Poll: Should the Pianoteq Harp include pre-pluck finger buzz noise?

Before even thinking about the sound, I must ask: How old is Alisa Sadikova in that performance? Six? Eight? She's astonishing.

I'm not sure that I hear a pre-attack buzz. Are there specific moments when it becomes more audible in those videos?

One thing I did notice about the sound is the plucky, nylony sound on the high, short strings, and how hard she had to pluck those strings to get a good amplitude. She either had a little trouble getting the right volume on the lowest strings or the mic was placed to catch the upper and middle strings.

Both players varied the attack point. Of course, on a harp, seeking the middle of the string means moving the hand to very different positions, since the  length of the strings varies so much, but at times they seemed to be intentionally not seeking the middle of the string, but instead seeking the buzzier sound of plucking off-center, controlling the timbre in the way that a guitarist can control timbre by exploiting the exposed length of string between the end of the fingerboard and the bridge. That seems to be an unavoidable limitation of any harp replication, either in samples or in modelling. The variation in attack point can't be assigned to velocity. Or could it be assigned to a pedal?

The beginning of the following video will perhaps make clearer to which noise I am referring: 
https://www.youtube.com/watch?v=MNkBaKb6IcI&t=32s
The piece begins with several repeated intervals.  When there is a long gap between the repeated notes, there is no audible buzz.  However, when the notes follow each other quickly there is a distinct buzzy sound before each pluck.

In the original two videos, the buzz should also be most prominent whenever there are loud, fast repeated notes.  For example, in Alisa's video at approximately 26s, and in Remy's video at approx 36s.

Well, wouldn't the damping buzz then would be on every note, instead of just repeated notes? But yes, the implementation is a problem. It would need to be a damping on the attack. The code would have to be something along the lines of "If amplitude of current string > x, damp before the note is played"?

Yep, "string-damping (between repeated notes)" of individual strings is exactly what pre-pluck finger buzz is.

Fortunately, the Pianoteq engine would not need clairvoyance!  It can work in the same way as legato/portamento transitions in string/woodwind/choir sample libraries: Only when the note is actually played (not before), do the engine decide whether a legato/portamento transition is required.  It plays this transition sample if needed, and *then* the note. 

Similarly, the Pianoteq engine can follow something like the rule given by Jake to decide whether or not to play a string-damping/finger buzz 'transition' sound before playing the actual note, and how loud and long this buzz should be.

If we consider that pre-pluck buzz sound for the harp is better implemented, as Piet wrote, as damping between notes, I think it may be currently simulated in a way by some kind of soft repetition, since in fast playing, a damped note may not sound much different than a softer repetition of the same. This is arguable of course, but the point is also that a keyboard is a much different interface than touching strings directly (as is also the case for guitar for example), and that it is difficult to think of a way to add randomly happening pre-buzz sounds without it becoming mechanical.

In any case, I tried something here, using the reverb pre-delay reflection (thus having a somewhat different tone than a straight repetition) to change to attack slightly so as to have more the impression of fingers damping strings at times. See what you think:

http://www.forum-pianoteq.com/uploads.php?id=2354

The problem is that the software can't guess in advance what notes you're going to play and when - implementing this pre-note buzz will unavoidably add a delay (latency) to the actual notes, if they're to be preceded by the buzz sound. However, when rendering midi data to disk, i would presume this can be done flawlessly, as far as i can see

/* edit */ .... at least with standard midi. I think the Yamaha Disklavier sends note positions for each movement of the keys, even before notes are played, similar to CC data - maybe if that becomes a dominant standard used by midi keyboards, this pre-note buzz thing could be done in realtime without adding latency. Unfortunately, standard midi doesn't send any note data before each note is to be played. Such keyboards would be pretty expensive due to types of sensors used tho.....

Don't think so.

I agree, that sound is essential to harp playing!  Plus it could be implemented in a very similar way to what's already happening with the piano, where the sound of an already-vibrating string getting struck with the hammer is different from the sound of a silent string.  Really, that's all this boils down to.  It would cause little-to-no delay and could be implemented as a slow, longer sound on note-off or as a quick sound immediately turning into a pluck when there is no note-off but just a repeated note.  I played the Harp demo and wished so much that I could hear those noises, but it didn't sound like an actual harp to me.  It sounded like someone sitting at a keyboard playing a harp sound.

My understanding is that Pianoteq models a new sound every time a CC or midi message is received, so there's always a degree of variability to the model.  Meaning every time the sustain pedal is pressed, for example, a new sound is created and we never get the so-called "machine gun" effect.  The same principal could easily apply to the harp with this harp sound where you can adjust the intensity and length, and adjust the randomness of those two factors, and the algorithm would decide the best standard setting based on the volume at which the string is vibrating.

IS the excessive buzz (like the one in a vídeo of a guy demosntrating i), something more noticeable in poor harps, something the manufactures would get ride if they could ?

Hmm, I didn't hear it when playing.

Unfortunately, Pianoteq's current implementation of string damping sound is not the same thing as pre-pluck finger buzz. 

As mentioned above, I have tried to hack this string damping sound and the "Damper noise" parameter to simulate pre-pluck finger buzz, but without success.