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

JFP,
In my opinion it sounds and PLAYS completely different. The samples (although taken from the same original instrument) are in an engine that allow them to be significantly more expressive and melodic. As Dewster says, more dynamic range and more timbre variation.

We have some more audio and video coming very soon too.

Thanks for the interesting review dewster, and kudos to Mike for submitting the recording.

Just to clarify, what is the approximate difference in sample size between the PX-330 and PX-350? I recall reading something along the lines of a 3x increase in memory, with lossless compression etc. Is that evident in this DPBSD submission?

Cheers,
James
x

Good on you Mike, for submitting the file for Dewster to look at.

This seems like a very welcome increase in attack sample length. This, added to the more powerful sound engine, improved action and onboard audio recording makes the PX-350 look like a fantastic choice at its price point.

Cheers,

Steve

Layers are reportedly the same (4). I'm not sure what to make of the lossless compression statements.

The DPBSD testing shows loop samples are roughly the same length, stretching is ~20% less, and attack samples are roughly 3x longer.

Commenting generally, these specs are difficult for me to interpret, so I usually don't put them in the pro/con sections of the text review.

A certain amount of timber variation is obviously necessary and good, but the timbre variation on (for example) Roland SN pianos is described by some as overly strident at higher velocities.

If dynamic range is too narrow the DP can sound compressed, such as the Nord DP voices which are in the low 30's (dB). On the other end some criticize Yamaha DPs for having unrealistically wide dynamic ranges in the 60's, which can perhaps make lighter playing more difficult.

So this is a dual example of "more is better, but only up to a point". And of course a lot depends on how these things are implemented (timbre variation distribution, how it plays, etc.).

Yamaha P-105



Another first for the DPBSD project - an anonymous submission!

An individual who asked me to not reveal their identity generously supplied us with DPBSD MP3 of the new Yamaha P-105. The MP3 (and pix) are of the default piano voice "Grand Piano 1". Those interested can also listen to the compressed layer test:

PIX: http://www.mediafire.com/?ei5m55d5s3oad5y
MP3: http://www.mediafire.com/?bh4tsq222247tyc
MP3: http://www.mediafire.com/?wc5813ne06pahdm (100:1 compressed layer test)

There's a lot of activity at the low end of the DP market this year. I'm on a constant lookout for beginning instruments for my wife's students, more portable fare for her to take on vacations & the occasional playing out date, what to recommend for those at PW on a tight budget, etc. So the re-spin of the Casio line, and CF samples showing up in Yamaha's entry level are potentially exciting developments.

Let's start with the P-105 note decay times. They are long for the lowest notes, but rather short for mids and highs. The attack and loop samples themselves are pretty short, and I believe this contributes to the crossfades and decays on the lowest notes sounding somewhat uneven. The loops are highly processed and bland sounding, lacking any realistic interbeating or "wobble".

This voice is fairly stretched with 29 samples covering 88 notes. Stretch group transitions are audible over most of the range due to timbre variation and some moderate L&R pan inconsistencies (I believe L&R are probably swapped in the recording).

Pedal sympathetic resonance for the P-105 is subtle, though it helps to smear out the slight audible quavering of the loops in the chord based test. The layers are well blended, but there doesn't seem to be much timbre variation at the highest velocity range where one would expect the most variation to be occurring.

Hats off to this anonymous person for supporting the DPBSD project! Should (s)he and you cross paths, please buy him/her a beer!

Some analysis pix and text review:


Figure 1. Spectral phase view of the pedal sympathetic resonance test, pedal down @ left, pedal up @ right, stimulus removed, normalized to -1dB. Sympathetic resonance is both visibly and audibly subtle.


Figure 2. Waveform view of the entire looping test, vertical zoom applied to see the noise floor. Note decay is long over the low end, rather short over the rest of the range.


Figure 3. Spectral pan view of the note C4, normalized to -1dB to increase clarity, cursor at attack / loop transition. Attacks are fairly short, loops are really short and processed sounding.


Figure 4. Spectral pan view of the stretch test, mid notes, normalized to -1dB to increase clarity. 29 stretch groups are clearly visible and audible over most of the range.


Figure 5. Spectral pan view of the layer test, highly compressed. Two visible layer switches, the second transition of which (@ cursor) is slightly audible to me (I believe Yamaha claims 3 layers). You can listen to this as a separate MP3 file (listed above).

----------------
- Yamaha P-105 -
----------------
FILE & SETUP:
- dpbsd_v2.0_yamaha_p105.mp3
- yamaha_p105_layers_comp.mp3
- Recorded using TC Electronic Desktop Konnekt, Sonar X1, WAV=>MP3 w/ Adobe Audition.
- This is the first patch: "Grand Piano 1".
- Recorded by an anonymous donor.
PROS:
- Passes the pedal sympathetic resonance test, the effect is very subtle but helps smear out looping.
- Passes the silent replay test.
- Passes the late pedal partial damping test up to the test limit of 0.5 seconds.
- Passes the half pedaling test.
- Low note decays are fairly long (~Pianoteq), mids and highs could be longer (~1/2 Pianoteq).
- This is a smoothly blended multi-velocity layer sample set (two steps visible @ v=72,90).
CONS:
- Fails the key sympathetic resonance test.
- Pedal sympathetic resonance is very subtle.
- Partially damped notes don't sound "buzzy".
- Obviously looped, the loops sound static rather than "wobbly".
- Attack sample lengths are (C1:C8): 2.0,1.9,1.9,0.8,1.1,0.8,0.6,? seconds.
- Loop sample lengths are (C1:C8): 1.7,1.2,0.5,?,?,0.25,?,? seconds.
- Stretching is visible over the entire range, audible over the low and mid notes.
- Stretch distances: 3,3,4(x3),2,4,3(x3),2,3(x3),2(x4),3,4,3,4,2,2,3,5,4,3,3 = 29 groups.
- Somewhat audible timbre step @ v=90.
- L&R pan of stretch groups up the keyboard is non-monotonic.
- No obvious key up or pedal down sound effects (via MIDI).
OTHER:
- L&R are likely swapped in the recording.
- Notes played @ vel=1 produce no sound.
- Dynamic range ~61.5dB (vel=1:127).
- MP3 levels: peak @ -2.5dB, noise floor @ -81dB.
- Dampered | undampered transition: F#6 | G6.
- Date reviewed: 2012-09-28

Thanks dewster (and anonymous)!

How about some comparisons with the P95, or PX-150?

Cheers,
James
x

P-95:
- Attack sample lengths are (C1:C8): 2.0,2.0,1.8,1.7,1.3,1.1,?,? seconds.
- Loop sample lengths are (C1:C8): 0.6,0.6,0.6,0.4,0.3,0.3,?,? seconds.
- Stretch distances: 2,3(x28),2 = 30 groups.
- 1 layer.

P-105:
- Attack sample lengths are (C1:C8): 2.0,1.9,1.9,0.8,1.1,0.8,0.6,? seconds.
- Loop sample lengths are (C1:C8): 1.7,1.2,0.5,?,?,0.25,?,? seconds.
- Stretch distances: 3,3,4(x3),2,4,3(x3),2,3(x3),2(x4),3,4,3,4,2,2,3,5,4,3,3 = 29 groups.
- 3 layers.

P-95 vs P-105:
- Attack sample lengths are about the same.
- Lowest note loops on the P-105 are a bit longer.
- Stretch group count is a wash.
- Three layers on the P-105 make this sample at least 3x larger than the P-95.


PX-350 (purportedly same as PX-150):
- Attack sample lengths are (C1:C8): 4.3,3.7,3.4,3.4,3.5,3.4,1.0,? seconds.
- Loop sample lengths are (C1:C8): 1.2,1.3,1.1,1.1,1.1,1.1,0.7,? seconds
- Stretch distances: 3,2,3,2,3,2,2,3,2,3,2(x5),3(x5),2(x3),3,2,2,3,2,3(x4),4,4 = 34 groups.
- 4 layers.

P-105 vs PX-350:
- PX-350 stretching is a bit less.
- PX-350 attack sample lengths are double in the bass, triple in the midrange.
- PX-350 loop sample lengths are about the same in the bass and at least double over the rest of the range.
- PX-350 has an extra layer.

Thanks.

I assume 'is a wash' means 'is the same as...', correct?

James
x

Dewster:
The PX-350 has a relatively long attack phase. But what can generally be said about the sound quality when it comes to length of attack phase on the one hand, and the amount of stretching on the other hand, compared to other DPs with lesser length of attack phase, but much less stretching. Such as: would you prefer a DP which has almost no stretching, though much shorter attack phase? What do you mean with "wobbly" loops?

As the sound decays and the vibrations of the string decrease, different harmonics are emphasized, which create a kind of "beating" of the note. In other words a static synth-like decay is unnatural. A lot of DPs have a problem recreating this natural beating, and so sound rather dead as the note decays.

Thanks for the explanation,voxpops. So, "wobbly" is more realistic.

If it's done well, yes.

Yes. Sorry to those who don't follow our own strange colloquialisms.

You're asking me to make something of a Sophie's choice when I believe technology and the economics thereof have advanced enough to make this issue moot. It all depends on the implementation, but I guess I'd generally take longer samples over no stretching.

Could anyone tell me which of the Yamaha tested piano voices would be the closest to the Yamaha CP33?
Thanks.

Not exactly 1:1 / apples-to-apples, but the closest would probably be P155, which was tested three different times, though the CP33 is 3-layer vs 4-layer and 64 vs 128 polyphony, IIRC. Not sure how much that factors in Dewster's tests.

If you're asking which sample, I'd guess GP1 for both.

Thanks, but I believe that these two are quite different (actually I was trying both of them last weekend and the piano sound is not the same).
If I had to bet, I think that maybe the piano sound of the P120 would be closer to the CP33 but I'm not sure...

I haven't reviewed the CP33, but that would be my guess too.

Thanks dewster, and by the way thank you for your excellent and interesting work.
Let's see if someone brings a CP33 sample...