Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Software piano's

A software pianos isn't all that complicated a beast. Basically a bunch of recordings that are triggered by various MIDI signals. Add a nice convolution reverb and you are pretty much done. That is basically what the first generation software pianos were.

But the first generation of software pianos earned a poor reputation because they weren't very "playable." In order to make the instrument respond in a somewhat realistic way you add 1) sympathetic resonance 2) partial pedal 3) repedalling. Once you get there things behave fairly realistically. To omit one or more of these is a grievous error in a current-generation piano. Their omission makes a big difference in how naturally the piano performs in my experience. There aren't really that many features like this, so they shouldn't be neglected.

I've oversimplified things here, but you get the idea. There's no point in adding other features if you miss the basic ones. That would be like adding leather, warmed seats to a car that doesn't have the ability to steer (or something else really fundamental). We don't see a lot of difference between pianos with 10 sample layers and 128. We don't see millions of different types of release samples making a difference. We don't even see layer blending making that much of a difference. I'm not sure what other features you may be thinking of.

At the end of the day the proof of the pudding is in the eating. Basic features I've class as "playability" items are important for a piano to respond naturally. More important than what's offered by other expensive but nonetheless first-generation pianos. The latter sometimes sound very good when processed or recorded, but often don't respond the way you expect them to while playing live. If you only care about sound, you can get a SW piano from 10 years ago and it will sound great. Sound quality and detail are not really areas of growth for software pianos.

With programs like Kontakt almost any yahoo with a mic can make a "software piano" but it won't behave like the ones discriminating digital piano players pay for. That's why lack of implementation of playability features is sufficient to disqualify a new software piano.

Yeah. I've read that thread before. I understand the issues involved. What I'm having a hard time understanding is the totally dismissive perspective that if this one feature is not supported then the library is no good.

Certainly there are people who use and rightfully desire a feature.

That said, a percentage of a point of view is based on hear-say. Some people repeat what appears to be common opinion. The same is true of a reaction.

People follow the crowd especially when going against it suggests you know less or are less skilled.

There is an element of implied knowledge and skill-level if you demand a certain feature in a product.........even when you do not use it........often or at all.

This parroting is common on internet forums where there is relatively little accountability.

Almost all sample based pianos are aimed to computer production musicians. Media composers, computer game composers, and the such. Those who are interested in getting the final result (the recording) out from their own studio...

In that sense half pedalling is hardly a problem for a lot. The sound, however, is. So having tons of velocity layers, and RR triggers and the such are much more important than having half pedalling options, which either way will get burried underneath enough reverb and orchestration.

Personally I find that the Garritan Steinway has a lovely sound (for classical stuff), but as far as playability is concerned (which is more the focus in PianoWorld) it's pianoteq hands down for me...

Thanks guys for corrections, I just quickly looked at DPBSD at could overlook something.

I don't use half pedalling in it's common understanding for 99,9% of my playing, BUT sometimes you may push damper pedal slightly too less - and in acoustic grand piano you have many levels of damper pedal. If a certain piano doesn't support half pedalling, the sound will be aborted. It will ignore any pushing pedal less deeper than let's say 50%. When you play fast piece with a lot of pedal changing and chords, half pedalling will forgive you some laziness or inaccuracy. But of course accepting pedal a few milliseconds after striking a chord is also very important in this case.

I think there are many cases when people use partial pedalling on acoustic piano, and don't even know it, because it is done automatically, connected with feeling of a moment.

EWQL doesn't have half pedal cause it's an older piano. Neither did Ivory I, NI Acoustiks, Grand 3, or the other pianos popular a few years ago.

I've only played piano for 8 months but I bought a Yamaha CLP340 new, so I started off with a good piano.

I can't believe that I am now actually considering the Vienna Imperial VST for $600+. It's either that one or the Ivory II American Concert D. Both of them are gorgeous even hearing them on the Try-Sound website.

They are the only one's I've been impressed with.

I also think the Vienna Imperial is one of the best sounding. The Ivory pianos occupy a different area of the universe. I have not been impressed with their tone. Have you ever seen this video? There are several that sound good enough to look into.

[video:youtube]Jj_9LHzvmFg[/video]

Don't make the mistake of picking a software piano for LIVE playing by listening to demos. I learned that the hard way after buying expensive software pianos. Most importantly, the demos tell you nothing about how the pianos will play. Try to find an acoustic piano that doesn't partial pedal (called half-pedal in this thread) or more importantly re-pedal. Of course they don't exist. So playing a software piano without those features will never feel like playing an acoustic piano. It will simply feel (and sound) unnatural to you while playing. That's why 2nd generation software pianos replaced 1st generation software pianos, not because of sound improvements in the samples themselves.

Demo's don't really tell you how the piano will sound (rather than feel) while playing either. Are you listening to the demo with the same speakers you will use to play the piano? Very likely you will listen to the demo over poor speakers attached to your PC. Or maybe over very good speakers in a high-end hi-fi system? Neither will sound much like the speakers typically used for playing live instruments.

Plus, (in my opinion), we listen very differently to a demo recording than we do while playing. The demo is simply recorded music (a final result) that we listen to and like or dislike as an audience member in a hall, or as we would sitting in our hi-fi room. Aside from our physical perspective (which matters - the piano doesn't sound the same sitting at the keyboard vs sitting in a room or an audience), we aren't involved in playing the music. While playing we hear the nuances in our playing as the piano responds to small changes in our touch. We receive feedback from the sound while playing and adjust our playing accordingly in real time. Playing the software piano allows you to hear how the piano timbre and dynamic range changes as you actually vary your attack and velocity on the keys, play repetitions, vary your pedaling, and so forth. You get none of that information listening to a demo. Plus inconsistencies between notes and non-monotonicity in volume or timbre will stick out while playing a software piano.

And the sound of demos doesn't tell you much because they can be doctored in so many ways. A sample set with major note-to-note inconsistencies in volume or timbre can easily be fixed after the fact in demos. Try that while playing in real time. For example, check out this note from Joe Felice, who did demos for EWQL :



I would say the 2nd worse way to pick a software piano is to listen to the same MIDI recording played by different software pianos. That simply does not work. A MIDI recording that was recorded with one software piano will not be optimum for another software piano. The first problem is that velocity curves for different software pianos are not the same. So what was optimum for the original piano may be not be optimum, or it may even be very poor, for another piano. Secondly, when we play we adjust our playing to the piano being played. I wouldn't play the same piece of music the same on the Vintage D as I would on the Ivory II American D for instance. So playing the same MIDI file on both can't be optimum for both. Finally, there is so much customization available on software pianos that an Ivory II (or Vintage D, or whatever) audio recording from a MIDI file is simply one of thousands (or millions) of very different audio recordings that could have been made from Ivory II (or Vintage D, or whatever) using the same MIDI file. So picking one tells you nothing about whether you would have preferred one of the thousands (or millions) of others you could have created using the same software piano.

So how do you pick the best software piano for LIVE playing? Listen to the people on this forum (or other forums that have people playing LIVE) that have used a lot of them. Listen to what they like and don't like about their playability, their sound, etc. Ignore people that aren't playing LIVE. They are interested in entirely different issues. You will find there is a general consensus about several pianos being "the best" for LIVE playing.

Listening to a recording made with a software piano plug-in can be a viable starting point along with recommendations or reading expressed comments and reviews. If you don't like how something sounds in a recording what is the point when you are essentially playing recorded notes of a piano when using software piano plug-ins? I havn't notice anyone suggesting people ought to base their decision on any one source of information. Unfortunately it will take some luck with your research to find something you like the first time you make a purchase. You will probably end up making an investment to chalk up as a learning experience before you are satisfied with something.

I don't understand the argument that the piano's sound while you are playing is more important than the sound of the recorded plug-in when it is what you will be working with if using software piano plug-ins. The idea seems to be that you must have the sound of playing at an acoustic piano. If you get a sound to match sitting at an acoustic piano chances are it will be alien to an audience and no one but the player will appreciate it. It might work if all you are looking for is a good playing experience and you are not interested in what anyone else hears. If there is any conversion to audience acceptable tones something is bound to be lost in the translation like working with a single MIDI file and multiple software piano plug-ins. If you are going to use software piano plug-ins you have to accept it is a different experience on multiple levels and adjust to it. It is impossible to recreate the experience of playing at an acoustic piano using an electronic device. It will never sound or feel the same with current technology...........who knows what could happen in the future?

Macy:
I agree with this ...
1. "Don't make the mistake of picking a software piano for LIVE playing by listening to demos."
A demo will not reveal the feel of a piano. A demo will not reveal the true sound, given the variability of auditioning equipment, especially the speakers.

... and this (at least in principle) ...
2. "And the sound of demos doesn't tell you much because they can be doctored in so many ways."
We seldom have a way to know whether any editing has been done.

You wrote this as an example of item 2, above. But it's really points out the inability to use a demo to judge playability. You quote Joe Felice ...
"I kept finding myself having to record my live performance into midi ... and then edit velocities to correct for notes that fired the wrong velocity-based sample set for a given note."'

This is also true ...
3. "A MIDI recorded with one piano will not be optimum for another. The velocity curves for different pianos are not the same. What was optimum for the original piano may be not be optimum, or it may even be very poor, for another."

Among all of your points, this is the heavy-hitter ...
4. "There is so much customization available on software pianos that an audio recording from a MIDI file is simply one of thousands (or millions) of very different audio recordings that could have been made using the same MIDI file. Picking one tells you nothing about whether you would have preferred one of the others you could have created using the same software piano."
But keep in mind that a Bosie or Steinway sample will still sound Bosie-ish/Steinway-ish after extensive configuration tweaks. Only grotesque and un-musical tweaks would mask their essential character ... and I this would be obvious in the demo. I would discard any such demo.

This makes sense ...
5. "How do you pick the best software piano for LIVE playing? Listen to the people on this
forum (or other forums that have people playing LIVE) that have used a lot of them. Listen
to what they like and don't like about their playability, their sound, etc. ... You will find
there is a general consensus about several pianos being "the best" for LIVE playing."
But listening to other people is no substitute for playing and listening to other pianos. Peoples' opinions come with unstated, hidden subtexts, difficult to discern in direct conversation and more difficult online.

Do you have any opinion about the "Piano Software Plug-in Sound Survey" thread. I'm guessing you don't approve? But I see great value in it. It highlights the enormous differences in the piano sounds that Ampy has offered there, enough to help me eliminate some poor-sounding pianos and choose a short-list of preferences. It helps narrow the field, which is valuable given the high-cost alternative: buying a series of pianos only to keep some and discard others. (We're still waiting for him to reveal the identities of the demos.)

I think Macy's last post is worth its weight in gold. Great advice there regarding sound alone (as in a demo) vs. playability (the response and feedback to one's own playing).

I also agree with MacMacMac on the point that no tweaking can mask the essential character of a piano sample library. At least I haven't been able to do it. I have however been able to substantially improve my playing experience (most notably of the Ivory II American D) by tweaking it to suit my preferences. So I did not tweak it to make it sound different than a Steinway D, but to make it sound like a Steinway D that fits my expectations in sound and, most of all, response.
In light of this, the demos of the American D certainly helped me to judge its worth compared to the previous Ivory II libraries which I already did own before.

Now I'd like to go even further on the point of inherent character:
Even the most advanced piano sample library is a recreation of a real acoustic piano. The better it is, the better it portrays the character and playability of the original (as well as its flaws!).
Even amongst real acoustics you will find pianos you like and pianos you don't like.
This ultimately means: You can buy the most detailed, perfectly crafted, super playable, true-to-original piano sample library that is recommended all over the internet by live players... but you still won't be all to happy with it if you happen not to like the acoustic piano that was sampled to create the library.

And from my experience - the better the libraries get, the more important this point gets.

Because you may not like the sound of the demo because of too much reverb, or not enough reverb, or too much sympathetic resonance, or not enough sympathetic resonance, or too much sustain resonance or not enough sustain resonance, or too much EQ or not enough EQ, or too much key release or not enough key release, or too much of dozens of other variable parameters or not enough of dozens of other variable parameters, --- were used to make the demo. i.e. You are not simply playing recorded notes, you are playing recorded samples that are then processed with hundreds/thousands of parameter variations/combinations that are included with the software piano program. So why reject the program because you didn't like one of those thousands of choices?

You simply don't have enough information about the range of possibilities from a few demos to make a choice between good programs. If a demo sounds totally hideous, then of course you will probably reject that piano. But in most cases the demos aren't that bad. My point is that making a piano choice between several decent sounding demos doesn't make sense even when one seems to be substantially better than others. I did that with my first purchases and there was no correlation between the demos and what sounded better in actual use after optimally configuring the pianos for my LIVE playing.


I said over and over that I was only talking about LIVE playing, and I was responding to Fscotte, who purchased a CLP340, which is not something that you carry around for gigging. And he has only played for about 8 months, so I don't believe he is using it for making professional recordings. I assumed that he was interested in playing his piano LIVE as a substitute for an acoustic piano, as I do. Hence, the objective is for the piano to sound, and feel, as much as possible the same as when sitting at and playing an acoustic piano in one's home. Unless one has 20 foot long arms, while playing you probably don't hear your piano as someone else would sitting across a room.

Nevertheless, within a reasonable approximation, if your piano speakers are located in close proximity to the physical location of the keyboard you are playing, and the sound at your seated position is a good replica to what you would hear seated at an acoustic piano in the same room, then the sound across the room will also be a reasonable approximation of what would be heard from an acoustic piano in the same room. Which is to say, within a reasonable approximation, the spatial convolution properties of the room will apply to both an acoustic piano and your software piano speakers. Of course, it is only approximate because the dispersive frequency response directionality of an acoustic piano sound emissions and the speakers are not the same. Yet with some care in speaker selection and positioning, one can satisfy their own selfish desires to optimize the sonic emulation seated at the keyboard without driving your "audience" out of the room complaining that your software piano sounds more like a Rhodes than a Steinway. (In my own case, the speakers are located in a grand piano like-case which may assist in this acoustical charade, or it may just be a placebo effect). Nevertheless, the bottom line is that I'm the one seated at the piano playing for hours on end, so I get priority since I don't sell tickets to sit on my couch to listen. And if you must imagine yourself on a stage in a much bigger hall, we can always dial in a little additional convolution magic without placing yourself in the audience.



Well, whether it is impossible or not depends on the criteria you use to judge what is close enough vs whatever tradeoffs you perceive to have with an acoustic piano. What is close enough for me, and what is close enough for a concert pianist are obviously two very different things. I will spare you the stories about the acoustic grands I didn't buy because I decided to maintain my current setup for now (and price wasn't a significant decision factor).

And I agree with everything you agree with me about. (OK that is suppose to be a joke.)


I think they are great fun. I've participated in them in the past and even supplied some recordings, etc. But, IMO, they are no way to pick between one quality piano vs another quality piano for the reasons I stated. But I don't think that was the purpose of that thread (didn't it say something like that?). The only reason I haven't participated in that one was because I haven't had time. I'm writing this at nearly 5 am and I haven't been to bed yet tonight. Us retired people keep very busy working when other people plead that we just do one more little thing for them ...

A lot of good points, Macy, A few I'd like to comment on...


It is true that the playback system for live performance will affect the sound of the piano, generally in a negative way. However, a playback system will tend to effect all sounds equally. That is, if the playback system has an unfortunate peak or dip at a certain frequency, that will manifest itself regardless of which piano source you use... and unless it happens to be at a frequency near where the source piano itself has some unnatural response, it is not likely to be significantly worse on one piano than another. Also, the playback system itself will sound different in different rooms, and in many cases, you may even be playing through different systems in different rooms (i.e house systems, or system provided by different sound companies). So I would say that you're still better off finding out what piano sound you like best with a good set of headphones, or whatever i.e. start with the best source you can, and proceed on the basis that whatever you select will degrade roughly equally through whatever you end up playing through. (Except that you do want to take care that whatever you choose will sound good in mono.)



I agree with your premise quoted above, but it's interesting that it, in a sense, almost contradicts that quote from your next section (which you quoted to illustrate a different point, that demos can be doctored). That is, I agree, you would tend to alter your playing based on audible feedback as you play, yet it sounds like, at least with that software in that example, the player was unable to sufficiently do so to create a natural result, that he still found himself needing to edit the results despite getting feedback as he played! This also points out an issue with velocity layers... the more there are (and especially if they are not blended in some way), the more points there are of potential abrupt transition, so it can be a trade-off.



I agree about those two problems, and would add that, to a large extent, they may be the same problem. That said, I don't think it is useless to compare pianos based on hearing the same MIDI file played through them, either. I would agree that it is not a good way to evaluate their dynamic response, as you probably would alter your touch as you played if you were hearing the actual piano sound as you were playing. But for the overall tone of the pianos, the decay envelopes, the overall realism (including the resonance effects and such), I think you can still tell a good deal even when hearing a sample generated while the player was listening to some other piano sound. Or put differently, a real piano, played badly, still sounds like a real piano... and so I think the fact that you may be hearing a sample where the player was unable to alter his touch in response to the sound would tend to result more in something that sounds like a poor performance rather than a poor piano. That's still doesn't give you enough information to know for sure that a piano will sound and play well to you when you get your fingers on it, but I think it can help rule out ones where you can be pretty sure you won't be happy with their sound, as long as you keep in mind that you should not give too much credence to things that sound like unnatural velocity response.

(Though as you also point out, in those cases where a piano has many possible adjustments, you might not want to unfairly rule something out, either... you may just not like the settings in use when the recording was done.)

As opposed to what? You can't get much use out of piano software unless you hit notes on a keyboard. As you have also stated yourself, the samples are recorded notes of a piano. They are not going to sound like an acoustic piano. They are going to sound like a recording of an acoustic piano. Having the software playing into speakers connected directly to your digital piano or computer isn't going to change that.

Hem... You can certainly trigger the notes through midi software, so you don't actually need a keyboard to "play" a VST piano.

And of course if you think about it, what you're listening to your mp3 player, or to youtube, or to anywhere IS a recording and not an actual acoustic piano, right?

A virtual keyboard is still no different in this regard than using a keyboard to trigger something in the moment. One is physical and the other is an illusion triggered by a computer mouse/keyboard/pad/dial/slider/whatever.

Not sure what you are differentiating here.

No need to get caught up in semantics. It's perfectly true that tons of VST use is done by people who aren't primarily pianists and are doing all sorts of things to the MIDI besides playing it live (post-processing it, tweaking the MIDI values, mixing it in with other instruments, etc.) and are less interested in whether it responds naturally in real time than whether (with the use of tools besides a keyboard) they can make it sound a particular way. That's what Macy and Nikolas are talking about, I believe.

Just as you can I can't tell by listening to a demo whether the piano responds nicely, neither can that group's audience. Therefore they don't care about some of those types of details.