I wish it were that simple.
 
The human ear, in the range of frequencies that we use in music, is as good or better than the very best microphones. Simply put, we hear about half the information that is physically present in the signal. That's the best physical performance of any of our senses. For example, with night adapted vision, in the dark we can see almost every photon that strikes the retina, but film or a sensor can do the same, and we can't resolve the photons spatially nearly as well as a modest lens -- or even a hawk, for that matter.
 
What this means is that if there is ANY flaw in the musical signal, a person with educated ears or musical experience is going to hear it -- and right away. 
 
And THAT means, even the best recording is never going to sound quite like the real thing.
 
The physical models, inevitably, make simplifying assumptions. That means that they are NEVER EVER going to sound just like even a... recording... of the real thing. Never mind the real thing itself. The models will ALWAYS be at least two steps removed from the real thing.
 
However, you are totally right about the tweakability. So, in my experience, even though the models can't sound REAL, they can actually be MORE USEFUL in certain musical textures because, with the tweaking, you can get the frequency balances to sit in the mix better, you can get the spectrum to be more useful for your purposes, you can get the attacks to be the right lengths for your rhythms, and so on and so on. In short... with models you can COMPOSE more.
 
Of course, in the future, the two steps removed will get smaller and smaller, and a skillful composer may be able to get us to forget about them for a time. But they are never going to go away entirely.
 
Hope this clarifies things a bit,
Mike
 
----- Original Message ----- 
From: Brian Redfern 
To: csound@lists.bath.ac.uk 
Sent: Tuesday, August 19, 2008 8:37 PM
Subject: [Csnd] Re: Re: Re: Re: Re: Re: Re: Modeled piano- Pianoteq


What I like is that pianoteq is more "tweakable," models are the future anyways, any problem with its sound could be addressed by future versions, since its just math, but with samples what you have is literally "set in stone."




On Tue, Aug 19, 2008 at 5:33 PM, Michael Gogins  wrote:

I found that by playing with the Pianoteq's resonance, high-frequency cutoff, and Q, I could get a sweeter, more ringing sound, more like the good SoundFonts.

Still not as much like a real piano as the good SoundFonts.

I know from doing this kind of work that with endless tweaking, and circling around different use cases and sets of parameters, you can usually get something quite a bit more usable.

Of course that should apply both to the Pianoteq and to the sampled pianos....

Regards,
Mike


-----Original Message-----
>From: Brian Redfern 
>Sent: Aug 19, 2008 8:08 PM
>To: csound@lists.bath.ac.uk
>Subject: [Csnd] Re: Re: Re: Re: Re: Modeled piano- Pianoteq
>
>The sampled piano on my Oasys kills the pianoteq, but at the same time, i
>can't load scala files into the oasys and I can't do really weird stuff with
>it, the modelling aspect of pianoteq works great for really strange tweaked
>piano sounds.
>
>On Tue, Aug 19, 2008 at 5:04 PM, Michael Gogins  wrote:
>
>> Thanks for your very interesting post on this very important topic.
>>
>> I tried the experiment you recommend just now, with J.S. Bach BMV 533, an
>> organ piece with a lot of range and a lot of stuff including block triads in
>> it, the Pianoteq, and the sfz freeware Sound Font 2 plugin with the free
>> Piano Steinway Grand Model C (21,738KB).sf2 SoundFont in Cubase 4.
>>
>> In short, I find both pianos quite usable, but I still slightly prefer the
>> Pianoteq.
>>
>> Some additional comments....
>>
>> With the Pianoteq, I can indeed hear the beating you describe. For me, in
>> the passage work and in melodies that I played or that were sequenced, it is
>> not objectionable. Also, I could reduce the beating to almost nothing by
>> increasing the octave stretch, or by using well temperament instead of equal
>> temperament. I suspect that in any given key, you can do something to
>> eliminate all the beating completely.
>>
>> The Pianoteq has a more even range, no recording artifacts (of course), and
>> is in better tune (in spite of the beats).
>>
>> Any given note on the SoundFont definitely sounds more like a piano,
>> because of course it is a recording of one, but the unified effect is more
>> jarring to my ears, because of recording artifacts, uneven range, and so on.
>>
>> But I tried several free SoundFont pianos, and they were each quite
>> different, so I suspect a commercial sampled piano could be better --
>> perhaps quite a bit better.
>>
>> I think for some textures, I would use the SoundFont for its sweeter, more
>> piano-like sound, but for most textures where the behavior of the instrument
>> is important, I would use the Pianoteq.
>>
>> In sum, for me the Pianoteq paints a more unified picture of a piano-like
>> sound, even if the sound is not quite as much like a piano. And, of course,
>> it BEHAVES a lot more like a piano than the SoundFonts do.
>>
>> Finally, I use these instruments in ways that most composers for the piano
>> do not. I use thicker textures, more notes, more rapid notes, more precise
>> timing, and so on. In these contexts, the more tractable behavior of the
>> Pianoteq is more useful -- until it completely runs out of notes! This is
>> something that the SoundFonts just don't do, since they don't eat computer
>> power the way the Pianoteq does.
>>
>> Regards,
>> Mike
>>
>> -----Original Message-----
>> >From: Michael Mossey 
>> >Sent: Aug 19, 2008 6:15 PM
>> >To: csound@lists.bath.ac.uk
>> >Subject: [Csnd] Re: Re: Re: Modeled piano- Pianoteq
>> >
>> >
>> >
>> >Michael Gogins wrote:
>> >> I have used, and continue to use, the Pianoteq quite frequently. In
>> >> my view it is not the only piano plugin one might want to use, but
>> >> for me at any rate, it is certainly the most useful. The big chords
>> >> are not as convincing as a sampled piano, but everything else is more
>> >> convincing.
>> >
>> >I guess I'm the only one then. I played with it some more---tried
>> >different stretch tuning, etc. The intervals are just harsh and
>> >artificial sounding.. they have very odd-sounding beats. Michael, have
>> >you tried playing a simple triad on Pianoteq vs. a sampled piano? Have
>> >you tried sequencing a Bach choral, in slow motion, so you can savor the
>> >harmonies?
>> >
>> >
>> >I agree that individual notes are fantastic. A single melodic line WOULD
>> >be great, except I can't get past the beats that take place in the
>> >release sound and in the ambiance. A single melodic line is often filled
>> >with major and minor seconds, which are the hardest intervals for me to
>> >accept on the Pianoteq.
>> >
>> >Best,
>> >Mike
>> >
>> >
>> >Send bugs reports to this list.
>> >To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>
>>
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>
>
>Send bugs reports to this list.
>To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"





Send bugs reports to this list.
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Michael Gogins wrote:
> So, in my
> experience, even though the models can't sound REAL, they can
> actually be MORE USEFUL in certain musical textures because, with the
> tweaking, you can get the frequency balances to sit in the mix
> better, you can get the spectrum to be more useful for your purposes,
> you can get the attacks to be the right lengths for your rhythms, and
> so on and so on. In short... with models you can COMPOSE more.

I agree with this point, which I would also phrase as: there's a 
difference between sounding "real" and being useful in some other way, 
such as sounding beautiful or musical. I think the Pianoteq sounds 
beautiful in many ways. I also think a lot of simple waveguide plucked 
strings can be beautiful.

Best,
Mike

The first listed Objective in the file CsoundVST.csd ("the Silence
orchestra") that comes with the Csound distribution always intrigued me:

 "- Most beautiful sounds"

I always thought: what does that mean?

It's dawning on me, thanks to this discussion thread, that there is EQUAL
validity in using modeled or sampled instruments depending on your
compositional goals. For a long time I hung onto the notion that if I was
writing a part that seemed to call for a small brass section, I should look
for the best samples of a small brass section obtainable. Now I realize that
that approach is not the only one...

Also I remember a remark by Frank Zappa (I think it was in the Real Frank
Zappa Book) to the effect that "timbre is king." 

Please carry on. Most intriguing.

-- 
View this message in context: http://www.nabble.com/Modeled-piano--Pianoteq-tp19008978p19091616.html
Sent from the Csound - General mailing list archive at Nabble.com.