Pianoteq Trial - Poor sound quality?

rdlom, unless there was an inadvertent mistake of using obviously dissimilar MIDI files to compare Ivory and Pianoteq, I am bound to conclude that your post has the intention to mislead readers of this forum.

Whatever happened to the sound on the Ivory rendering at 32.250 seconds?  It just disappeared.

The E flat octave (with B flat above) in the bass at 45.815 seconds is virtually non-existent, the second one is slightly better.

But I don't think these problems are due to the midi file used.

More to come.

Glenn

Hi rdlom,
I have posted for you the exact setting that was used for recording Hugh Sung's demo Clair de Lune that is on our "discover_pianoteq3" page. You can download it here: http://www.forum-pianoteq.com/uploads.p...835%29.fxp
I hope you will enjoy it!

Not necessarily, I use the soundcard that came with my Mac and I'm happy with the results, I believe Hugh Sung uses an M-Audio Fastrack Pro which isn't an expensive interface.

http://www.m-audio.com/products/en_us/FastTrackPro.html

I've had Ivory for a few years, and it's sound, most noticeable on the Yamaha C7, seems to emphasize the attack and deemphasize the midrange, both of which give the sound a lot of definition. There's less of that, but it's still there, with the Steinway. That type of sound can be very desirable at times. However, tonal consistency from note to note is not Ivory's strong point.

There seems to be a lot less attack and more midrange with Pianoteq which, at times can sound muddy to me. But the tweakability is amazing. I've found that adjusting mic position, hammer hardness, EQ and the velocity curve can give the midrange a bit more definition. No doubt those who actually know what they're doing can acheive more.
But can Pianoteq duplicate all the 'good' aspects of the Ivory type of sound at this time? I don't believe so, but I could be wrong.

That mic setup sounds interesting, I'll be trying it. As far as listening from a playing position, I agree, that's the sound I'm used too when I'm playing. When I'm listening to a recording I'm hearing the piano, most times, from an audience perspective, except that, if a recording sounds very intimate, chances are that it's a closely mic'ed piano, or a player's perspective.
Many of the major sampled piano developers offer 2-3 or more perspectives, which is only one of the many reasons their 'libraries' are so huge. EWQL Pianos take up over 240GB of space, which I think is ridiculous given the fact that they don't 'breathe' the way Pianoteq, at 20mb or so, does.

rdlom,

If you're listening to both the Ivory and Pianoteq files on the same computer, then the soundcard has nothing much to do with the differences you hear.
A better soundcard would probably make them both sound better, but the sound relationships/differences you're getting now would remain the same.

I don't mean to be deprecative of Pianoteq.
It pursues a more sophisticated approach in reducing the piano to a mathematical model. I would love to get rid of my sample VST, but what
counts in the end is the quality of sound and for some reason I cannot reproduce the demontration samples quality.

If I had to change Pianoteq's velocity to make it sound more authentic using the same midi, then the velocity algorithm is faulty;
because if you judge each sample by their respective accuracy,
Ivory's velocities seem to be correct.

I don't think the velocities are explictily at fault here.
I just cannot narrow it down.

I have a acoustic Kawai, couple of hardware digipianos and then Ivory and Pianoteq. PTQ is in my opinion much better INSTRUMENT than Ivory: PTQ responds to players fingers, has these marvelous resonance features and can project players feelings. Ivory can sound more natural when you play a midifile with it but for me it isn't a real solution for PLAYING piano. But I understand these first impressions about PTQ: there's still some artificiality in its sound. I must conclude that there's no single perfect choise at the moment: some hardware pianos are quite good for a player; they sound good enough for a stage situation and they work well with their own keyboard and they don't have latency problems. Many giga level sample based software pianos sound very good for studio purposes but you wouldn't take them on a stage. PTQ is maybe somewhere in the middle: It sounds great if you aren't after the most realistic piano sound, it's expressive and highly nuanced.

Whole thing depends on what you're after for...

P.S: Still I think we're close to moment, when modeled piano takes the final victory over sampled piano.

Ah yes, Loony Claire, I remember her well...

What a great memory to have, Glenn

I'm going to suggest a little test:  I will upload a midi file of the first half of Liszt's Hungarian Rhapsody No.2 (RhapsodieHongroiseNo.2).

The full piece is simply too large to upload on this site, even in mp3 format.

I will also upload an mp3 I rendered in Pianoteq.

Then someone can render the same midi file with samples for an interesting comparison.

Glenn

PS - Edited:  Correction - the original piano roll was recorded by Paderewski and converted to midi.

There are a couple of places where it seem like he might have torn the ivory off the keys.

thank you rdlom for the new recording. The fxp that I uploaded is exactly the one used for recording Hugh Sung's demo on our site, so if you still find a difference in the final rendering, it shouldn't be due to Pianoteq but to musical interpretation.

Concerning velocity:

• the default velocity for a given midi file is not necessarily the best: it depends on the keyboard that was used for recording the midi file,

• Pianoteq does not behave like a sample library. In a sample library with a limited number of layers, many velocities share the same sample. For example, if velocities 70 to 100 use the same sample, then part of the velocity 100 crispiness is present in the lower velocity 70. In Pianoteq the variation is much smoother, and mezzo forte is really mezzo forte. If you find the sound too mellow, it is very easy to change, either increase the hammer hardness or the velocity curve.

As a final thought about default setting, you cannot have one setting that is good for all keyboards, midi files and styles of music. Even in the real world, the piano tuner will adapt the piano to the pianist and the room, the pianist will adapt his playing to the music, the piano and the room, the sound engineer will use different recording tools and techniques depending on the context, etc. Pianoteq offers you many possibilities, they should be used if you want an optimal result.

Are you sure you used the midi I uploaded in my first post?

It does sound much better, but the reverb at the end of each note is a tad bit too strong sometimes.

Can you please upload the FXP.

Why did you decide to use Garageband's reverb instead of Pianoteq's own?

Did you want the full midi of Hungarian Rhapsody No.2?

I'll upload it.

Glenn

On the Pianoworld forum (Synths and Keyboards) there is a thread entitled "Distortion in Pianoteq3.5?"

One responder made a comment I agree with, and I replied as follows:

The "limiter" is to my mind, one of the often overlooked and abused settings in Pianoteq.  Most of the default settings in Pianoteq use four or five microphones (headphones being the exception).  On my older dual core with these settings, the output level must be turned down to prevent clipping, OR, the limiter must be turned on.

The problem with the limiter is that it is essentially a compressor (preventing clipping of peaks).  Compression results in some distortion (sometimes this does sound better).

To put less strain on my system, I use two microphones, . . . to achieve the sound I want.

There is an "audio load indicator" in Pianoteq - switching to two mics from four mics, the load drops by about 1/2 - which seems to indicate that the mics use a lot of processor power compared to that used for production of sound.

Does anyone have any comments - pro or con - to my observations?

Glenn

Don't forget to reduce/remove the reverb, lower the impedance, adjust the strike points, etc. -- you can achieve a _ridiculously_ clear sound this way, though you'll obviously lose a little of the "body"...

Speaking of body, I prefer the YC5 add-on to the C or M models.  It has a bit more "punch" and clarity in the middle and upper registers.

And who knows what "miracle pianos" may be in development?  I consider that YC5 to be a step in the desired direction you seek, so here's hoping for more of the same!!!

(And don't forget to sample the other instruments -- I love the vibes and the new Clavinet.)

Oh, goodness, you're going to make me work for this...

Okay, will you be happy with a very clear sound for the "Clair de Lune" MIDI file, or do you have some other MIDI or composition to use as a reference point?

I'll see what I can do for you, but it may take a few days.  ;^)

Fortunately I wasn't holding my breath waiting for a response to post No.25 above  wherein I was hoping for renditions of Liszt's Hungarian Rhapsody No.2 done in samples (or other Pianoteq pianos).

I think HR No.2 is a much more difficult midi to render to wave than Claire de Lune because it has such a range of dynamics and speed of notes.  Some of the chromatic scales are extremely fast which may challenge some systems to play in real time.  (Starting at about 2:10 and again at 3:10, the repetition rate of notes is very fast).

My five year old dual core Athlone can handle it with Pianoteq.

Glenn