I've been shopping for a new digital piano, and trying to
find out where and how far the technology has advanced since
I bought my Kurzweil Mark 5 three or four years ago. I found
as much information as possible on the web (some - from the
manufacturers, some from consumers in newsgroups) and am now
trying to locate stores that have some of the models I find
promising, to actually try them out. So far, I've been
interested in the Yamaha CLP-880, the Kawai CA950, and the
Baldwin Pianovelle series. Each brings a different approach
to the fundmental problem of emulating the acoustic piano
through the fakery of solid-state electronics.
Approaches to Digital Piano Sound Generation
The sound of any acoustic instrument is hard to emulate
electronically, and the piano's sound is absurdly difficult.
All digital pianos that I know of use sampled acoustic piano
sounds at some point in their sound generation. Sensors on
the keyboard detect the velocity of the key (how hard the
key "touches bottom") and adjusts the loudness and other
acoustic characteristics of the note accordingly.
Of course, there are other important sounds in a real piano
besides individual strings vibrating. There's the sound of
the hammers striking the keys. And there's the soundboard
and other strings, which vibrate sympathetically as you
play. This is maybe the most complex acoustic component of
the piano sound, and it's the most daunting obstacle in
emulating it.
There seem to be two main approaches to sound generation in
digital pianos today. Recognizing that a string's tone
varies when you play loud or soft, each approach attempts
modify the tone of the original sample in a different way.
Until recently, the standard approach was to generate the
closest sampled sound available, but filter it on the way
out, brightening or dulling the tone. High-end models
contain more samples to choose from, for example multiple
samples of the same key at different loudnesses. Presumably,
these samples will capture some of the ambient acoustic
variations for a more realistic sound. The better models
also let more notes sound simultaneously when you use the
sustain pedal.
There are a lot of limitations to this approach. To begin
with, there are an infinite variety of loudnesses and
touches, and it's not practical to store and use more than a
few per key. Variations in attack and decay between the tiny
set of sampled sounds must are simply not there. To my ear
anyway, covering for these missing samples by filtering the
nearest available one seems to degrade the original sampled
sound. Since nearly every keystroke requires this kind of
manipulation, the overall sound is seldom as good as the
samples. Filtering is after all a subtractive process - it
removes elements of the sound. Maybe for the same reason,
filtering also seems to be totally inadequate for handling
the more complex acoustical overlays of real pianos, like
sympathic resonance.
To tackle the full range of acoustic effects more
effectively a new approach to sound generation is now
emerging: using synthesized sounds instead of samples.
Typically samples are used in development, but not as the
direct source during playing. Instead of filtering a sample
source, the digital piano applies a set of algorithms to
generate a sound from scratch. Of course, the results are
only as good as the algorithms' ability to simulate
acoustical effects, which must be devilishly complex to
model. But in theory at least, this approach is capable of
taking any acoustical effect into account. The first
generation of these pianos is now appearing. I imagine it
will take some time for them to live up to their potential,
but I am hopeful that they will.
For a much better technical overview of this and other
issues around digital pianos, I found the following web page
very informative, if it's still available:
I've recently tried several representative digital pianos.
See my very subjective reviews below:
* Yamaha CLP-880
* GeneralMusic RP10
* Kawai CA890
* Conclusion
Digital vs. Acoustic Pianos
Ever since digital pianos became a marketable commodity, a
furious debate has raged among musicians over digital versus
acoustic pianos. I'm a firm agnostic. Well, no - if I had
the space and money, I'd get a concert grand. Since I
haven't enough of either, I'd like the best-sounding, most
responsive instrument I can afford and house. I have no
illusions that my Kurzweil is a great piano (nor that I'm a
very good player). I simply maintain that it sounded and
played better than any dinky spinet or worn-out upright I
could have found for the same money.
What's more (I say straight-faced to the piano purists),
there is nothing inherently superior about acoustic
instruments: they just happen to be at a higher state of
development. That's because they've been around longer and
have been lovingly refined by thousands of devotees. The
piano, more or less perfected over a hundred years ago, is a
marvel of engineering and craft. It represents an intimate
connection between machine and our deepest thoughts and
emotions that no modern invention can rival. But let's not
forget, it is a machine, and so are digital pianos, whose
developmental obstacles can be overcome by technology and
ingenuity just as the pianoforte's were in the eighteenth
and nineteenth centuries. The only question should be how to
make an instrument as exquisitely responsive to the human
hand and ear, with a sound as rich and beautiful.
Aside: Just maybe, the question should not be how to make
digital pianos sound exactly like pianos, just because
because it's what we're used to. The future of digital
pianos may instead be a different instrument, responsive and
beautiful in a different way than the acoustic piano. It all
depends on what people do with it. I confess to my own
hypocrisy here: today I want the best possible approximation
of a good acoustic piano. But in principle I think that
music lovers should keep an open mind about the long-term
development of the digital piano.
Yamaha CLP-880: Review
Last Wednesday I tried the Yamaha CLP-880. The store had
some badly-regulated and untuned Yamaha baby grands, but
still it was handy to compare the CLP to the acoustic piano.
I found the keyboard to be excellent, but the whole effect -
responsivess and tone - kind of dull and lifeless unless I
cranked the volume up all the way. Even at full volume,
though, the sound lacked acoustic richness. For example, I
held the sustain pedal down and hit a key in the lower
register hard. I hoped to hear some fifth and fourth
overtones - nope. It sounded exactly the same without the
sustain pedal. So much for resonance.
The soft tones were especially unacceptable to my ears. The
Yamaha literature brags about the range of samples listed:
ff, f, mezzo-forte, and damped. So by its own admission,
with no samples at p pp, the Yamaha apparently depends on
the damped samples to play softly. The result is dull,
thunking notes with no projection and no possibility of
legato playing (except by using the pedal).
Bottom line: The CLP-880 has a nice keyboard, but its sound
leaves a lot to be desired. It seems designed to best when
played loudly . Yamaha has not even attempted to emulate the
pearly soft tones of a good piano - one of the most
beautiful of all piano sounds. Yamaha seems to have pushed
conventional, sample-based sound generation to its limit.
The demonstrates to my satisfaction that this is a dead-end
approach, and - I hope - not the future of digital pianos.
At nearly twice the price, I wasn't convinced it would be
much of an improvement over my Kurzweil.
GeneralMusic RP10: Review
But there is hope. The night before seeing the Yamaha, I
went to the neighborhood music store and played the Italian
GeneralMusic PRO 1 keyboard (also goes by the "RealPiano" or
RP10, I think). The sound-generation system of this keyboard
is also used in the Baldwin Pianovelle. Now, the PRO 1 is a
pop/jazz performing keyboard, so was not tuned to sound like
a good piano and lacked a decent touch. But my impression of
it supports some of the manufacturer's claims about their
"synthesized" approach to the digital piano. The apparent
resonance of the sound was uncanny and frankly stunning.
What I mean is that if I pressed the sustain pedal and
struck a note, I heard the sympathetic voices of other
"strings" and the distinct effect of a resonating wooden
box.
According to the GeneralMusic literature on the web, their
sound generating system uses samples only in the initial
development. But in performance, the sound is synthesized
using complex algorithms, rather than filtered from the
sample. This obviates the need to store a massive database
of samples, and at least in theory makes it possible to
account for any acoustic conditions. They claim that they
have built in algorithms for all sorts of things that happen
inside the piano, including hammer noise and sympathetic
resonance from the soundboard and other strings.
Of course, the results can only be as good as the
algorithsm, but the demo I heard suggested that they're on
the right track. If so, it's such a palpable advancement
that I believe all high-quality digital pianos will
eventually adopt it.
So, obviously I want to play the Baldwin next. Of course if
it sounds and feels like the PRO 1 that I played, forget it.
To take advantage of this technology and advance the state
of the art as I see it, they need to tune the sound settings
to be more piano-like, and provide a good keyboard. That's
what I'll be looking (and hoping) for.
Kawai CA890: Review
Tried the Kawai yesterday. The showroom had some
decently-maintained acoustic baby grands, which was nice for
comparison. The CA890 has most of the bells and whistles
that the Yamaha had and a few that it doesn't. The action
felt just about as good if not better, but the sound was
fundamentally different: Kawai uses the synthesized approach
rather than just filtered samples. As a result, there is a
better simulation of ambient acoustics, and variations in
color depending on how you attack notes. I also liked the
way soft notes sound clearly and decay slowly - much better
than the Yamaha. The CA890 is also more reasonably priced
than Yamaha's comparable model.
On the down side, I couldn't seem to control the variations
in color: a slight increase in pressure brought wild
variations in color. For example, in a melodic line some
notes would sound okay, then all of a sudden there was the
bright metallic sound that you get when you really punch a
note on a real piano (ff). On the baby grand, there's enough
resistance in the action to control this. Maybe that's part
of the problem - the action on the digital piano is not hard
enough. More fundamentally, however, my sense is that the
piano is over-responding to variations in key velocity. They
should be more gradual for a more even sound. The Kawai
features presets for controlling the color: several piano
"types" (concert grand, "studio piano"...) and several
shades of "brightness". None of them brought the tone under
control except settings like "mellow" which were
unacceptable as piano settings anyway.
Bottom line: I think Kawai is on the right track with their
action and synthesized approach to sound generation.
However, in execution the sound modeling has some fatal
flaws that are surprising for a maker of such fine acoustic
pianos. Unless I can learn how to surmount them, this is not
a piano I can live with. I hope Kawai continues to develop
and improve it.
Conclusion
I'm going to sit tight with my Kurzweill Mark 5 for now.
Nothing I tried is better enough to reinvest all that money.
The only other piano I'd like to try is the Yamaha GP2 -
have seen some rave reviews - but it's pricier than any of
the models reviewed above. Overall I'm disappointed but not
discouraged. This is daunting technology, and it's just
gonna take time.
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
>snip<
I've recently tried several representative digital pianos.
See my very subjective reviews below:
* Yamaha CLP-880
* GeneralMusic RP10
* Kawai CA890
* Conclusion
>snip<
>I'm going to sit tight with my Kurzweill Mark 5 for now.
>Nothing I tried is better enough to reinvest all that money.
>The only other piano I'd like to try is the Yamaha GP2 -
>have seen some rave reviews - but it's pricier than any of
>the models reviewed above. Overall I'm disappointed but not
>discouraged. This is daunting technology, and it's just
>gonna take time.
Try the models in the Roland HP or KR series. I went through a search similar
to yours and was also disappointed until I came across the Rolands. You too
might find something you like, especially since you think that Kawai is at
least headed in the right direction with a synthesized approach. Roland uses
what it calls "advanced structure adaptive synthesis."
Richard
I looked at a Roland HP 230 series in a piano store the other day and liked
it.
The simulated wood grain body (HP 236) apparently raises the price a great
deal.
I'm gonna go price the basic black at the music store tomorrow. Why pay
more
for fake wood?
An interesting thought - I guess I'd have to take delivery on a trial basis
to make the experiment.
>Now that I think of it, you can change your synth too,
>using MIDI IN/OUT/THRU and the boxes they sell.
> Did you explore this option?
No, I'm not sure what box you're referring to. Do you mean that Yamaha and
others sell other types of piano "sounds" for these pianos?