STS latency test, worse than Sofware sampler.

lagoausente · Post by **lagoausente** » Thu Apr 06, 2006 2:37 pm

I tested the Scope I have got just today. I´m disappointed with the result. I take a mic, and used as trigger. Used Roland Spd11 trigger input to get midi.
Took the soundcard analog output and recorded an estero file, on channel the mic, the other the sample.
Using STS the sample arrived 7-9 ms later then the mic.
Using my Terratec EWs88mt, and a sofware sampler in SX, the sample arrived 4-5 ms later than the mic.
I was hoping 1.5 ms, but I can´t get this.
Ullit settings don´t affect, result is the same. STS is conected directly to midi A input (not emulated midi at all), and just to analog dest.
Hope, I´m having any issue on my settings to someone of you can find, and help...!
If no, want a disappointment.

voidar · Post by **voidar** » Thu Apr 06, 2006 5:47 pm

Could you give a more detailed representation of the signal-chain you use, for both tests?

Also, a picture of your scope-project would be nice.

MD69 · Post by **MD69** » Thu Apr 06, 2006 11:07 pm

Hi

If I do agree with the first part of your method, I do not agree with your conclusion.

The first part (STS recording) measure the STS + ROLAND latency, while the second part measure ... the roland latency due to plugin delay compensation.

This give us roughly 4 ms latency for the STS ... about the same than an AKAI sampler!

cheers

lagoausente · Post by **lagoausente** » Thu Apr 06, 2006 11:39 pm

The test was just same conditions for both, sofware and STS.

The signal goes this way:

STS

)
Mic-preamp(split)-
channel 1 to analog record (L
channel 2 to Roland SPD11- (midi out) to (creamware midi in),

routin window, midin in A to STS- STS out to analog source,- audio out to analog record (R)
So have on L the mic, on R the creamware out.

Sofware:

Mic – preamp (split)-

channel 1 to analog record (L)

Chanell 2 to Roland SPD11- (midi out) to (Terratec midi in),

-to Cubase SX,-to sofware sampler-,- audio out to analog record (R)

Again, on channel L the mic, on R the Terratec out.

Recorded Stereo file, plugin delay compensation moves the entire file, not only or R from a stereo file. The comparison, is a real, from the mic (trigger), to samplers analog out.
In my opinion, something goes wrong, since I suspect that Midi in A , on Creamware routin window, seems not to go “really” , from in to dsp, for the resutl 8 ms, seems it goes to the computer and comes again to dsp, even when was not routing that way at all.

[ This Message was edited by: lagoausente on 2006-04-07 00:43 ]

djmicron · Post by **djmicron** » Fri Apr 07, 2006 2:27 am

could you share your project ?

i don't understand the test you've done....

MD69 · Post by **MD69** » Fri Apr 07, 2006 3:11 am

hi,

Initially I thought you were recording your VSTi and the mike separately in two files! routing seems correct from your description as well as the protocol.

You could connect the roland to creamware midi in and see if latency change when youo record your Vsti.

cheers

valis · Post by **valis** » Fri Apr 07, 2006 3:12 am

I have no idea what you're doing really, but I get the impression your testing setup is the issue atm. Share a screenshot and/or project file please. You can post a Project and a screenshot in the "Projects" area at the bottom of the forums and post the link in this thread.

I suspect part of the problem is a language barrier. What nationality are you? Perhaps someone here speaks your native toungue...

Also it takes a bit of time to get your head around the routing window, even though its mostly straightforward studio technique.

[ This Message was edited by: valis on 2006-04-07 04:18 ]

miguel · Post by **miguel** » Fri Apr 07, 2006 4:52 am

I believe lagoausente is spanish, aren't you? I or Nestor can help with that. If you wish just post whatever in spanish I'll do my best to translate.

From what I understand lagoausente is recording both the sound of his finger hitting the key through the mic and the output of STS to a stereo file. I think that adds the keyboard's MIDI latency that you'd need to subtract from those 9ms. Anyway that's my guess at it

Edit after thinking a bit with a coffee in front of me: Absolute numbers don't matter here. I think lagoausente did the right thing just comparing with whatever other software. So if a softsampler + some other soundcard is faster there's no much to argue, no matter the actual numbers.

[ This Message was edited by: miguel on 2006-04-07 06:21 ]

voidar · Post by **voidar** » Fri Apr 07, 2006 5:43 am

What is this "analog record" that he speaks og? I want to know what you are recording on.

Not that it has to matter. You're doing it for both tests.

MD69 · Post by **MD69** » Fri Apr 07, 2006 6:12 am

Hi,

I think he means scope physical audio in!

If I am correct, signal get OUT then back IN.

cheers

voidar · Post by **voidar** » Fri Apr 07, 2006 6:51 am

MD69: Right

lagoausente: I suggest you try the same test meassuring midi-latency - just record midi using both interfaces, in SX or something.

Anyway, your findings might be correct of course. Though I would neverswap out my Pulsars with Terratec gear. Thing is, you could add a gazillion scope-plugs after the sampler, and latency would stay the same. It's very much fixed.

Shayne White · Post by **Shayne White** » Fri Apr 07, 2006 8:44 am

Are you using an open-air microphone to measure latency?? That doesn't seem very efficient to me...or am I wrong? Like others here, I don't understand your setup.

Shayne

lagoausente · Post by **lagoausente** » Fri Apr 07, 2006 10:13 am

Hello again, I´ll explain it clearer.
The Roland SPD11 (for the ones who doesn´t know it) is a drum machine, that has 8 pads, midi out, and 4 analog inputs for external pads. A mic can be used as a pad.
I conected one sm57 to a preamp, and took the line out into the Roland. When touch the mic, the Roland sends a note on through it´s midi output.
Playing latency, is the total time since we press a key (keyboard), or touch a pad and hear the sound, so the test is really very easy to do. Just compare the time between the mic signal and the sound coming from the samplers.
To do this, just record an stereo file, on left channel, the mic signal (as reference), on the other channel the sound coming from the sampler, first with STS, and then with a VSTi one.
On the STS setup, just use the CRemaware analog output, and on the VSti, used the Terratec analog output (what we really hear)
For the recording, I used 2 analog inputs of the Terratec one.

On STS setup, simply conect the midi out from the roland to the Creamware card. Load sts, and route it to the analog output, very simple.
On the VSTi setup, just conect the midi out from the roland to the Terratec midi in, load the VSTi sampler in Cubase, using the Terratec driver with buffer at 64 samples, and route it to the analog output, just the same as before, and very simple too.
The result, is not only the STS sampler latency, but is a A-B comparison.
On the 7 ms result, 1ms is the time the roland takes to convert the pad input to midi (what have measured some time ago), and perhaps another ms for the DA on the card.
But for the comparison, both a present in both tests, so no one have advantage on the result. Both have D/A, and the Roland (audio to midi).
For give some calm for the "believers" of business advirtements, I have a good new.
I replaced the STS for Modular II.
I got only 3.5 ms, comparing with the 7 ms on STS.
If we substract the 1ms of the Roland, and another for the D/A, we have 1.5 ms, that is just what Creamware talks about, so Modular II itself has 1.5 ms.
But if we do the same cound on STS we have 7-2= 5 ms. So STS requires more time, I suppose, because of the needing to use the system RAM.
Here I probably would need to be engineer to can setup properly my BIOS to can improve the performance of pci to ram speed. Tried Turbo mode, in Bios, sigle and dual channel, decreased the Video card latency, with just the same result.
So as conclusion, total latency comparison (included roland):

Modular II: 3.5 ms
VSTi, with Terratec.: 4.5- 5 ms
STS-4000 : 7-8 ms

lagoausente · Post by **lagoausente** » Fri Apr 07, 2006 10:20 am

http://www.planetz.com/forums/viewtopic ... &0&start=0

here is the proyect, really nothing new.

djmicron · Post by **djmicron** » Fri Apr 07, 2006 12:33 pm

how do you record the sts output ?

In your project i don't see any asio destination connected to sts, how do you make the latency test ?

lagoausente · Post by **lagoausente** » Sat Apr 08, 2006 4:36 am

As I explained here, I route STS to analog output, then just take a jack-rca cable, and conect Creamware analog output to Terratec analog input and record it. As well as the original mic signal. So I can compare the time between the mic signal, and the final Creamware analog output.
Don´t need Asio at all, just route from STS to Analog output.

kensuguro · Post by **kensuguro** » Sat Apr 08, 2006 6:01 am

I wonder what the difference is between different STS models.

But anyway, I doubt it's worth it to beat the STS for performance. I'd stick with a soft sampler whether it's slow or not. Unless there's a specific reason you really need to use STS.

[ This Message was edited by: kensuguro on 2006-04-08 07:04 ]

lagoausente · Post by **lagoausente** » Sat Apr 08, 2006 6:10 am

The reason I was going to STS was just because of it´s supposed performance, just have a look to my other thread, there is a midi issue here, in my opinion a quite big bug.

voidar · Post by **voidar** » Sat Apr 08, 2006 7:41 am

On 2006-04-08 05:36, lagoausente wrote:

As I explained here, I route STS to analog output, then just take a jack-rca cable, and conect Creamware analog output to Terratec analog input and record it. As well as the original mic signal. So I can compare the time between the mic signal, and the final Creamware analog output.
Don´t need Asio at all, just route from STS to Analog output.

When using the software-sampler, do you route out of the terratec and then into it again?

like:

trigger -> terratec MIDI -> Cubase w/sampler -> terratec D/A -> terratec A/D -> Cool Edit
+
Mic -> terratec A/D -> Cool Edit

[ This Message was edited by: voidar on 2006-04-08 08:42 ]

djmicron · Post by **djmicron** » Sat Apr 08, 2006 8:35 am

On 2006-04-08 05:36, lagoausente wrote:

As I explained here, I route STS to analog output, then just take a jack-rca cable, and conect Creamware analog output to Terratec analog input and record it. As well as the original mic signal. So I can compare the time between the mic signal, and the final Creamware analog output.
Don´t need Asio at all, just route from STS to Analog output.

in this way,
the sts/scope latency is added to the terratec latency.
i really don't understand this kind of test and why do you record the sts output from the terratec when you can use the asio destination....