Single wave file for voice storage versus one per note.

[Analogicus]

HOW,

Not sure what you mean by "indefinite pitch", but having pipes slightly out of tune with each other and sounding together adds a vibrancy which is very desirable if it is not overdone. As with all such matters, it can become very much a matter of individual taste. The standard soundfont protocols allow you to set the "fine tuning" of individual notes to achieve that result, but as the smallest "off-tune" available is usually one cent, this is not very useful for the notes at the top of the range, because it produces beat rates which are too fast to be pleasant. One workaround is to make samples for those high notes where you have detuned the pitch by less than one cent, and the Audacity Speed-Adjust function comes to the rescue for that. (I prefer that to the Pitch-Adjust function, as I seemed to get better results that way when I was learning to do these things. I find Audacity invaluable, and also Viena from the SynthFont website).

I should warn you that whether or not the beating together of two tones is a pleasant result, depends very much on the nature of those tones and their loudness relative to each other. I tend to avoid tuning differences for tones with the same basic pitch in case it sounds too much like a celeste effect. An exception would be reed stops, where low in the range I certainly detune them from flue stops of the same basic pitch. However I remember one instance where the 8' reed caused an unpleasant beating when used with the main 8' flue stop in the same department, but it only seemed to happen on one particular note, so on that one note I abandoned the reed detuning, making its pitch identical to the flue stop pitch.

Analogicus

 

[HOW]

Hello Analogicus,
I know indefinite pitch normally refers to the slight variation in pipe frequency depending on the air regulator but yes, I was using this as a "catchall" term so to speak and as you stated that is the great feature of the pipe organ. However, there is another feature that is seldom addressed that we call a traveling wave effect which is similar to the Doppler effect. If one uses an audio signal generator amplified and driving a speaker, the sound is very dead. However, if you walk towards the speaker, you will introduce the traveling wave effect and the sound becomes more alive. I believe the pipe of an organ produces this effect by the air flow causing the wave to travel in outward from the mouth (albeit the wave within the resonant pipe chamber is stationary for the most part.)
Anyway, I further experimented with the difference of using long recorded pipe samples of 1 per note compared to the MOS 1 Allen single period (32 words X 8 bits) which is the same waveform for each note but scanned or read at different rates. (The address points may be repeated for the lower notes which Allen called phase angle processing). The swell division was set up to play the same sounding Allen stops as the great division which was set up to play 30 second recordings or wave files. Other than variations in the recorded samples on certain pipes that might have been "out of spec" I was hard pressed to find any difference in the formant or harmonic structure from C1 to C6. I was using the Hlabs Traveling Wave Generator on the Allen voices to approximate a long recorded sample so far as the "indefinite pitch" is concerned. Keep in mind that I was only concerned with the particular voice identification parameters, not the variation in phase or what not. In other words, the tonal qualities that identify a particular voice were constant from the low notes to the high ones for both the simplistic single wave versus the actual recordings. From this I would conclude that the harmonic structure is constant (so far as the voice identification) throughout the division for both.
Best regards,
Stan