Sampled Pianos NOT up to par for playability AND sound in one package

[ohernie]

You should probably state which controller you are using.

I think the problem is more with your pocketbook than the actual piano softsynths. You are getting what you are paying for, which isn't much. Furthermore, your selections are, for the most part, a generation behind.

Before you write off softsynth pianos I suggest you 1) do some research on this forum 2) loosen up the purse strings and buy a modern one.

Ernie

[Glenn NK]

I am still looking for the right piano sample for me so I can play live and do recordings at home and actually for a change, be happy with it. As well as not throw my money away.

Thanks

A few rambling thoughts on this matter - hopefully I won't get shot at too much by sample afficionados.

Sample technology is based on recording a sound with a microphone, storing a physical or electronic model of the sound, and replaying the model back by means that are physical/electrical (vinyl disks), or purely electronic (tapes, etc).

In essence, recording a piano with hundreds of samples, saving them, and re-working them to be usable to reproduce the sound with the aid of a computer is the same technology; we've added the computer.

This technology is quite old now - Edison was one of the pioneers of this technology (phonograph and gramaphone).

The technology has come a long way, but it will always have its limitations and weaknesses as follows - note that any one of them will affect the sound:

1. What and how many microphones should one use?

2. Where should the microphones be placed?

3. Should the room be large or small or acoustically dead?

4. How many samples are enough?

5. Do the recording technicians consistently get the hammer velocities correct in order to accurately achieve the tonal variations with hammer velocity?

6. Should the dampers be up or down - or partly up?

7. Is the piano in "perfect" tune? What is perfect?

8. Is the piano voiced "perfectly"? Perfection varies from person to person and is extremely subjective, thus there is no one definition of "perfection".

9. Is the piano regulated properly?

I'm sure others can point other elements that affect the result.

If any of these aren't done "right" what is the result?

Probably the most difficult things of an acoustic piano to emulate is Sympathetic Resonance. It is also likely the most striking difference between a digital piano and an acoustic piano.

With the dampers up, striking a tenth in the bass (C and E for example), will generate as set of overtones that will be reflected in many other strings in the piano. When one strikes a fifth (C and G), a different overtones (sympathetic vibrations) are generated. The important thing to note is that using the same notes results in the same overtones - in other words, they are repeatable and predictable.

Extrapolating this thought process to the complexity of actually playing the piano, one soon realizes that to record all the possible combinations of overtones that result from literally millions of possible note combinations, rapidly approaches infinity.

So to truly emulate the Sympathetic Resonance of an acoustic piano would seem to be well beyond the capability of recording samples - unless they are willing to record an infinite number of samples. I think not.

What is the answer?

There was a study done by Scientific American magazine circa 1965, and it was found that if a piano tone was electronically generated from the fundamental and a minimum of twelve partials, piano experts could not discern the tone from the recording of a real piano. This was possible 43 years ago.

Since then, mathematicians and musicians have learned that even the attack and decay portions of the sound can be modelled accurately (including the hammer noise).

Although complex, for any given single note struck with the dampers up, it is possible to accurately predict all overtones that will result, and their relative strengths.

Why not generate these mathematically? After all isn't that what computers do best?

I will make a bold prediction - in five years or so, we will look back at sample technology much as we look back at cassette tapes or wax cylinders today. An interesting and temporary state of musical technology.

Have a nice day.

Glenn

[kensuguro]

well, you may have heard of pianoteq, and several other up and coming hardware units, such as the roland vpiano. There's another one called pianoid which is promissing, but is very much in development.

The comparison of samples to phonographs is quite extreme, sampling technology has come a long ways in realism and usability, so it's not really fair to dismiss it as yestertechnology. The technology you speak of from 1965 is most likely additive resynthesis from fft analysis data, which was all the rave with the Fairlight systems.. A form of synthesis still widely used, but not for piano modeling.

More modern piano modeling comes in a form of a bunch of differential equations, each equation tailored to simulate a specific component in the signal chain. It takes an incredible amount of computing power to render this in real time, and unfortunately, CPUs that we use for aren't optimized for that. Hence, the DSP and outboard hardware route taken by roland and pianoid.

Physical modeling is a double edged sword. Theoretically, it is so easy to idealize physmod and say that it rules most supreme, because it is mathematically and scientifically more correct than a sampled piano. It is easy to say that because it has continuous velocity, that it will be more playable than a sampled piano.

But that doesn't mean the model produces a particularly attractive result. Physmod strings, brass, and percussion have been under research for quite a long time now, and yet very little people use it for production. Film scoring still turns to VSL an EWQL orchestra. It's still very debatable whether pianoteq sounds aesthetically pleasing or not. (personally, I think not)

It's because many times, a specific "sound" carries with it more than just a replication of an instrument. It is the entire production line. The performer's technique, the engineer's micing technique, the mixer's aesthetic, the ambiance of the recording location... So while it may be thought that physmod instruments are accurate representations of the instrument, it's still very much "locked" in terms of all these other elements. Actually, even worse.. physmod instruments don't even possess all these other elements, and therefor sound artificial and out of this world.

That's like getting Micheal Jackson as a kid, and not getting any of the brothers, no engineer, none of the writers from motown, nothing. Chances are, you're not going to get the signature Jackson 5 sound that way.

An acoustic instrument recorded in the same way as a physmod instrument (meaning, sound is recorded without going through a mic, mixer, but somehow recorded directly to disk) perhaps the 2 may sound similar, but since we have never heard of anything like that, we cannot associate very well with the sound.

But regardless, physmod is the way of the future. And it's becoming more and more a reality every day. I'm very much looking forward to roland's rendition, and yamaha's, if they have one in the works. Particularly Yamaha since they can put their acoustic piano building experience into good use. Pianoid is interesting because the hardware processor they're using is insanely powerful, but I'm going to keep my fingers crossed. (hope it makes it to production)

[Ashermusic]

IMHO, when people decide to start seeing all these virtual instruments and sample libraries as tools in their own right and dealing with them as such, and stop hoping for/expecting them to be identical to real instruments, they will make better music.

[SCARBEE]

Eventually it is all about the sound and playability - then it doesn't matter HOW it is achieved.

It's like making special FX in movies - you still use dolls and robots, real locations and real actors though you have computers that can do it all (not all great though).

it depends on the project and the situation. In Rolands case it makes sense to use physical modelling as it is probably easier to deal with that loading a 2 GB piano.

But for Computers it is not a big issue as we get bigger hard drives and more ram.

But one thing that is totally different is the production - sampling pianos is a complicated and time consuming process - whereas physical modelling is more of a science project including all kinds of patens etc. Not easy to get into for the average developer.

I always use my ears and see if I get touched by a sound - if so I don't care how it was made...

[jcfelice88keys]

Hello Maestro2Be,

You made a very valid point in your opening message that started this thread. Many pianos that sound well do not "play" well. You went on to name a list of products you have tried.

One product not mentioned was the East-West Quantum Leap Pianos Boesendorfer. I am Joseph Felice, the person who submitted a live performance of Gershwin's Rhapsody in Blue, and Beethoven's Hammerklaver Fugue. If you wish to hear ppp through ffff and every gradation in between, I strongly suggest you go to the following site, and listen to my performance -- first -- of the Gershwin Rhapsody in Blue. You are free to make your own decision how you think the dynamics of this particular instrument are articulated. And remember, this is is a live performance, played on an ancient Roland A80 keyboard, using Digital Performer as a pseudo tape recorder to capture the performance. The reverberation you hear is what comes straight out of the box.


I had to encode the URL as follows because it would not be legible at this piano forum site. Following h t t p : [/ /]

w w w [dot] s o u n d s o n l i n e [dot] com followed by a backslash,


then type in after the backslash: Quantum-Leap-Pianos-pr-EW-171.html


Cheers,

Joe <jcfelice88keys>

[tmhuud]

+1 for quantum Pianos. Nothing quite like them out there IMHO.

[jimmymio]

TSO has a big brother (The Big One) but I find myself going back to TSO very often as I find the programming of the velocity layers in it more biased towards the softer velocity layers which I like a lot.

I have several Sampletekk pianos but i was unaware of TSO. If you like softer pianos there is an "Alicia Keys" piano coming out soon. I believe it was sampled by Scarbee. Sounds really good.
JP

[SCARBEE]

If you like softer pianos there is an "Alicia Keys" piano coming out soon. I believe it was sampled by Scarbee. Sounds really good.
JP

Thanks. Demos on www.scarbee.com are played quite softly with pedal down.

Transformation (dry)
Transformation (wet)
Flying (dry)
Flying (wet)
Baroque_inspired_melody (dry)
Baroque_inspired_melody (wet)
Tomheten (dry)
Tomheten (wet)

Songs by Nils Liberg

Here are some more demos - in funk, jazz and pop style...

these are originally made for my electric keyboards but shows a new side of the piano. Note that there are several small pieces within each mp3!

All dry off course..

Songs are by Christian Vinten and sevan Gökoglu

www.scarbee-downloads.com/demos/ak/Christian_Vinte n_Jam.mp3

www.scarbee-downloads.com/demos/ak/Christian_Vinte n_Jam_2.mp3

www.scarbee-downloads.com/demos/ak/Sevan_Gokoglu_J am.mp3

enjoy

[esperlad]

I'll throw in my opinion: When I was writing a cue for a film, there was an actor pretending to play an old piano. I used a couple of different pianos, including VSL. I decided to try the Bardstown Bosendorfer to see how that sounded. I was very pleased with the results. I thought they were so good that I strongly suggested that those takes replace what was submitted earlier.

I still use that piano. Lately, I have used the Kontakt format because GVI is less than stable.