Better Piano Soundfont

"Ugly Baby Lizard" would be an awesome name for such an abomination of sound :-)

Alright... so comparing the two side by side in Polyphone to find the differences between us:

1) You sampled from a Salamander C-5 grand piano and I was sampling from a Yamaha Digital grand piano (YDP). There are differences in tone right there. Enough said about comparing apples and oranges. Let us distinguish between original instrument quality and soundfont craftsmanship quality now.

2) You had stereo samples to work with. I only had mono samples to work with. Using the exact same .wav sample twice for a left and right speaker does not make it stereo, that just makes it bloated. That's the grief that I had with the original YDP soundfont design that I made my own .SF2 version of it. So I used the panning feature on the instruments section to create that keyboard sound of bass on the left, treble on the right, middle C in the middle for the speakers. The same mono samples are played twice, but with two different sets of pans to create the stereo sound of L and R instruments when you're listening with a headset, the bass a little softer in your right ear than in your left, and gradually getting louder to your right ear as you climb towards middle C. You did not take advantage of panning at all, every stereo sound centered on middle C in your instrument design, relying on the mic pickup to make the stereo.

Only one note ever gets played though (mono on 2-3 hammered strings), and it's two different ears that are hearing it from two different points of view that sound would appear to travel from one side of your head to the other (stereo). Added math weighs nothing. Two mics weigh twice as much in the .wav samples as one. I chose math, you chose mics. If you were in the audience in a concert hall, stereo sound would make no difference to you at all, seated too far away from the piano to notice that the lowest bass is less than four feet further away than the highest treble that a mono version would be fine (What does the audience hear? YDP Mono). When you're seated at the piano, the travelling stereo sound is definitely noticeable (What does the pianist hear? YDP Stereo).

3) You used A and D# for every octave, those two notes getting stretched to cover the full span. I used A, C, D# and F# for every octave because I didn't like the way the two note stretch deformed the sound. I could afford that added weight because I wasn't doubling up on the same note for the stereo R and L .wav samples. You used 30 samples for the piano (15 notes) plus 14 samples for added harmonics. I used 30 samples for the piano (30 notes) and 0 added harmonics. I miss the harmonics, but missing harmonics is the truth of every digital piano I play. If you want to claim that the YDP is a lesser quality piano soundfont for the lack of added harmonic sounds, I'll agree with you there. But I would claim that the YDP is a higher quality soundfont for using 4 unique notes per octave than two.

4) Your .wav sample lengths are significantly longer than mine, and that's where the bulk of the Mb weight comes from. But I would disagree that longer .wav samples equal better. Degradation occurs naturally that an A4 root 69 needs key corrections once the note has been held beyond a certain point. If I'm going to loop a note to be held at that same volume level indefinitely, it's not going to be at the point where the note is already out of tune due to natural wavelength degradation, even if that faded timbre is true to the reality of a live piano. All of your looping .wav samples need key corrections. All of my looping .wav samples are clean.

5) If I was going to be playing music where the piano note was held for more than two full measures, or if the grand piano was playing a solo, I wouldn't be using a digital piano soundfont for that. I would want to hear the full resonance of a genuine grand piano and I would spend the limited Mb allowance on that, being the only soundfont I would need to add to the score. Concert hall solo piano music differs from rock 'n roll band music in so many different piano ways that we're comparing apples with a fruit basket now. But even in the pianos, you didn't do the math to make the Bright Piano and Honky Tonk piano out of the same C-5 .wav samples. Change the filter cutoff point in the instrument to trim out the bass tones and you'll make a C-5 Bright Piano for no added .wav sample cost. Double up on the instrument to make two tones for two piano strings on the same note (as a real piano is made), one string in tune and one string slightly out of tune to make the C-5 Honky Tonk piano for no added .wav sample cost. Or is that added versatility just a couple of "ugly baby lizards" to you, lol.

So in the renovated house analogy, my renovated house has 3 rooms (3 different midi instruments out of 1 set of .wav samples), and your renovated house has 1 room. I could use the original, unaltered YDP .wav samples if I want longer samples (up to 16s in length) to make a heavier Mb package to equal the C-5, and the instrument and preset designs of my YDP soundfont would be the same: Stereo, Mono, Bright and Honky Tonk. But I don't think the overall quality of the soundfont would be any better by adding more Mb weight in the .wav sample length. How often do you hear a piano note that been sustained for 16 seconds that you would notice the difference? That's four sustained measures on a tempo of 60, and eight sustained measures on a tempo of 120. I don't have any sheet music that sustains a piano note for that long that I made the .wav cutoff decisions that I did. Is there a particular need for a 16 foot ceiling when an 8 foot ceiling is the standard? ;)