Symphony + Soft Synths - latency killer?

I think you'll find the answer here:

http://www.apogeedigital.com/products/symphony_performance.php

Very important to note, however, is that the 1.5 ms round-trip latency spec they quote is based on a sample rate of 96KHz. So if it works like most systems (note the emphasis on "if"), the round-trip latency will be double at 48K (3 ms latency), and even longer at 44.1 ( ~3.5 ms). If my figures are correct, then IMO such latency figures would be less-than-impressive. But... perhaps a call to Apogee would be in order to find out for yourself what the numbers are.

Maybe I'm just losing my touch with old age...

nah mate its fatherhood sapping your energy. i know exactly how that is....

iS has it exactly right, though i would disagree with him on how 'unsatisfactory' 3.5ms latency is. most analogue desks have a latency of around 2.5 ms. and that is consider zero latency!

the figure that they talk about with symphony is round trip latency, so your soft synths will share that latency so that they are exactly in time.

i have quoted this often, but if you work out how fast sound travels at room temperature and sea level, 1.8ms is about how long it takes for sound to travel 3 metres. when you consider that the round trip from mic to cans AFTER processing (of non-latency inducing plugs of course) this system theoretically improves on the latency of....well...sound through air! err.. if you are in a band and the drummer has really bad body odour so you have moved well away from the rest of the band....

hi,
i often find the same thing with softsynths.
hardware ones have more character.
iron in the circuits is one factor.
output circuitry, another.
the symphony card will give very low latency, PCI cards are superior to FW most of the time.
it'll also put a huge dint in your credit card.
DR9.

Rohan,

Beg to differ with you my friend...

True analog desks have actual zero latency. You should be able to play audio into an analog board, take the output of the board and patch it back into some channels, flip those channels out of phase and hear silence (provided the gain staging across the board is the same). Though if you were to do such a test I'd suggest keeping the volume down to a whisper because of the possibility of a loud feedback loop that might occur as you adjust levels to achieve cancellation.

A single MIDI note takes approx 1 ms. from beginning to end. By this time we're all kind of used to that bit of delay, if we notice it at all. The amount of time a synth or VI reacts to it gets added to that (and some plugs have a fixed amount of latency they need to calculate their audio outputs, and in many cases it can be significant). So add to that 3.5 ms. of latency and... well, you get the picture.

The problem gets even worse if you need to increase the buffer size in order to accommodate proper playback of a song using lots of plugs, etc. So in my way of thinking, every millisecond counts. The only amount of time that can be considered zero latency is zero. Otherwise, claims of "near zero" latency are akin to, say, a food product labeled "now better tasting!". The question to ask is, "How much better tasting? Can you quantify that?"

In a recent post on the Lynx website, I inquired about their "zero latency" specs on an interface card. While their latency specs seem to blow the Symphony's latency specs completely out of the water, their "zero" actually constituted a few samples. That's not zero. Sure, a few samples is truly insignificant, but my point is this: it's not a good idea to take for granted the label of "near zero" or "zero" latency figures, because one company's "near zero" can be tens of times longer than another's.

The whole "software monitoring" thing is a crock AFAIC. More people have posted more woes about latency problems on forums such as these because they (erroneously) believed the sales pitch that they could replace their mixing boards with software-based monitoring systems.

And when in the studio, I always used to feel detatched from the tracks I was recording on, regardless of how loud playback was in the room, if I was too far away from the speakers. I always liked to be right close to the them. It felt better. But that may just be me.

Message was edited by: iSchwartz

=iS=.
true.
electrons running thru the console are at a rate of ~300,000/sec.
so, no latency, evenr.
DR9.

that's ~300,000MILES/sec, sorry.
DR9.

just to go slightly back on topic...

I've inverstigated the very matter of softsynth latency with symphony to some depth, including talking to apogee directly about it. after using our symphony setup for a while, we've found that while it's able to give respectable soft synth latency performance, it's not a magic wand.. in the sense that it doesn't suddenly give you faultless and consistent low latency performance with no chance of crackles or glitches and so on. just FYI, we were able to get at least 130 or so voices playing live (by bashing on a keyboard with the sustain pedal on) with ivory piano at 96k sample rate, with a sample buffer of 32 on a mac pro, with no crackles. however, this test was not really useful for anything, because the minute we tried to use such fine settings on a real-world song with a (large) variety of (big) sampler plug ins, synths and plug ins etc, it seemed to be no less prone to wigging out than any other logic system I've used such as with an RME. but hey, it sounds good anyway..

I'm going to suggest, based on latency specs only, that you investigate the Lynx system. Check out this link and scroll down to the reply to my question by one of their forum admins, PaulTech:

http://www.lynxstudio.com/forum/forum_posts.asp?TID=1581&PN=1

He's talking about latency on the order of a few samples, not milliseconds.

If we're going to compare thruput latency specs on a system, and the info as provided is assumed to be accurate, the Lynx is much better than the Symphony system in this regard.

Following up on ischwartz's comments on the lynx system, I did some online research and make a call to lynx studio. It seems that the latency they are reporting is indeed well below the symphony setup (which surprised the heck out of me) as well as increased stability compared to their competition.

I have been quite set on the symphony system for some time now, but I am quickly shifting my plans. It seems that a new PCI-e card from lynx is scheduled for release sometime in september, along with an intel mac driver. Performance from this system looks to be quite stable, and getting 16 ads AND 16 das for the price of one Apogee 16ADX is quite apealing. Most users are quite happy with the sound of the Aurora systems from what I have read (comparing it to the Apogee 16AD/DAX series more than rosetta or ensemble) and this seems to be a no brainer to me. Am I missing something perhaps? Apogee may have just lost my vote.

Also- to get slightly back on topic, the latency from a soft synth will have a lot more to do with the speed of your computer than the interface you are using. Having DAs that perform quickly will help as well, but the majority of the latency (afaik) will come from how fast your processor is, and correspondingly how low you can set your buffer size. Please correct me if I am wrong here.

I will also agree that an analog console truly is "zero" latency (Well, maybe a few trillionths of a second or something 🙂