Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Hi-End DP's sound engines vs. Hi-End software pianos?

Hi,

I there are lots of very expensive DP's on the market recently (such as Avant Grands or new Rolands) so I'm really interested how are their sound generating capabilities in comparison with the top class software piano solutions such as VSL, Ravenscroft, Synthogy, Pianoteq? Better if sampled vs. sampled, modelled vs. modelled (I actually prefer software modelled ones) And, say, headphones mode vs. headphones mode, i.e. something bypassing that built in multispeaker systems. And, obviously, since I'm asking means that I can't check them by myself. Audio demos are also not that much helpful because the pianist is always the main component. So one's personal experience is the most valuable.

I mean it was always unquestionable when speaking about, say, Privias, so what about that top level models? V-pianos BTW?

Check out the "The DPBSD Project" thread. That's exactly it.

Actually I dearly hate the idea of checking the piano audio engine by copy-pasting some MIDI file - it's a so wrong thing to do.

The digital piano vendors sell a "package". I would consider this package to have about the following components/features:

- cabinet / looks. People have different standards for what is acceptable in the living room / music room / "man cave"
- action. A lot is being written about actions. Not enough is being measured in that respect, which I find bizarre (see my post in the DPBSD threat).
- sound system (as in speaker, set up, amplifiers, soundboard and whatnot)
- sound engine (nobody does just samples anymore. But most still have an approach that is sample based, with looping of samples because of limited memory)
- interfaces (some have USB host/device, line in/out, bluetooth, etc.)

and that delivers a instant-on piano experience.

When you sell a package, you can focus on the total experience and therefore have less to focus on the quality of the individual component. (Can, not must).

Maybe a bad example/analogy:
Digital camera vendors choose to compete on megapixels at some point in time. Boy was that a bad idea for photography. At around 6 megapixels (up for debate) this becomes irrelevant for the picture-taking of a "normal person" and noise increases and picture quality gets worse.

Good camera reviews do photos of standard cards (you know, with colors, circles, rectangles, text etc.) with standard studio light etc. It is one of the tests and it help. DPBSD is like that. One important component of a test.

I always marvel at the reviews e.g. dpreview.com does. Fantastic. I wish there would be the same for (digital) pianos.

But there is not, maybe because the market is too small. Maybe because too much in a piano is linked to individual preferences in sound, touch, visuals.

DP and software pianos are products of audio engineering after all. I see it as brilliant to come up with a MIDI file that reveals the characteristics of these products and compare them scientifically. Personal experiences are great to discuss and share. However, if you want a comparison, I suppose you will want something objective and not just opinionated.

Anyway, here's my own opinion: I have a HP506 and, in headphone mode, it's the best Roland had until just recently. However, I still prefer Galaxy II and Galaxy Vintage D better for their details in resonances and nuances.

Edit: ... or may be I just like more authenticity of sounds. When I want Steinway sounds, I want Steinway sounds and not a "Roland" if this makes sense.

That's it, thank you!

...imagine a car being tested in some service center - you can see the specs, the engine power, etc. But to actually evaluate it you need to be able to test it on the road by yourself. Same with a MIDI file - you don't know at which loudness conditions it was recorded/played, what dynamics settings was used while the piece was played originally - this and many other factors affect the performance and hence overall impression. The performer skills are also extremely important. It may sound quite well on the original setup but with any other audio engine it will sound totally different. Also the decay times, reverberation, pedalisation - all this factors are interdependent and can't be abruptly ported.

Well, I find it useful to play on the dp, record it on sms and play it back through pianoteq. I could play it thro pt at rhe outset, but it needs wires and things to do it with.. .but in any case the original dp sound is good to go. Imo of course. But one likes a change now and again. . .:)

FWIW, I know of one very competent pianist/teacher too, heard him play all sorts of stuff, from Rach to Beethoven. He has a youtube channel, but his name eludes me at the moment. Anyway he likes modelled pianos most, has a V piano himself, however he still prefers to use pianoteq.

What his opinion on the latest Roland models with the modelled engines would be, as found in the HP6xx and LX7/Lx17 I don't know. I recall him saying this well before these new models came out, but pianoteq 5.x is what he based that on.

Of course that is just one opinion from one person, make of that as you will, but he sure could play at a very high level from what I heard

Personally (as a beginner I should add) I like the change of weather occasionally, I have time for both a sampled library and modelled sounds as well. There are days where the synthetic/artificial nature of pianoteq still gets to me though, as expressive and resonant it can be made to sound, I also need my dosage of ivory to gimme that Steinway sound or something more authentic sounding, depends on the mood or what sound I want to get on the day

Personally I prefer these software options over the build in Casio sample for sure in the AP 450 I own.

I heard some recordings of CLP 585, even played that model in a shop once for a bit no too long ago, and heard demos of the Kawais CA series, they sound good too, however still not at software levels ... IMHO, for no technical reasons or whatever number of velocity layers they may have when sampled, or how long the samples may be to avoid looping, I just say that based on what pleases my ear, as simple as that

Ewql can be considered a premium Vst piano?? There is some "Bechstein" Vst piano with the quality of others like Ivory ??

It might be Hugh Sung (but I don't think he has a V-piano) -- one example:

http://www.forum-pianoteq.com/viewtopic.php?id=3043

One comment states that it was done with Pianoteq version 4, "D4" piano.

And Ludovico Einaudi (Pianoteq version 4):

https://www.youtube.com/watch?v=oo0EYA2jAEg

And many more:

https://www.pianoteq.com/references

End of Pianoteq marketing material.<g>

Charles, none of the above, I remember now, new day, fresh brain, and as soon as I saw your post, his name came to me immediately. Phil Best.

https://www.youtube.com/user/philbestmusic

This is a great concept with a noble intention but the execution in real life can be problematic.

With hardware keyboards, the actions are rarely/never just sending raw MIDI output to the keyboard's digital brains. There's usually a velocity map in the software that re-maps the MIDI output of the action to smooth out the response and make the whole package a bit more playable. When I was at Kurzweil we spent weeks and months hand-tweaking velocity maps for playability (in addition to receiving help from smart software tools).

There are a number of variables at play regarding the velocity maps - the velocity response of the synth engine (filters, EQ, effects, and just plain old amplitude), the velocity switch points of the samples (or the response of a synthesized model), etc. And different actions require drastically different maps.

So between all of these variables, the generic MIDI file poses some problems. If the file was created on a certain action but played back on an instrument with a drastically mis-matched velocity map, you could end up sending MIDI values to the engine that the machine is not normally used to receiving. Passages could sound softer/louder than intended, or there could be abrupt jumps from in volume.

Not the end of the world, but also not ideal in some cases.

Today I attended a really good semi professional singers group in a theatre with a 400+ audience.
There was a small group of professional musicians.......a cello, flute, guitar, drums and three keyboard players.
The main piano sound was coming from a Roland RD800.
I had an ideal position...third row ,slightly off from the middle.
There was no mistake at all that this was a digital piano sound....I am a concert piano tuner, so that was very easy indeed to detect, because I occasionally stay to listen to top grands in top theatres during rehearsal after i've tuned the grand prior to the final soundcheck and they want me to check upon the tuning after that.

What surprised me the most is that it sounded more or less okayish, but nothing really was worth mentioning in favour of the RD800.
No real sparkling or defenition, a little bit muddy and above all an arteficcially ring that made it so very clear that it was nothing more than a digital to begin with.

The recent recordings of my band show a different picture, I use the Integra for synth and other sounds and know what the Roland piano's sound like (althaugh the RD800 is better than the Integra when it comes to the main piano sounds).
I use a Nord (Silver Grand) for all my piano sounds and I must say it sounds freaking awesome on every recording we made.
Yes it is a digital, but the pure sound is alive and warm.
At home I use Garritan CFX and it is great, but since it is run from a big and extremely fast computer connected to my Kawai MP8 it is not an option to bring it with me on stage.

I know this is a little off topic, but after today my admiration for Nord (Silver Grand that is....)and how it sounds op front in the audience and part of a mix is appreciated once more.
I'll take the subpar keybed, the limited amount of memory and everything else.....but boy does it sound good up front.

Totally agree. There are many measurable criteria in original image / object to errors of the captured image. But merely little in sound processing. The closest things in audio are frequency response / crosstalk / audio noise measurement etc which is still talking about inside the loopback line signal. Once getting out to the air, all become subjective.

Some DP cannot replace the sound samples.
All DP cannot replace the internal speakers.
All DP cannot replace the keyboard.

To say which DP is better than that is simply unmeasurable. In other word, this is art!

I have recently tried AG N3. This is the closest thing to a real grand in my opinion. There is a guy here who quite often posts some very technically demanding classical piano pieces he has recorded by using AG as a MIDI controller to some of the high end software piano libraries and the results are excellent, probably better than what you can achieve with the AG's internal sound engine when recorded through the line-out's, however the guy admitted that with most of the software libraries there was no repedaling and half-pedaling at all. And it is a well-known fact that even those that support it, are not doing it very good (Pianoteq being the only exception since it is modeled but its timbre arguably sounds artificial and is not on par with the sample-based pianos). So, he's doing it just for YouTube where you, as a listener, can't feel if half-pedaling is missing and the pianist is forced to compensate for the lack of repedaling. So, for audio recordings, the sample-based libraries are probably better. For pure piano experience from player's perspective I believe an integrated approach with real hammer action such as the AG with well-crafted multi-speaker system, tactile resonances, proper half-pedaling, repedaling, etc. can't be beaten.

You will have to explain to me what's wrong with the partial pedaling and repedaling in the Vintage D, and Ivory II American D. Works just fine for me. And those aren't the only sampled pianos that have it, although they are the only ones I play much anymore so I don't recall any specific pedaling pro's/con's with others that have it.

Vintage D:
1. press the damper fully, play a key and release it, then release slightly the pedal, say 50%. The note sustain should become faster. This didn't work when I used Vintage D few years ago.
2. Press the damper fully, play a key, release the key. Then very quickly release-press-release-press the pedal. The repedaling works only the first time. BTW I think that it doesn't work even the first time, can't remember exactly.

I guess newer versions of Vintage D might be fixed but I doubt it. Can you confirm? Regarding Iviry, I remember someone complained about same issues but I haven't played it myself so you can confirm.

And my description of point 1 above is oversimplified. A more realistic scenario is to start with the fully pressed damper pedal and then release it slowly and smoothly observing how dampers engage progressively by muting the sound more and more and decreasing the sustain time more and more.

I guess for some styles of music such as new age, pop, jazz, etc that's not a big deal but for classical romantic repertoire the pedal work is equally important to everything else, it's an art of itself. An example is worth thousand words, so just listens to the first piece this pianist plays in the following clip, Chopin's Nocturne op.62, No.1. Note how he never engages or disengages the damper abruptly. Instead he uses a lot of slow release to connect phrases where the final effect is of a smooth legato rather than just too much damper. Listen to the consecutive trills he starts after 5:50 where he gradually releases the dampers throughout each trill in order to prevent notes to cumulate and at the same time to flow and connect trill after trill.

[video:youtube]_LGTmn4mQMk[/video]

Ironically digital pianos with looped samples are closer to that real continuous pedal behavior since they vary the sustain artificially by decreasing the repeated loop while full samples are really hard to be used for half pedaling. What I've noticed on e.g. Vintage D is that it detects the half-pedal value at the start of each note and applies the corresponding degree of sustain to it regardless of consequent pedal position.

Works fine for me, but I've only been playing for a bit more than 50 years, so maybe I don't know how it should work.

And yes, there was a major mod to the Vintage D pedaling a few years ago.

It's great they have fixed it on the Vintage D then. Can you record a demo on how you slowly release the pedal and the sound get progressively muted? I'd be interested in how it has been finally implemented.