Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards The DPBSD Project!

Galaxy Vintage D Review

Yesterday setchman kindly provided us with a DPBSD MP3 of the very latest Galaxy piano, the "Vintage D" - a sampled 1920 Steinway D using Kontakt 4 for playback. Thanks setchman!

It is of course fully sampled, with no stretching or looping evident. The velocity layers appear and sound blended through the broad middle velocities, with semi-audible switching at the soft and loud ends, but the switching is by no means objectionable, and the timbre variation with velocity is nicely done.

A few nitpicky things with the damper pedal though. It fails the pedal down silent replay test in two different ways, the vel=1 note causes a distinct muting of the note, and the pedal up event following causes total muting even though the key is still being held down. And I don't hear any sympathetic resonance when a single note is played first and the pedal pressed a bit later. Not surprisingly, DPs that use delay effects for the sympathetic resonance usually get this right, but a sampled instrument would have to somehow crossfade over to the pedal down sample of the note (perhaps they are doing so and I just can't hear it?). Finally, the pedal up/down sounds are triggered around the 50% pedal point, which is rather non-realistic when partially pedaling and the pedal is only moving a bit.

As usual, the text review, the MP3, and some pictures of the analysis are located at the share point for your viewing / listening pleasure.


Spectral frequency view of the layer test - timbre changes smoothly with velocity.


Spectral phase view of the layer test - some layer switching is visible, particularly when zoomed up, and midrange is nicely blended.


Spectral phase view of the silent note repeat test - cursor @ pedal up; major damping before that caused by replaying the note @ MIDI vel=1 with pedal down.


--------------------
- Galaxy Vintage D -
--------------------
FILE & SETUP:
- dp_bsd_v1.5_galaxy_vintage_d.mp3
PROS:
- Passes the pedal down sympathetic resonance test.
- Passes the key down sympathetic resonance test.
- Sympathetic resonance sounds realistic.
- Pedal up/down sounds are "thunks".
- Key up sound/effect is a realistic "buzzy" sound.
- Nice long decays on the order of Pianoteq.
- Responds to partial pedaling.
- No visible or audible looping, notes appear and sound 100% sampled.
- No visible or audible stretching.
- Large dynamic range (~55dB, vel=1:127).
- Appears to be a somewhat blended multi velocity layer sample set (I see at least 9 layers).
CONS:
- Fails the pedal down silent replay test @ vel=1 & pedal up.
- Pedal down after playing a note doesn't cause sympathetic resonance.
- Pedal up/down sounds trigger @ 50% pedal, even when partial pedaling.
- Visible velocity layer switch @ vel=7,15,22,36,49,53,57,61.
- Audible velocity layer switch @ vel=7,15,36,53,57.
OTHER:
- C7 key-up during looping test makes a "distant thud" damping sound.
- MP3 levels: peak @ -1dB, noise floor @ -90dB.
- Date reviewed: 2010-03-11

Just a quick question, has anything ever passed the "pedal down silent replay test"? It seems that every review I've read has that as the first con. So should it really be in the test at all since it seems nothing does it?

Honestly, I didn't read every review, so I don't know if one does it or not. Just seems a majority of them don't.

Yes a few would pass it but Dewster has only just added this element to the test and it happens to be that the pianos recently tested do not feature behaviour that would pass this test.

Steve

The Yamaha AvantGrand N3 passed. And as Steve indicated, this particular test is new. It's located near the front of the tests, so it tends to get commented on first.


All of the reviews are in a single text file located here.

Ah, thanks. Guess I should read more of the older reviews too.

Dewster: Yes, you are spot on re: the crossfading. I have at least one instrument that employs crossfading for the sustain resonance. (Pianowave's "Kalamkarian Bechstein"). It appears that both the pedal-up and pedal-down samples are always played simultaneously (with the attendent hit on polyphony!), with the cross fading invoked by the sustain pedal.

Unfortunately, the EW Quantum Leap Pianos does NOT employ cross fading, and has the same problem as the Vintage D appears that it may have. (I'm taking your word for it - I realise you are not 100% sure though)

I'm curious to see how well Ivory II does sustain resonance - I'm feeling optimistic.

Greg.

Interesting! BTW CA63 does support cross fading, i.e. activates damper resonance even if the pedal was pressed after a key was hit.

Oops - it's actually the K-Sounds "Signature" that seems to play the sustain samples at the same time and cross fade - the Pianowave Bechstein actually seems to invoke the sustain samples only when required, saving polyphony! Both are Kontakt based.

Greg.

Most likely a dispersion delay effect, rather than actual pedal down samples. The pedal probably crossfades between dry output and the output of a short reverb.

I'm not aware of any DP in hardware that has a pedal down sample set (like they do for PC samplers).

No it's definitely not a reverby effect, I can distinguish the difference. It's a very realistic pedal down effect. It sounds like additional overtones. I don't know if it's a real sample or some virtual piano calculation.

BTW even the CLP-340 has stereo sustain samples...

Arrgh. It now appears that both the Pianowave and the K-Sounds have both sets of samples playing all the time, with the crossfading. However, for some reason K-Sounds simply uses more voices, all the time. The reason for my confusion is that I always thought Kontakt counted a stereo pair as two voices, but it does not - it counts it as one voice. One thing's for sure - they both smoothly switch on the sustain resonance when the sustain pedal is pressed.

When testing this, watch out for pedal clunk samples temporarily increasing voice count.

Greg.

I could be wrong, but I believe a "stereo sustain sample" is just a recording of all the strings vibrating after being excited by an impulse, and not by the specific notes being playing. Couple that with a little reverb and it could likely sound somewhat realistic. Genuine note samples with the damper pedal down would double the sample memory requirements, and we all know how crazy that notion is among DP manufacturers.

Well I could be wrong, but I guess I have a pretty well trained ear regarding those subtle real acoustic sound effects, as I play acoustic piano quite often. When I have the damper down each individual keypress produces unique resonances. When setting reverb to "off" this can be observed much better. There are also specific overtones ringing depending on which key you've pressed which sounds very realistic to my ears. This can't be accomplished by simply recording all the strings vibrating. Since KAWAI keeps this as a company secret (?) I cannot really tell you how they accomplish this, but it's done very well. There is also a nice smooth transition from non-damped to damped sound when you press the damper pedal after you stroke a key.

Speculation: This might also be quite polyphony intense, which is why they might have this polyphony issues from time to time and 192 notes polyphony is really needed to activate all effects and still have enough sounds available.

Dewster, may I suggest two tests for you to add. It is something like a "re-pedalling" but not exactly This is a feature on real pianos, digital pianos but not all of the software pianos:

Test 1
1. Press a key, preferably a bass note (because it has longer key-off)
2. Release the key
3. Immediately after the release press the damper pedal.

Result: Depending on the key, the note velocity and the duration between the key release and damper press, you will be able catch the sound and keep it sounding.

Reason: The lower the note, the longer it takes for the damper to mute the string when you release the key.

(Result with e.g. Steinberg The Grand - even if you are "Speedy Gonzales", there is no chance to catch the note. You must press the sustain BEFORE you release the key if you would like to do a sustain)

This is a very realistic real scenario. If you play on a real piano, it is very possible that you release a key and in a matter of a thousand of a second you press the damper pedal. Both happen virtually at the same time and the result would be the same as if you have pressed the damper pedal before you have released the key. However if serialized via MIDI, these events are sequenced and it depends on the algorithm to interpret them correctly.
-----------
Test 2
1. Press a key, preferably bass, and preferably loudly
2. Press the damper pedal.
3. Release the key.
4. Do a multiple quick "pumping" up and down with the damper pedal.

Result: Each time the damper pedal is pumped up, it begins muting the vibrating string, but since the muting is not instant, on the pump down motion you are letting it sound again, etc.

Reason: same as in Test 1

(Result with e.g. Steinberg The Grand - on the first pedal-up the notes are dropped)

The second test is very common technique in the classical piano repertoire, to simulate a sostenuto pedal if it is not available - you press a bass note, hold it with the damper pedal, then move both your hands in the highs to do some arpeggios but in the meanwhile you pump-up-down from time to time in order not to let too much high notes to create a mess. The highs get muted by the damper almost instantly, so you can clean the high notes by that pumping motions and at the same time the bass note will not get muted if you are quick enough with the pedal and if you are not lifting it too high (half-pedalling is a must).


Most of the sampler libraries are not supporting that feature. In fact only Pianoteq (because it is modelled) and Ivory (because it is professionally implemented) support it. Digital pianos have no problems maybe because they are looped and it is very easy to restart looping.

No. 2 is implemented on many DPs especially those that support half pedalling. No. 1 would be indeed interesting, Never checked this.

I am sure you have faced the Test 1 many times, but you never knew this existed simply because digital pianos are implementing it correctly. I realized all those effects for the first time when I tried Steinberg The Grand recently and there were many times when sustain seemed to not be working and I thought it was a faulty sustain pedal... After analysis it appeared I was pressing the damper pedal slightly after I have released keys. I've experienced that with Steinberg The Grand and Native Instrument Akoustik Piano only but I suppose it is a bug on other non-looped sample libraries.

Yes, I think the issue with the non-looped software pianos is that they often have release samples. It's difficult to produce the half pedalling whilst at the same time retaining the real release samples. Without the release samples, they could do the half pedalling in the same manner as looped digital pianos do. (just apply an envelope to the full length sample in the same manner)
Ideally, the software piano would provide an option to the user: half pedalling without release samples, or no half pedalling, but with release samples. In fact, I have a third party script that works with Sampletekk's "White Grand" that does provide half pedalling (or at least, the ability to "catch" the notes as described earlier), but forgoes the release samples.

Greg.

Interesting. I'll certainly take it under advisement. I would need to experiment on a few DPs to see if similar timing would work across all of the ones that support these effects, and to see how difficult it would be to reproduce and analyze.

For the second test, some DPs that support partial pedaling have the transition between damped and undamped centered more towards pedal down, and that might stymie a single test. Perhaps multiple pedal levels could be used in two or three tests in order to have a better chance of capturing it.

Just curious:

What do people think of the DPBSD MP3 files? Do you download and listen to them? Do they help you? MediaFire keeps track of how many times a particular file is downloaded, and there seems to be a certain amount of activity there.

How about the analysis pictures? Do you find them useful?

I like the analysis pics and I also listened to a few of the MP3s of the pianos I'm interested in (such as the HP-307, V-piano)... Keep it up