Apogee Symphony Users... 1.6 ms of latency? Hype, or the real deal?

Hi GP,

Thanks for the detailed information about differences in delay times. I appreciate your observations especially because you've tested this out on several system.

If you record a loopback via digital outs with RME hardware it shows up early, if you record that via analog it is recorded more or less 1ms late which IMO is the DA conversion time.

I would agree.

Cheers!

Hey Rohan,

Velvet ears

They are indeed velvety, but only so because my wife is away for a while and there's no one to pluck. Er, my ear plumage, that is, pluck my ear plumage... 😉

On to the point of 1.6 ms's of latency and whether or not it's perceptible, I find it educational (and yes, fun) to read all views on this. But I've already expressed my views and I don't think I have anything more to add. It also occurs to me... if whatever latency you might be experiencing (and finding acceptable) wouldn't be acceptable to me, it certainly doesn't reflect poorly in your music (as you know, I've heard some of your stuff and raved). The emotion and sound quality in the few pieces I heard should serve as a benchmark for quality and taste for everyone.

Many times the music transcends the technology. In fact, just last night on the radio I heard some Stravinsky. It sounded to me like a live broadcast (kinda lo-fi). I thought maybe it was being broadcast from Tanglewood (this is the season for that kind that kinda thing around these parts). Turns out it was a recording from the 1930's with Stokowski conducting. Surely a restored recording, but the emotion came through even on the limited bandwidth equipment they had a the time.

This isn't to say that I've changed my views on how important 1.6 ms is or not. I just know that I can sense 1/3 of that timing and it will bother me. But if it doesn't bother you and others, and you can still make good music, than so be it.

The recording delay thing is a whole other ball of wax that I would do battle with the mightiest of mighty non-believer-in-the-importance-of-setting-its-value-correctly warrior type. There can be no argument that the A/D process takes time and that recordings will arrive (typically) late into Logic unless, like the RME and Apogee stuff, their drivers compensate. Since not all do, recordings will typically be lage. I mean late. Dam-n ear plumes falling on the keyboard againmn... Tch... there's another one!

In all my time in online audio forums have I have come across such complete nonsense. This is the kind of debate between a people with an IQ off the charts and could not figure their way out of a paper bag. Audible latency between the snare and the ear of the drummer, what a load of crap.

Ok, I own a Protools system, and I’ve owned 2 symphony systems. There is no latency that you are going to hear, even at 44 when running a buffer of 32. I could track with a ton of tracks, with a ton of plugs, all while tracking through as many plugs as I want, with no latency worth ever talking about.

The only type plugs that will cause latency issues are DSP/FW plugs, like UAD, Liquid Mix, Powercore, and Duende.

So anything you’ve read to this post, you have just wasted precious minutes of your life that you will never get back reading post by people who should get a life.

iSchwartz, if that little latency bothers you so much, how the heck did you ever deal with the latency from other players on the other side of the stage playing live?

Small degrees of latency don't bother me, I adjust quickly.

A few random comments to some of the posts above (I haven't read them all)

The latency of a Symphony system is 3.36ms @ 44.1kHz according to this review in TapeOp: http://www.apogeedigital.com/pdf/symtapeopreview.pdf . The latency of a PTHD system is 2.38 ms @ 44.1 - using the 192 I/O interface - according to Digidesign (confirmed by my own tests when I still used PTHD hardware). The difference is just below one millisecond.

The latency of Lynx Aurora was reported in some Gearslutz thread a while ago - I don't remember the exact value.

Since PTHD always is in Software Monitoring mode, you need to use a non-Digidesign interface (not supported by Digidesign) or a native rig if you want low ITB latency. This can be achieved by avoiding software monitoring.

Some people seem to forget that humans naturally compensate for sounds coming from a distance (the ears/eyes/brain seem to cooperate a little here....) - which doesn't mean that it would feel 'natural' to hear one's own voice (in headphones) a few milliseconds too late (along with the sound of our voice from within our heads - simultaneously).

In short, people who want lower than 2.38 ms latency in their DAW (@ 44.1 khz) either need an external monitor mixer, use direct monitoring natively, use TDM with non-supported hardware or find a native system/interface with lower latency than Symphony. I, for one, don't which system that would be - but maybe others have some info?

Some more numbers, posted by Lynx' David Hoatson in another forum:

"Round-trip for software monitoring is then 27 samples plus 2X the buffer size. So for software monitoring with a buffer size of 64 that would be 3.23ms total. With hardware monitoring it would just be 0.56ms."

This probably means 2.06 ms @ 44.1 kHz using the 32 buffer - equalling 2.24 ms @ 44.1: http://www.gearslutz.com/board/new-product-alert/104282-lynx-aes16e-pci-express- card-3.html#post1802057

According to another post in the same thread, Symphony (and probably other native interfaces as well) also have circa 0.5 ms roundtrip latency when bypassing Software Monitoring. AFAIK, the fastest converters out there use only 8 samples, meaning that DA and DA would use a total of 16 samples (0.36 ms @ 44.1) for converting alone, and when monitoring in direct mode, there's isn't much that can add latency other than the converters, except possibly if you use a 'hybrid' system with separate monitor effects in the I/O itself.

Both Symphony+Logic operate in the 2-3 ms roundtrip latency range, Lynx seems to be a little faster - even when using software monitoring, and for for the optimum "No Latency Experience", new systems can probably offer ITB latencies in the sub-millisecond range since newer converters are faster than those made a few years ago. Since PTHD always is in Software Monitoring mode, direct monitoring isn't possible here (without an external mixer).

I used Logic with PT hardware for many years before I went native, and can't recall anyone complaining about the 2.38 ms latency in their headphones, and nobody seemed to notice that PT HD had higher latencies than PT Mix, which was in the 1.5-2 ms range either.

ZXC wrote:

This probably means 2.06 ms @ 44.1 kHz using the 32 buffer - equalling 2.24 ms @ 44.1:

I can't edit my post, so here's a correction:
This probably means 2.06 ms @ 44.1 kHz using the 32 buffer - equalling 1.9 ms @ 48 khz.

Since we now can run hundreds of tracks on a fast 8-core using the 32 buffer, maybe we'll see a 16 samples buffer soon, offering latencies around 1.3 ms (@44.1 khz) and 0.6 ms (@96 khz) with Software Monitoring on?