Interesting comparison; thank you for taking the time to compare recordings made with U4 and with a real Yamaha upright instrument.
Regarding the physical tuning of unison strings:
As a piano tuner, in trichords (usually from approximately note F3 upward), I first tune the middle string. Then in conjunction with the middle string, I strive to make one of the remaining strings resonate with the octave harmonics (even harmonics) and the remaining string emphasize usually the interval of a twelfth (third harmonic). This tends to add a singing quality to the instrument. Please note, however, if I am unable to emphasize the octave harmonics in the leftmost string (in conjunction with the middle string), I switch to making the rightmost string emphasize the octave harmonics, and vice versa. The general makeup of a real acoustic piano prevents me from guaranteeing that all of the octave harmonics are associated with the leftmost string and the center string.
Personally, I have come to believe this is what is meant by "deliberately de-tuning" of a piano by an experienced tuner. As a rather cruel experiment, I used an electronic tuning device to tune each of the three trichord strings independently of one another (to within a tolerance of 0.1 cent, according to a particular electronic tuning device's published specification -- cheap electronic tuners are only accurate to within 1.0 cent, so buyer beware). The not-so-shocking result was that there seemed to be no rhyme or reason between how the overtones worked themselves out -- except to state that a whole piano tuned this way (by an electronic tuning device of all strings, independently of one another) resulted in a tuning that sounded approximately in tune, but somehow weak, thin, and lifeless.
Ironically, I can walk into a given piano showroom and tell immediately when their "house piano tuner" simply went through each piano with an electronic tuning device: It happens way too often. These house tuners pride themselves in being able to tune a whole piano in less than 45 minutes. Their tunings might sound "in tune" at first, but upon critical listening, you can tell the tunings are thin and lifeless! (Think: Cheap electronic big box store 49-note keyboard.)
* * * * * * * *
Back to application with Pianoteq PRO: Here is one of the unsung applications/justifications of Pianoteq PRO. Often times, many Youtube recordings of professional pianists include some individual notes that the unison tunings are "not-quite-right" in certain notes. I am able to reproduce this effect in Pianoteq PRO by increasing the detuning slider of a very few individual notes to a value approaching 2.00. The effect gets ruined if you simply "randomize" the detuning, because nearly all of the notes get affected. Rather, only less than 5 notes per 88 notes of a piano should be treated this way. Of course, there is no way to predict which note's unison tuning will be this far off.
As an example, listen to Marney Laird's rendition of Brahms' Intermezzo in A Major Opus 118 No. 2 on Youtube. At first you may think the piano sounds "glorious" -- until you realize that many of the unisons are horribly out of tune. Generally, the better the professional performing artist, the piano is tuned better (often touched up during the recording sessions), such as those by Rubenstein or Glenn Gould.
Many individual velocity ranges within individual notes of high-priced Piano Sample libraries are horribly out of tune, with regards to unison width. The problem often seems to occur in certain note-on velocity ranges. This implies that the piano had to be serviced during the recording sessions, but the producer did not know or care about notes that slipped through to the final customer in certain velocity ranges. Thankfully, with modeling, Pianoteq does not suffer from this problem.
Enough of my rambling, and thank you for having read this far.
Cheers,
Joe
EDIT: Originally I was going to divulge the creator's name and commercial product name of a huge sample library of four famous piano brands, each consisting of from 12 to 18 sample layers per note, then repeated with soft pedal engaged, sustain pedal engaged and in "repetition" mode -- these were some of the most RAM intensive samples ever made into a commercial product. I am not here to complain in a public forum about a competitive product, because this multi-multi-Gigabyte library was extremely popular at one time. Of the four pianos, three of them were "unplayable" in the final version, due to individual notes that were in tune for most -- but not all -- of each velocity layer. I complained to Mr. X via telephone about my findings; his only reply consisted essentially of shrugged shoulders and the statement that "Well, the piano was re-tuned twice per day." Perhaps this was the truth, but certain notes remained horribly out of tune in the finished product. When I made demos for these instruments (back in the days before discovering Pianoteq), I recall having to record my live performances into midi, and then subsequently editing velocities of certain notes, so as to avoid several instances of out-of-tune notes.
END EDIT
Last edited by jcfelice88keys (25-09-2017 05:36)