Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Who are we fooling - digitals are junk at this juncture

Of course, because that's what trolls do. It's the easiest thing in publishing just to pick a controversial subject or argument, tart it up with bit with bigotry or intolerance and then sit back and watch people fall for it.

Well, I think we all realize that acoustic pianos will not be beaten in realism department for the time being, simply because they are as real as it gets. Digitals are still sort of an approximation of their acoustic counterparts and I'd be the first to admit it.

With that said, however, I don't really believe that digitals are "junk" or "crap" as you say. I don't know, maybe you are used to the AP's so much that you can't settle for the "artifical" flavor of a digital anymore or you have had access to many acoustic grands and whenever you have sat at a digital piano, you mentally put it side-by-side to to all those AP's and see/feel the deficiencies of the former. That's probably fine by me, I guess. But the list of institutions buying Clavinova's that someone has posted in this thread says that the DP's can't be all bad. I assume that the decisions to buy these pianos were made by people who did know what they were doing and if the Clavinova's really were "crap", it would mean a lot of wasted money.

I can't really comment on the pricing with respect to the technologies used, though, since I lack that knowledge. I would agree that especially the high-end digitals are quite spendy (in my country, the top model of the Clavinova CLP series carries pretty much the same price tag as a new Petrof upright) but I don't really know if the technologies used in DP's sold nowadays are the best available or not. I've seen the Physis piano in the NAMM 2013 thread -- that one seemed to have some pretty nifty gimmicks in it but at this point in time I consider it maybe a little more than a proof-of-concept kind of thing.

+1

Once new model is released, repeatedly old tiny sample set is used, many short cuts are done and again we are disappointed. Even Clavia, which is quite forward looking company: if it releases 400MB piano or 100MB rhodes it would be a miracle. But at the same time we have cellphones with multi-core processors and many GB of memory.

My wife teaches private piano here at home and also has a church gig, so I hear a real, decent, maintained, brightish grand piano a couple of hours a day. Played through speakers, I can't say I've ever encountered a situation where a DP fooled me into thinking it was an AP. Looping, stretching, too few layers, and lack of resonance invariably add up to a dull, lifeless sound. The cheap speakers don't help either.


DPs are bought by all sorts of people for all sorts of reasons, some rational, some irrational. If you need the things only a DP can give then you obviously need a DP and should probably buy one. That doesn't mean they are anywhere near as good as they could/should be at this point, or that they are reasonably priced.

Yes, even at 400MB they're leaving ~95% of that gorgeous piano on the cutting room floor.

Rather like the movie industry showing you a trailer on an iPhone and claiming it's a feature length film in an IMAX.

the idea is the same as in light bulb conspiracy, it's a greedy business and there are no place for sentiments or drive for excellence.

I don't particularly see why this would be necessary. Two switches to determine the velocity of the hammer at release and one to determine whether the dampers are engaged or not. There *might* be an issue with DP implementation in that the two switches are not as close together as we would like so instead of giving us the release velocity they give us an average velocity. If this is the case I suppose it's due to the tolerances on the switches that are currently being used. That being the case I would prefer a better type of switch, but I don't see any reason to think that it should be continuous. Do you?

While I could see the possibility of a theoretically more complete modelling by using continuos velocity measurement (or position which could be equivalent) of key movement, but how much would it be better than the n sensor approximation - I cannot say for now either. I think it could be sufficient even with n=3. some better n=2 solutions yielded already playable expressive keyboards.

Yes. So that the final velocity is real, and not a rough approximation. And so note-off damping is a changing position, rather than a velocity. It's trivially easy to apply non-constant pressure to a key while playing (almost certainly the norm) and it's also easy to apply note off damping on a real piano in very complex ways. Two and three sensor actions might seem pretty OK (I'm kind of shocked they work as well as they do) and many can obviously easily adapt to them, but a real continuous sensor would give you tons of extra control over what actually happens, when, and how.

In many ways it would be nice to move past the bandwidth limitations of MIDI, which is how we got all this velocity nonsense in the first place. I mean, it's OK to a first order, but if you're going after realism I think you need more detailed data that key position can give you. And this isn't pie-in-the-sky stuff, it's all easily doable with 10 year old (or likely older) technology.

@dewster,

Have you thought of seriously sending your ideas (about the continuous sensor) to Roland, Yamaha, Kawai, etc., and see if they could make a prototype model?

Or, are you currently working on a keyboard that would implement it?

I would even consider buying one if there is someone that can do it.

As of now, there's nothing better than the V-Piano.

@Dewster, Pv88 - if only I had the time and resources to develop a DP -- I would build one with a dual bios, with the second one customizable, and lay waste to every DP manuf on earth. Imagine a Linux rom with a 128GB ssd behind it (300MB/s), and a 4 core x86 around 2GHz and 25W TDP for processing power, 8GB ram (30GB/s). About $350 of electronics would destroy anything offered today ...

Even with a "100%" sound engine we still need a speaker set-up that can communicate the sound as a real grand piano does if we are aiming for acoustic piano realism.

A DP is simpler than an AP, in some negative and some positive ways. Let me illustrate it in a way that should be intuitive to most by posting a picture of a "DP":

Torhu: Can your version accept full-length samples?
Does it have hammer action?
What about letoff?
MIDI in and out?

Infinite Response are already using continuous key position sensing, however they are necessarily outputting standard (hi-res) MIDI.

Quoting from: http://www.infiniteresponse.com/expr.html



Greg.

I really don't think the engineers at Yamaha/Roland/Kawai etc. haven't thought about this already but for some reason they stick to whatever we have today. I'm not sure I share dewster's "conspiratory" point of view that manufacturers purposely sell products with sub-par technologies but it's an interesting question nonetheless.

- just a minor side-issue on bandwidth, I don't mean to divert the OP but do you (or does anyone) know how fast midi data is pushed through a USB port into a PC? I know both MIDI and USB are serial ports but the latter has a theoretical max rate of 480kb/s (v2) while MIDI is a little over 31kb/s. I was never convinced, when using a sequencer for real-time input, that otherwise tight chord timings were getting accurately reported by MIDI whereas I've a bit more faith in USB doing a better job. Does anyone have any insights?

I'd rather we'd refrain from using marketing speech.

With digital systems there is no "infinite response" and no "continuous sensing" and I'd be pretty astonished if you'd really need a full blown DSP just to get a reading from a triple sensor setup.

I'd even argue that even if you'd measured the velocity distribution of the key presses from a couple of great pianists with sufficient accuracy you'd still get a distribution you could model with sufficient accuracy on a digital.

Even so, a digital sensor with just 8 bit of resolution still can discern 256 different levels of velocity, a setup with 16 bit resolution could even measure up to 65535 unique levels of velocity per sensor.

That's pretty much enough resolution to map out a pretty realistic velocity response curve and you wouldn't need a DSP for that (not at the low frequency even a fast pianist presses the keys which is < 20 Hz or < 20 keypresses per second)

You'd still only be able to use 127 levels over MIDI though, which probably would still be quite good if the sample sets in the pianos used all 127 layers and had enough samples for all of the velocity levels and notes.

So until a decent DP isn't even able to give you the full resolution of MIDI (127 levels of velocity) with a decent polyphony and samples which are long enough and numerous enough to play most chords without looping then it's pretty moot to talk about 'continuous response' or other marketing gibberish.

This by itself wouldn't even be much of a technical hurdle, take the innards of a Korg Krome (which is able to use the German Concert D sample set of the Kronos) and pair it with a great action and you'd probably blow most of what's available today out of the water in terms of fidelity with it's 3.8 GB of available memory and even the Krome only samples each key with 8 layers of velocity.

The Kronos X uses a 64 Gb solid state disc and 4 Gb of RAM, which would be more than enough to store one or two of the ivory 2 grand sample sets and use them.

I'd be pretty surprised if MIDI over USB used anything other than the 31.25 Kbaud/s rate. With that rate you could send a MIDI event every 32 Mikroseconds.

Nigeth: I agree, in a digital system, there is no such thing as completely "continuous", however the VAX77 does NOT use a triple-sensor action - each key has a hall-effect sensor, which produces an analog signal that varies in proportion to the key-press depth, and this signal is then digitised and processed by a DSP, as evidenced by the quote I supplied.

Greg.