppose the latency of a newer version is probably comparable (not necessarily better is suppose). see specs on motu.com.

… better than an edirol in any case (does it not have a direct instrument out or separate monitor output? this wold solve your latency issue eventually, at least for listening…)

i should mention, that i am generally unhappy with the converter quality of the motu 828 mkI, especially with bassy and/or distorted audio, which remain to sound "okay" only on much more expensive converters, like RME or Apogee…

jrp


November 27, 2008 | 6:34 pm

the edirol doesnt have specific monitor outputs. The latency issue isnt with the dry guitar coming out, its the hefty latency coming off max msp which needs to be cut down.


November 27, 2008 | 8:28 pm

awesome, at least that means you have to spent zero of 200 GBP.

jrp


November 27, 2008 | 8:33 pm

no point in getting a firewire device then?


November 27, 2008 | 8:43 pm

wel, you answered your own question. if the issue is related to max as you say, then it is not related to your audio interface (aside setting a low buffer size in your dsp panel, which should be set to 64 normally. that would mean roughly 1.5 seconds of latency. you probably wouldn’t notice this, unless your autistic…)

so what remains is looking for large buffer sizes anywhere in your patch, fft is a usual candidate to increase latency. a buffer of 1024 on an fft would mean roughly 25 milliseconds of latency. that starts getting noticeable, especially for the rhythmically talented…

jrp


November 27, 2008 | 10:20 pm

ok thanks for the useful info. Is there anyway to drop the vector size below 512 without distortion and crackling?


November 28, 2008 | 7:39 pm

is it possible to run signal through max patches in real time


November 28, 2008 | 8:07 pm

Quote: Nano59 wrote on Fri, 28 November 2008 11:39
—————————————————-
> is it possible to run signal through max patches in real time
—————————————————-

There are no computer-based systems that have zero latency. Different audio interfaces have different sizes of "safety buffers" that impact latency. Different device drivers have various minimum buffer sizes, and performance levels.

Max has two different vector sizes: signal and IO. Keeping them both low is better for latency, worse for performance.

Keep in mind the 1 ms is about one foot at sea level. Increasing latency by 1 ms is the same as moving one foot further away from your speaker.


December 1, 2008 | 4:47 am

–>

Is there anyway to drop the vector size below 512 without distortion and crackling?

–>

AFAIK this depends upon your hardware and drivers. ASIO is the best for latency, I’ve heard, and there’s ASIO4All which is an "add-on" ASIO driver for many soundcards which don’t natively support it. Haven’t tried it though.

Combinations of vector size and Latency (ms) can give you different performance options, just play around with them. Not sure how they interact exactly, one would think that the vector size would determine the latency, but perhaps that’s another latency in addition to the menu one?

Also sometimes I’ve worked with low latency and it’s good for awhile, then starts to crackle. Turning the DSP off then on often fixes this for awhile, but might not be desirable to do in performance.


Viewing 15 posts - 1 through 15 (of 15 total)