Sample pianos - R.I.P.? (Page 2 of 3)

In all the years I've been working on PC I never found a better solution for making Audio work proper than RME. (needless to say:Intell processor/Motherboard AND Graphic Card!! etc.)
With the new RME HDSPe  AIO you can afford 0,7 ms latency watch out for their products if you wanna get the most out of pianoteq!

sorry for that commercial break-I don't sell their products I use them.

http://www.rme-audio.de/products_hdspe_aio.php

After I've listened and played the new git. FX
AXE FX witch is a revolution for e-git. I always think of Pianoteq in a 19" Box with the same feeling- I would prefer it. What this guy is getting out of this normal processors is fabulous and shows that  a good Idea (and PTQ is one) can be transformed with the right tools.

You've made my point very well.  This soundcard costs more than my planned PC upgrade which I'm suggesting should be the PTQ minimum hardware spec!  I'm sure it's a very fine card but the law of diminishing returns suggests to me that throwing more cores at a more advanced model would give the biggest bang-per-buck.

Yeah, pretty much.

As the Borg would say, "Resistance is futile."

Or, as Sam (De Niro) says in "Ronin":  "If there is any doubt, then there is no doubt."

Let the OCD run free!!!

:^)

What scares me is putting hundreds of hours of work into learning software interactions only to have the programmer completely rewrite major sections, change the interface, remove or change tools beyond recognition.  I've had this happen before with synthesis software (the product was eventually cancelled completely) but I'm sure this wouldn't happen with PTQ, it's a professional product.

Yes, I am, with some caveats.  One can choose the data sampling period so that it is at a particular point in time or over a user-defined timerange (not really applicable for Piano). 

Example 1:  Plot the first 0.5s of the output of the fundamental (graph) with y axis as either frequency or amplitude. I'm not sure how useful this would be (in freq mode) because we don't have direct control over the oscillation.  It's not like organ where software could deliberately set to start a particular partial flat or sharp.  If it was possible to work out what controlled amplitude over time, then that graph would be very useful.

Example 2:  Plot (in a table) the first 8 partials for level at <spot time>

This is where things get tricky, because SNDAN will assume that if the fundamental is 440Hz, that frequencies *around* 440Hz are the same data set.  I presume this is so you can track frequency/time as per example1.  This however could be problematic, because at a particular timecode, the output level *could* be the sum of the actual harmonic *and* extraneous noises, especially in the mid-bass region.  I hope I'm making sense.  Put another way, organ pipe analysis in this way may show strong output of the 20th harmonic on a Rohrflute.  Chances are that there is no harmonic output as such, it is all stochastic edge tone (noise). Put this level of output as steady state harmonic in an additive synthesizer and it would sound ridiculous.

What you'd therefore have to consider is, what else other than the vibrating string could produce output *near* the harmonic you're analyzing, then ask whether inclusion of this in the test output would influence the results or not.  I'm pretty sure that a vibrating piano string does not have a perfectly steady pitch output.  Again though, that is the model working; we don't have direct control over any of this.

I suppose what I'm saying is, one would need to be very careful about the subset of data collected, what is done with it and what assumptions are made.

Regarding partials evolution over time, the main problem with this is that SNDAN can use different colours to show different traces on a graph, but it isn't consistent about which ones it uses so I generally plot one partial at a time.  I also don't find much use for a waterfall 3D plot of amplitude for the first 8 harmonics - looks nice, but what to do with it? 

What I would suggest is that if you are trying to understand how tools react, you should limit yourself to examining e.g. the fundamental.  Every time something is changed, record a wave file and analyze it.  I'm fortunate that my KX8 can be told to only output a specific velocity value, which aids testing.

I suppose a valid goal would be "what influences/causes/suppresses a "hump" in the attack profile of the fundamental?" which would require rigorous testing of single parameters then parameters in combination.  As to whether a hump (or not) is required to achieve a specific sound...that's what we'd be wanting to find out.

Even without changing any voicing parameters, the analysis I suggested (maybe for the first 8 partials) could show the effect of lid position, for instance.  I find with the standard version, closing or mostly closing the lid on both C3 and YC5 presets, gives a much more "plummy" sound than having it open, which was a surprise.  Yes I expected selective frequency damping, but the effect of the reflections is very pronounced.

Sorry for the length of the post - I'll try to remember how to use SNDAN over the weekend and post some sample plots.

Best//Neil

Have anybody here tried to show pianoteq to a piano tuner ???

I would like to hear from pianoturners whay they think about pianoteq pro.

With regard to Ecaroh's original concern, I've got Ivory, and before purchasing Pianoteq 3, that was my main piano software.

IMO, in one way, Ivory has a very pleasing tonal or spatial aspect to it- whether that's caused by the tuning, or the mic placement when it was recorded, or other things, I don't know. I think the Pianoteq M3 has similarities there.
In another respect Ivory leaves me very unsatisfied with the sound as a whole, and this I'm pretty sure has to do with it's lack of sympathetic resonance. Also, certain individual notes don't sustain well, so there's inconsistency, and to me it's distracting.

As far as the attack,  as some have said, the Ivory C7 does have a bite to it. It works well for ensemble playing, or in a mix. For solo piano I much prefer Pianoteq. Can you get the same attack characteristics from PT as with Ivory? I don't know. That would be useful, but personally, I'm more interested in warmth, depth and breadth of tone, such as that of a Steinway D.

My post wasn't about any fears over Pianoteq, Philippe and I'm greatly impressed by the quality and consistency of Modartt's product. 

I and another organ fanatic spent literally hundreds of hours alpha-testing software for creating organ samples through additive synthesis (and a mixture of other techniques bordering on physical modeling).  I know the amount of work which goes into a product such as PTQ and also how much work would result from Jake and Glenn's experimenting. 

It was my opinion at the time, and still is, that real-time physical modeling of organ tone would be very successful, especially for situations where matching to existing pipework or providing just a couple of additional ranks, is needed.  I've attempted flue and reed modeling with NI Reaktor and Harm Visser's toolkit but as Harm advised me himself, the models aren't organ specific, they're for general flute and reed instruments, not to mention that the models by their nature aren't or won't really work polyphonically.

Looking forward to Modartt's future developments!

[/HijackThread]

Best//Neil

Jake, the links are active again.

_

De rien.

I finally got my Ivory 1.7 and after hours of installation and few hours of testing I can share with you few of my experiences. Ivory is my first "professional quality" software based sample piano which I purchased as an option for PTQ for mainly recording purposes.

Ivory sounds quite nice and powerful in the place (middle C and below) which was to my experience the most difficult for PTQ. So if you are about to record pianoparts like "Let it Be" and "Angels" etc. from those two pianos I'd choose Ivory.

But if you'are recording more sophisticated piano stuff which uses the whole thing, question of Ivory or PTQ is much more complicated. I didn't expect to get those resonance features from Ivory, so lack of them wasn't any surprise to me. It depends on a player and his/her priorities how much he/she puts value on these qualities. Thanks to PTQ I have learned to listen those things much more closely and for me that's very important part. But as I said before, the most important thing is the basic sound and how well it fits to the musical needs.

If Ivory is quite powerful in the lower part of piano PTQ is beating it very clearly in higher register. In fact Ivory sounds terrible if you play high notes (and lack of resonance doesn't completely explain it - maybe it's badly sampled, I don't know). 

All 88 keys having up to 10 velocity layers (in Italian grand, which I have not, they have 12) you coud suppose to have a very expressive and responsive piano sound. No looping either. Compared to harware pianos this is supposed to be the main advantage. But this is maybe the one of the weaknesses too: It seems to be very difficult task to sample it evenly. I was expecting to hear this but it was a surprise how much Ivory has this unequal character. In other words for me there are too many individual keys that are sounding different than the keys near to them. When you hit hard you can even hear small amount of distortion in some plases. C7 is maybe the poorest sound in this respect and it was a great disappointment to me. Steinway seems to work better and in fact with some tweaking you can more corea-type or rock ballad-type piano out of it.

So I must conclude from this sample vs. modelling competion (right now Ivory vs. PTQ) that maybe the game is 7-6 for PTQ. As people here have said it, PTQ may not allways sound like a real acoustic grand but it IS like a real instrument.  But after all it depends on a task what are doing with it. I'm quite satisfied to have both. As we said earlier, it seems allmost impossible to get this corea-type sound out of PTQ which for me is a shame. And I just don't have over 5000€ to put to V-piano which is after all maybe the best in producing this corea-sound.

P.S: Many people have said that sample pianos sound marvellous when listening to demos but when you actually get it it's a disappointment. With V-piano it seems to be quite opposite: when listening to demos it doesn't sound very good at all to my ears. It's much more better when you sit in front of it and start to play it....

Yes, I have it. Most of it sounds really very good (for certain types of repertoire anyway). It has a couple of notes though which have a rather annoying 'thunk' or 'ploink' (or whatever you wanna call it) in their sound at higher velocities and that's a real shame, because it's very noticeable.
Also: it's got a very large dynamic range. Normally I would say this is a good thing, but it can make the Garritan Steinway pretty difficult to place among other instruments in a mix, unless you use a compressor or a limiter (but that's not always desirable either because it messes with the natural balance of the instrument).

But it's definitely a very nice sampled piano.

_

Piet, have you played the PMI libraries with Sympathetic Ressonance Scripts ?

WHat about Pro MEga 3, have you tried? 
Video demonstration:  p://www.generalmusic.us/Media/DRAKEmodelingtechnologyexplainedVHS.wmv

The Scripts for PMI library, to use with Kontakt 2, turn the libraries quite CPU hungry.

I noticed that the Garritan Steinway requirements are quite CPU Hungry too:

•2.8 Ghz CPU Pentium 4 or better for PC, 2.0 Ghz
Core 2 Duo MacIntel or better for Mac.
•2 GB RAM Recommended
•Professional Edition 24 bit - 67GB free hard drive space (45GB 16 bit)
•Basic Version - 3.5GB free hard drive space (download size 300MB).
•Hard drive speed of at least 7200 RPM

Remember the requirements for PMI Libraries if used with Sympathetic Ressonance Scipts.

Here some tests for Sympathetic Ressonance efects (elaborated by Niclas Fogwall):

Test 1:
Damper effect. Here I depress the sustain pedal, play a loud chord, release the chord but hold the pedal (the chord still rings under the sustain pedal).
Result: When holding the pedal, the piano sound becomes more spacious.

Test 2: As a continuation of test 1 (you will here listen to the same file) I here silently press the chord (while holding the pedal) and finally release the pedal while holding the notes of the chord with the fingers.
Result: The chord that is held still rings loudly in the end.

Test 3:
Here I play some loud chords and immediately after releasing the keys, heavily press and hold the sustain pedal.
Result: The chord is caught by the string resonance.

Test 4:
Here I play an up-down scale while holding the sustain pedal.
Result: Sympathetic string resonance.

Test 5:
Here I am laying down a chord in C major, without the hammers touching the strings, and pressing and releasing some keys to hear the effect.
Result: The string resonance reveals the chord that is still there.

What about try out those tests with Garritan Steinway and V-Piano?
Pianoteq emulate all them perfectly.

I agree.  My brother's East/West installation sounds more like a piano on single notes than PTQ2/3.0 did, but the playability just isn't there, for all the reasons Piet mentions.

I don't think any sample player can produce realistic SR.  It will also be computationally expensive as well as hard work for the HDD.  It seems like a sledgehammer to crack a walnut.  If one strikes e.g. middle C with the dampers off, then adds the higher G whilst C (and everything else) is still ringing, this will not only cause the G-resonant strings to vibrate but then all the sympathetically-vibrating strings will interact, producing secondary beats which may set off still more strings.  By the time you've worked out how to simulate that by triggering samples (if you even bother) you might as well use full-blown physical modeling.

As has already been said, the problem with samples is that's all you've got - a snapshot of a particular element at one point in time.  There's no interaction between any two disparate elements.  If you ever have the chance to sing in a reverberant space such as King's College Chapel, Cambridge, it can be immense fun to play with interaction of "single" notes in such an environment.  But, sample single notes of your voice and play them back through a PC - the effect is nowhere near the same.

Have you noticed the guys that the suposed great digital pianos manufactures always hire to play, talk positive things and promote their product ???
I never saw them before, or heard about...

I think we will know when areally good near perfect digital piano rise, when some real famous respeced classic pianist say that he choose that piano to a concert.

When Nelson Freire said that he really enjoy play a given digital piano, I will know that digital pianos finally reach a quality level so good as a real one.

Greg,

That link is to a Roger Hodgson performance (unless my browser's messed up)

Can anyone else provide Jake's request? It would be nice to hear the sound for comparison.

Thanks. Btw, I realise that in the video he never actually states that the V-Piano will be his preferred piano from now on, which would have a lot more weight than simply saying nice things about it. I guess this is the point that Beto-Music was making.

Greg.

Beto-Music,
Ok, thanks for the clarification. Yes, that would be most impressive if/when digital pianos reach that level of perfection!

Greg.

Neil, If a pianist was able to adjust a real Steinway Grand in a halph hour, to fit better his ability, changing velocity sensibility, hammer hardness, along keyboard, I'm sure he would do or request a time to do it.

I saw on web something like Nelson Freire comenting about a new Steinway-D that was bought to a theater Brazil. When asked about what would be play in the new piano, he said: 
"I hope he likes me."

Maybe conservative people fear someday a concertist get very bored, and just send a midi from home to the digital piano on the theater.
háaaa háaa...

I just watched that youtube vid, but more importantly, I listened to the podcasts Hugh linked there.

Is it just me or that Cunningham sounded really... crappy? Yamaha was really bright as expected, and Bösendorfer... ahhhh what a warm, mellow sound. Clearly the best out of the bunch

The Cunningham sounded ok to me, FWIW. Yes, the Bosey was nice and mellow, but still had a subtle metallic bite for the forte playing.

I really enjoy Hugh Sung's presentations - what a great speaker!

Greg.

katie,

Welcome to the forum, and oh boy, are you gonna get some arguments here

No argument from me - she's subject to the same preconceptions/misconceptions as everyone else.

Glenn

Edit - I'm not interested in debating about a technology that was developed by Edison (samples, recordings, whatever one wishes to call them).  I'm interested in how sound will be produced in the future.

That's why I'm here.

Agreed - I can hear detuning, but overall I thought it sounded good. Maybe a bit percussive, like an upright, though. (it's 5-foot something, yes? Does that make it a baby grand?)

Greg.