damper pedal off "crunch" harmonics

hmm, that's a good point.  I do have and use a continuously variable pedal.  however, when i let off the pedal -- no matter how slowly -- the strings just start sounding muted.  i don't hear any extra harmonics on the bass strings.

people who also play guitar might know what i'm talking about -- you can very lightly mute the bass strings near the nut and pick them and you'll get a bunch of harmonics depending on little adjustments to the position of the mute on the string.

Regarding key off velocity - this could be manually checked by entering the events manually with a mouse in any modern sequencer. No need for midi keyboard here.

I entered the notes - with respective release velocities: 1, 64, 127.

It seems the only effect that this has on pianoteq is the volume of the key release. I don't hear any evidence for modeling damping velocity of the strings... they seem to be damped at the same speed regardless of the key release velocity.

Even of Mordatt choose to model this in some future version, I don't see too many (affordable) keyboards out there that can produce such midi events.

Ethanay:

I haven't attempted to produce the "crunch" you described, but I agree with you that it isn't related to key velocity, but to damper pedal position and rate of application (how fast the pedal is applied or removed).

With my Roland KR7, I can strike a chord (staccato) with the pedal down, completely release the pedal and immediately re-apply it.  The notes will sound - although diminished in volume - much the way an acoustic piano would.  And this effect is more pronounced in the bass notes (as in an acoustic piano).

The aforementioned Rolands are not keyboards, but digital pianos (albeit with many other sounds built in).

When I look at the Event List of one of my recordings, I often see eight or ten consecutive Pedal Controls that vary with the position that my foot was in.  The Roland piano generates all values from zero to 127, so in effect it models the speed and position of the damper pedal (more values are generated if the pedal is applied/removed more slowly).

It seems to me that in order to have damping velocity of the strings, continuously generated pedal contol positions are required.  Values of zero, 64, and 127 might approximate velocity damping, but values of zero and 127 would in effect just be an ON/OFF switch (which wouldn't model the gradual action of a damper pedal any more than an ON/OFF switch will act as a dimmer switch on electric lights).  I believe that Doug pointed this out.

I've looked at literally hundreds of piano midi files from the internet, and don't recall seeing any that had values other than zero, 64 and 127, and most only have values of zero and 127 (even the midi files from the Classical Archives use 0/127).

I'm a bit confused.

In his original post, Ethanay refers to the damper pedal, which when let off slowly (after striking a combination of notes with the dampers up), creates a unique sound on his acoustic piano.  What he describes seems to be only affected by the damper pedal (midi controller number 64, which incidentally can have values from zero to 127).

My confusion arises as to how "key off velocity" will affect notes that have been struck and released, but are sustained by the dampers being up.  Can someone enlighten me?

A related question of mine is "does Pianoteq recognize and respond to all 128 steps of pedal control if the keyboard generates them (midi controller 64)"?

Julien:

Thanks for the clarification; I'm relieved to hear that PTQ will handle the subtle steps of Pedal Control (damper control).  I have already given up on EMU's Emulator X as it will not - in fact when used to render a midi file, EX makes my piano midi files sound very choppy so it is quite useless (it doesn't know which of the 128 Pedal Control values is ON and which is OFF).

Now some "rambling":  Do the piano sample systems recognize the 128 steps of Pedal Control?  It seems impractical that they could have a sample for every Velocity value, and thus must interpolate from a finite and reasonable number of loudness levels to keep the size down and be workable without a supercomputer.  (I realize that 128 steps IS finite, but to achieve this level of nuance would require far bigger sample libraries than they presently have and they are already huge).  This seems to me why the concept of Pianoteq will win over the sample based systems - even if Pianoteq is tripled in size to achieve more control over the bass notes to produce a "bigger" bass, it will be far more efficient and effective than sample based technology.

Sampled instruments utilize, for the most part, a typical subtractive synthesis architecture where an oscillator (or sample) is fed to an amplifier and a filter which are both controlled with envelopes or other waveforms.  The multiple velocity values are usually programmed to effect the amplifier and filter sections and may do so for the entire range of 0 to 127.  There may be 3 or more samples per note and these are cross-faded between each other at various velocity levels.  They are also effected by the amplifier and filter at each velocity level.  So there can be subtle differences at each velocity level even on sampled pianos, but the base oscillator -or sample- only changes a few times per note.  Virtual instruments based upon sample playback are not inherently sterile -in fact they can behave very organically- but they cannot compete with a modeling paradigm that regenerates at each note velocity a unique "modeled oscillator".  Then combine that with true sympathetic resonance based on real-world physics, and you have Pianoteq.  UIAM...