Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Kawai announces Concert Artist CA98 and CA78 digital pianos

I think it is very difficult to objectively compare the two modes or even the difference between the pianist mode characters in terms of "quality".

Nevertheless, I have heard several digital recordings produced by the CAx7 and the CS models with the "old" sound mode. One of the main differences I hear is the "soundstage". The pianist mode demos are "wider" and have more spatial "depth". The sound mode is flat. it sounds like the mics are just capturing a basic stereo image. A related difference is that whenever there any multiple sounds being heard at the same time they seem to originate in different locations/sources in pianist mode (as different independent strings producing sound at the same time). In sound mode I hear what I hear on my DP: a "mass" of sound that blends together all individual notes. Just listen to the Full or Balad pianist demos and then to the sound mode demo. And then there is the resonance. Listen carefully to the Ballad and Vintage demos. This is what I am used to hear in Pianoteq, not in DPs that try to simulate resonance with a set of tweaked DSP effects.


I agree that the sound is rather dry. But I actually think this may be a good thing! I would say that what we hearing on these pianist mode recordings is mostly the result of the acoustic resonance model and not of reverberation effects at all. If this proves to be true, then there is plenty of margin to apply reverberation afterwards. And if the speaker system of the new models is good enough, then maybe we can rely on the natural reverberation of the room instead of additional effects.


I hear what I described above. The Chopin piece on the youtube video has a wider and deeper soundstage than your recording. Yours is a flat stereo recording. The second part of the youtube video also has a narrow soundstage, like your recording. Your recording also has a lot of reverberation. This is very noticeable during the rests, especially during the first couple of measures. In the video the only sound that remains is that of resonance. In your recording the sound continues to reverberate. The fast repeated notes also sound clearer on the video. And congrats for your performance :-)

BTW, I have listened to these demos with a pair of Beyerdynamic DT 990 (open) and Beyerdynamic DT 770 (closed) headphones. I would only point out that the soundstage of the pianist mode is even more pronounced than that of the sound mode on the 990s.

Hello Stuart,



The domestic (Japanese) market offers the CA78 in a 'White Maple' finish:


...however, this is not available overseas.

I shall pass on your suggestion to the product planning team, however I expect for the time-being the colour finish line-up will remain Rosewood, Satin Black, and Satin White.

Kind regards,
James
x

arc_turus, I think that's a great description!

Cheers,
James
x

I think arc_turus covered it quite well.

Overall, I find that the Pianist mode sounds more like you are surrounded by Piano sound in all directions. I also found that the resonances were better with it in some parts.

Osho

White Maple


White Maple looks lovely James!

Have to say the miserable colours would give me pause for thought about buying this.

Satin white feels a little too 'Elton John'. White maple took the edge of it and was more subtle in my opinion. Would make sense to offer it elsewhere.

Hello Stuart,

May I ask where you are based, please? (country)

Kind regards,
James
x

Sure, UK. London.

Okay, thanks!

I understand that the new sound engine requires more information than what is available via standard midi.
However, if I understand the tech correctly, MIDI supports custom extensions for describing data that is not covered by standard midi.

Has Kawai defined the extensions for doing this? Because if it did, then any software pianos connected via MIDI could make use of this extra information.

Can we hope for a future software update that would support this?

James, what it this data the Rendering engine requires that can't be sent through MIDI? This is really awkward. You play the piano through the keyboard and pedals. Both are supported in MIDI. If you are saying the keyboard generated high-resolution values that can't be transmitted through MIDI, this shouldn't prevent the engine from receiving low-resolution values. For example if the engine distinguished between a value of 27.5 and 27.6, why wouldn't it receive a value of just 27? The same about pedals. Unless there's some audio feedback that needs to go through the speakers and the audio system and then gets feed back into the engine for some reason (resonances?), there's really no reason for this limitation. Can you (are you allowed to) elaborate on this?

I am afraid this is not that simple

IIRC, for the keys, what you can transmit via MIDI is a single, 8-bit number describing the "velocity" of the key press. (That is, an integer number between 0 and 255.)
I can very well imagine that this keyboard collects a lot more data than that, and feeds it into the engine.
Standard midi might not have an explicit specification for covering that data.



James did say that some data is transmitted via MIDI... just not all the data that is used by the new engine.
So you can take the MIDI recording, and feed it into a VST (or even back to the DP), and it will control the engine exactly as it was with the previous generation.

So there is no loss here, compared to the previous functionality; except maybe some loss of opportunity, in making better use of the new information.

Since one of the main factors that makes the new engine so great is the better integration with the keyboard, I am speculating that we could get the same benefits when using with VSTs, if the extra information was transmitted, via custom MIDI extensions. (See https://en.wikipedia.org/wiki/MIDI#Extensions )

Obviously it would take some effort to craft a new protocol for transmitting this data embedded in a MIDI stream, but I see no technical reason blocking this.
On the other hand, this would only benefit the external VSTs (and the people using them), and Kawai might see that as a competitor to their internal sound engine,
so they might not want to invest the extra effort required to do this...

Did James actually confirm this? I recall there's been a lot of analysis and speculation based on the details that have been released, but has it been confirmed that one of the key points of the new engine is some kind of sensing outside the capabilities of standard MIDI?

Yes. The post by Csillag above CyberGene's post (the one that you quote) quotes James saying this. James' original post if even further up in this thread.

Ah, I missed, that, thanks! Carry on...

Played the CS-11 at the local Kawai shop at lunchtime.

Lovely feel!

I really hope the new ones feel as good, but sound better over headphones. (I think the new ones have the same action don't they as the CS-11?)

(And if they also came in White Ash! Then that would be very tempting!)

Have been toying with getting a Yamaha CLP-645. But I think the CS-11 has a better feel to play (though it is quite a bit more expensive). But I do find myself wondering about the new 68 and 98.

Pity they don't upgrade the CS-11 to a better headphone mode.

Guess they don't want to 'muddy the waters' for the new NV-10, which will also have the polished ebony finish.

Anyhow, FYI Rose Morris on Tottenham Court Road (London) said the end of the month (Oct 2017) for getting the new models in store to try, which is quite exciting!

James - Bend your development guys ears about getting the white ash model to the UK!

MIDI results from translating the physical actions on the keyboard to digital events, i.e. key on/off events, key velocities and the pedal events. I have read on this thread that the CAx8 is able to output MIDI data through its USB/MIDI ports. So I believe the limitation is the inability to internally record the MIDI data and not generating and outputting MIDI data.

I understand that playing standard MIDI files with the "pianist mode" rendering engine would require making many assumptions. This engine does not need to support most standard MIDI commands, such as channel changes, bank changes, DSP effects, and, especially, multiple parts/instruments. Supporting these features makes no sense on an acoustic piano rendering engine. But it could be done by playing just one part and ignoring the non-supported MIDI commands (like Pianoteq does).

But I would say that the reason for not having a MIDI player + recorder at all is manly because the "pianist" engine is likely based on a new architecture and hardware. This means that Kawai would have needed to develop a whole new MIDI recorder/player that works with this new architecture. And many SMF files may end not playing "correctly". So, they probably skipped the MIDI player altogether. And without MIDI player it makes no much sense having a recorder since one can use the existing recorder in "sound" mode or just connect to an external source and record MIDI in "pianist" mode... but I may be wrong on this...

Not sure concern about muddying the waters with the Novus is the reason, it's in an entirely different market segment. Also Kawai has previously offset the luxury/polished black models for previous generations.

Looking at this from a lower level of abstraction...

The CA-98 shares the GF2 action that was used in the CA-97. The action sends note on/off data from the triple-sensors system that is received by the CPU and translated to MIDI (I'm assuming here the action itself isn't processing the raw signals from the sensors and converting to MIDI data on its own, otherwise the point is moot). I presume this is still done for "Sound Mode."

So it would be interesting to know whether the GF2 has always been outputting more detailed/nuanced data that the CA-97 stripped in order to fit into MIDI and which the CA-98 preserves, or if the electronics in the action are actually different in the CA-98 that allow it to capture more data than the GF2 in the CA-97 could?

Or perhaps I'm thinking about this all wrong

I don't think that the delayed CS model has anything to do with the Novus. The delay was the same in the past: The CS11 is equivalent to the CA97 (with the nicer cabinet) but came out more than a year later. I assume the CA98 equivalent CS?? model will come out with a similar delay.

I would love to see what the guys at Ravenscroft (and other similar groups) could do with their virtual instruments, if they had access to the extra data used by the engine in "Pianist" mode.

(For those have haven't seen it yet, earlier they custom-tailored their virtual piano to work best with a (customized) Kawai VPC1 ... see here: https://www.youtube.com/watch?v=BUC0xU7OX54 )

I bet they would even be happy to help with designing the MIDI extension required to transfer that data from the KAWAI keyboard to their software.