Anyone else noticing higher than normal latency when using this otherwise wonderful sounding piano?

I have my Midiman Delta 1010 card set for 64 samples (2.4 / 2.0 msec at 24 bit) and I still hear a perceptible latency.

I have turned the "Finger Attack" adjustment all the way down and while it's better, there is still an annoying lag.

For the record, playing the NI NY Concert Grand through the same Kontakt player gives absolutely no latency at all and my other pianos like Ivory and Garritan are equally responsive with no perceptible latency.

Anyone else experience this with the Alcia Keys program?

I think 2ms - it is a about key pressing time... (B.Joel - 13 times/sec - about 30 key presses+key releases (I think key press times less then key release) - about 30ms or less for key press) and 2ms you can not fix in any cases. But latency depand on software and hardware also, and its can be reason for big latency (CPU? HDD? RAM?)

Yuri, I don't think it's a hardware problem since Ivory and other software pianos which play through the same software sample player are just fine regarding latency.

even when its turned all the way down there is still finger noise and the hammer "swoosh".

Scarbee said he made a patch to turn the latency and remove the extra latency, so it's in Native Instrument's hands now as to when they publish the patch.

Basically he'll add a bit more room on the latency/finger knob so you can cut off a bit more of the pre-tone attack noise.

What are you opinions otherwise on this pretty awesome product?

I have found it to be really "smooth" the tenor bass and treble all meld nicely, sounding more like a modeled piano than a sampled one. This piano really shines when you add some vocals or if you play jazz or broadway tunes. I think it lacks a bit of bite, this sort of rawness that you get in a big steinway that helps classical pieces stand out.

btw when I turn the latency knob all the way down it is perfectly satisfactory for me. Maybe I can tolerate the extra few milliseconds better.

It can be hardware problem because some software pianos have different system requirements, and Alicia Keys can require faster HDD (I don't know, but for example in my system I have no delay with Ivory and Pianoteq, but big delay for Grand 2 - I use asio4all).

Alicia's Keys it not software. It is a sample set. (Just data.) that is loaded into the Kontakt sampler (which is software) He is saying that he gets different latency with this sample set compared to a different set of samples loaded into the same software.

In Kontact you load can play more then one sampled instrument at the same time. So with the Steinway and Alicia's Keys both loaded playing you still notice one is delayed. Because both instruments are playing at once inside the same software, both see the same effect of slow disk or CPU.

All that said, we know the true answer. The person who made this decided to emulate the latency that is in every real acoustic piano. It takes time for the hammer to move, about 5 ms. and then your ear is not on the string so there is a speed of sound delay even after the string is struck. That delay is about 1ms per foot. He wanted to make this work just like an acoustic piano.

You have to read the description of the product. This was intened for Alicia Keys own kind and style of music. She does not play complex classical solo piano. In her music the piano is down in a mix and not a solo instrument. It plays a backing role mostly to a female voice and hence wants to stay out of the range of her voice as much as a piano can.

Most of use cannot hear these small delays (I can't) but apparently some have better ears. I wonder if those people are bothered when they play acoustic pianos?
Most of the time the brain compensates for the delay. for example when you talk to someone who is 25 feet across a large room you do not notice the 25 ms latency or that his lips are out of sync with the sound. But if a DVD video is 25ms out of sync it looks horrible It's possable the brain can adjust for an acoustic pianos but but a DP or TV set?

Really? How do we (you, actually) know the true answer? You know this for a fact, or this is your opinion? (Perhaps you have been in contact with the developers?)

You seriously think that a developer of a sampled piano would purposely introduce latency? Really? (I DON"T!)

I would suggest that it is likely a setting in the Kontakt player that needs to be adjusted. (BTW, there are solo piano demos of this piano, by Alicia herself, so I'm not inclined to believe that her piano is used only in a mix.

Speaking of latency, I was playing around with the parameters of the Roland FP-7 that I play at school. I introduced a ton of latency - and this with just the digital piano - no computer involved at all! (Kind of demonstrates the importance of parameters). Of course I immediately changed the offending parameter back. Why would I want latency introduced?

I would suggest that the OP check the FAQs and contact tech support, if needed.

Lawrence

well yeah we are in contact with the developers, go look at the thread here

http://www.gearslutz.com/board/new-product-alert/358437-alicia-keys-piano-18.html

ps scarbee just described a 32 millisecond latency bug and posted his fix for the 2ms latency that you can do by hand, and he apparently also got the approval to push out these changes via native instruments (I didn't read the whole thread), you can try these two fixes if you think you are running into this particular bug or want to cut the extra 2ms off of the pre-tone hammer swoosh sound.


3. latency knob at 0% (but this time sustained - pedal down): I got about 30 ms latency (peak at around 31-34 ms)!!!
3. latency knob at 100% (again sustained - pedal down): I got about 30 ms latency (peak at around 31-34 ms) again!!!
Guys.... we just found this bug. The pedal down groups seems to have cc#1 to control latency and not CC#110. That explains the problems!!!

THANKS a lot...

If you have Kontakt you can try this:

Open editor and select a PD group.
Click on SOURCE - MOD flag.
Look at midi CC - it says 1. It should be 110.

there are 8 PD groups -all has to be changed this way.

To add the extra 2ms latency cut try this:

click EDIT All groups
Click on SOURCE - MOD flag.
Look at midi CC 110
Move slider so it displays 59,8%

we will fix this in the update.
__________________
__________________
Thomas Hansen Skarbye
Singer, Bass Player, Composer
& Creative Director, SCARBEE

Thomas Skarbye himself posts on gearslutz.com using his own name. He answers questions on a public forum.

BTW that 32 ms thing is a bug, not designed in intentionally but the smaller delay is I think on purpose.

Unfortunately, most of this is impossible, at least for real time playing, and using typical MIDI controllers. On a real piano, the click of the finger hitting the key reaches us very rapidly - if our ears are, say, 3ft away from the key, then it will take just 3ms. In a software instrument, we have a whole chain of things that add delay, and the click won't reach us until AT LEAST 20ms later - often more. We're behind from the very beginning, just due to the time it takes for the key to move down far enough for the controller to sense that the key has moved at all, which I think is between 10 to 20ms easily for a forte note! A continuous MIDI controller would start to make "pre-attack" sounds like this become more plausible for real time playing, because things could start to happen as soon as the key starts moving. (albeit with the rest of the delays in the system) Regarding the 5ms for the hammer to move, from what point are you taking the measurement? For a forte note, it takes typically between 20ms to 30ms for the strings to start vibrating from the time our finger touches the key - ref:
http://www.forum-pianoteq.com/viewtopic.php?pid=8196#p8196

For real-time work, and using today's typical systems, I think that everything should be done to minimise the latency. The samples should start playing as close as possible to the beginning of the attack sound of the hammer hitting the strings. For non real-time work, then pull out all the stops and do the rendering as authentically as possible. (and for pre-attack sounds, I guess this would require special processing to ensure that these sounds occur a bit earlier than the sound of piano proper!)

Greg.

Wow! I had no idea this would generate such interest !
In a nutshell, THANK YOU for leading me to the gearslutz thread.

Yes, this IS a software problem as I run a small project recording studio and the machine is an i7 with 16g memory and very fast Seagate 32mb buffer drives.
I run Nuendo and Protools.
Trust me when I say there is no hardware problem.

There is some validity in that Ivory for example *does* use it's own engine so in effect it's apples (sorry for the pun) and oranges. However, a 2 msec latency is just that. The time it takes through the hardware from input to output. While the *playability* of the various different pianos will be subject to their own idiosyncrasies, the latency should at least be close especially when running dry with no bells and whistles added.

This is the simple version, there is some cheating especially with firewire interfaces and internal buffering, but the Delta 1010 is a PCI card that is known for it's very low latency and stability.


I'm glad Scarbee is aware and they are working on a patch.
That's great response!!!


So now that we have gotten that out of the way, how does it play and sound?

In a word, WONDERFUL......

It has it's own character, but this is *not* Akoustik piano re-regurgitated.
I hate Akoustic Piano BTW.

The playability with a good weighted controller is quite pleasant and the pedal samples, mic noise, sympathetic vibration and so forth all add to the sound quality.

Notes are smooth, ramp up in volume and tone when playing louder and I haven't heard any obvious sampling artifacts. In my opinion it's pretty close to Ivory, Garritan but different. It is far better than any modeled piano I have tried because unlike the modeled pianos, the dynamics are alive with this program.
I start playing this sample and my creative juices get flowing!

For $99.00, this is the deal of the year for people looking for a very, very nice sampled software piano that competes with software costing 5 times as much.

Well done!

HTH

What controller are you using? Thanks.

Lawrence

I use a Roland RD-700 and also a Roland RD-150.
hth

I believe it is impossible for a software to emulate the latency and the early noises of a real piano.

Scientific research has shown that the key travel velocity is about 1 m/s at forte level and that this velocity is rarely exceeded. I myself measured the key velocity that is necessary for a value of 127 with my Kawai piano. The speed required is about 1,5 m/s. So that is ff or fff level.

The key travel distance is about 10mm.
So the key reaches bottom after about 10ms. But if the key is stroken hard, the impulse of the stroke flows through the wood faster than sound and the hammer reaches the string /before/ the key bottoms out.
This was found by measurements and scientific physical measurements on a real grand piano.


It is also possible on a real piano (and with my Kawai keyboard) just to apply a stroke to the key and the hammer transports this stroke to the string, but the key does not bottom out at all.

At pianissimo level, the hammer normaly reaches the string after the key has bottomed out, because the hammer has a free flying phase (the blow distance) and at slow velocities and large blow distance the hammer needs more time to travel than the key. It is, however not desirable to emulate this behaviour. Ideal where a blow distance of zero and then this delay would not occur. Piano technicians try to reach this ideal, but this adjustment is so sensitive, it must be repeated after some weeks, so this is only done for very expensive pianos (and expensive classical pianists ;-) So it is not a disadvantage if this delay is eliminated with a digital piano.

The Softwarepiano cannot emulate these delays. I think it is nonsense to try this.
It should also be considered that a masterkeyboard will introduce some delays that are inherited from the construction and the mass inertia of the mechanism.

If there is something, that should be emulated then it must be done inside the master keyboard mechanism.

Another Question for the Alicia keys piano:

In another (german) forum I have read, that it is almost impossible to play pianissimo with this instrument, because even the lowest velocity will give a much louder tone than other software instruments (e.g. Galaxy II)

The poster said, he where unable to adjust this behavior.

Is he right?

Peter

Peter,
With continuous controllers (and sufficient processing power!), it should be possible to emulate those finer details pretty well I think. Maybe pressure sensors on the top of the keys will also be required(?)

Greg.

Thats theory ;-)
I am mostly interested in possibilities that can be made with todays masterkeyboards and softwarepianos.

I believe the Kawai keyboard ist already pretty good in emulating the real thing. The only thing missing is a 3rd sensor for the damper.

However, I have the idea to get a keybed and mechanism from an real acoustic, fit sensors to it, buy a Casio PX 130, and connect the sensor switches to the keyboard scanner.
Dont know, if I ever will do this. First I must find out if the PX130 will really generate the Noteon and Noteoff events in that order, that is required...

Edit:
Of course I personally dont need that. I am pretty much happy with the AWA Grand Pro now as it is. I would only do that if there where a chance to expand this into a small business.

Just got another idea:
Get a cheap 88 key masterkeyboard. Fit this into the Kawai CP136 in this way that the keyreleases will press the switches of this keyboard. Filter off all Kawai Noteoff events. Convert the Noteon events from the additional keyboard into Noteoff events and merge the MIDI streams. This way I get a 3rd sensor with keyoff velocity. It happens that I have an unused Masterkeyboard at home. Possibly I will do this ;-) Will cost almost nothing for me.

So far I have read, the upcoming update for synthogy ivory will support keyoff velocity. If that is true and if it works well, that would be worth some work.

;-)

Peter