PART of me believes that a transducer on a sheet of plywood could do a credible job.
Other parts of me think that a baffled box behind the plywood can help.

MOST of me believes that piano sound boards are incredibly non-linear in their response to excitation by the vibrating strings - - and to a very large extent that is the "character" of wooden pianos anyway.

Hahh, all fun stuff to contemplate

...time to practice...

1)  I still haven't figured out the velocity vs. timbre balancing.  There are several potential controls here to balance:  Pianoteq Volume, Pianoteq Velocity, Pianoteq Dynamics, Pianoteq microphone levels, Pianoteq microphone distance from virtual soundboard, and stereo amplifier volume.  Plus other factors, such as having both Microphone 1 and Microphone 2 additive into each left and right stereo output (each of which drives the left versus the right soundboard transducer on my piano).

2)  I am still trying different transducers:  Tonight I started by gluing two small 'coin-type' transducers to the upper treble section of the soundboard, hoping that they could drive the high treble sound very convincingly without driving bass information, to add to the one that really drives the bass (but unfortunately it also drives treble information).  While that is the hope, with the PIanoteq microphone locations passing relatively pure bass and relatively pure treble information each, in order to simulate what the strings do to the soundboard on a real piano, I cannot find a place on the soundboard that really drives just the bass or just the treble.  That being said, I gave up on trying to get treble output to drive the right channel, and bass output to drive the left channel.  Instead, since both microphones present both treble and bass information, I decided to place a second Hidden Audio 401 more on the treble side of the soundboard, so now I have Channel 1 going to the left/bass-sided 401, and I have Channel 2 going to the right/treble-sided 401.  So far the treble is too loud, but maybe it's time to start using Note Edit to quiet the high treble keys, if I can't get the FXP gross controls to do enough.

More to do, more to do....

Nothing to add but applause for already great results from your dedicated and well documented work!

But your comments on my early-capping or flat-topping the velocity, and then going back to allowing velocity to expand fully go back to my balancing the Pianoteq velocity versus the amplifier volume to keep things under control.  The reason that I went back to a straight 1:1 velocity curve is that Pianoteq 8 didn't seem to want to freeze my velocity curve between different presets, despite me setting it that way on the "Snowflake" menu -- maybe a bug?  I haven't spent more time investigating it, but thats why I went back to default.

Yeah, that's my gripe with pianoteq GUI: it's beautiful but requires the manual, not easily discoverable. I wish they had opted for a "ugly" standard one, like (say) the one of a browser.

Anyway, what you search for the "global" velocity curve which you can enable by clicking that little G under the velocity curve windows.

I continue to applaud your determination, your achieved success and the even bigger accomplishments that feel like are just around the corner!!!

@Qexl - Ben, your 'moonshot' mics are clean, but a bit thin.  I really had to crank the mics to get convincing volumes, especially for mic1, the way I currently am hard-panning.  Part of the issue is that whenever I get the bass loud enough, then the treble is too loud, as my left transducer (bass) is large and strong, driving both bass and treble on the soundboard, and my right transducer (treble) is relatively small and delicate -- that's partly why I have been trying to separate the bass and the treble inputs that I feed them with, meaning trying to find soundboard positions that predominantly exude treble and predominantly exude bass -- because then after 'hearing' it with the Pianoteq mics, then I feed that right back in to my own soundboard via the transducers.

As far as external processors and DAWs go, I guess that I am trying to do the best that I can do from within Pianoteq first, before 'venturing out'.  As far as my desire to learn a DAW, I place that as a separate goal, and my interest in Ableton comes mostly from its looping construct and utilization of Session view versus Arrangement view - a separate issue.

But your comments on my early-capping or flat-topping the velocity, and then going back to allowing velocity to expand fully go back to my balancing the Pianoteq velocity versus the amplifier volume to keep things under control.  The reason that I went back to a straight 1:1 velocity curve is that Pianoteq 8 didn't seem to want to freeze my velocity curve between different presets, despite me setting it that way on the "Snowflake" menu -- maybe a bug?  I haven't spent more time investigating it, but thats why I went back to default.

Anyway, too much to chew on now, since it's time to shower for work.  And since I spend my weekends at my fiancée's house, I won't get back to work on this stuff in earnest until Sunday night.

Thanks for the continued support!

Glad to see such good progress - so happy for you! (for simplicity, posting here now instead of emails)

Observations of current vid..

Those hammers now sound louder in the Pianoteq ver to me (at least on the 1st vid above - if you also hear that, wonder how turning down 'hammer noise' a little might work out - or perhaps something else over time.. maybe in the room they sound more bassy when on vid they may seem more sharp/mid/treble) - but all else.. getting there.

///

Your Q1:

It's possible the smaller transducers may give some comfort in high velocity on treble side - but as far as mic placement here's what I noticed:

You have 2 figure 8 mics (with some modelled hardware) - and possibly.. I'm thinking they may give some benefit to the overall acoustics - and some flavor of Steinway B recordings.. but I'd definitely try this (moonshot mics! - and only thing missing is 'pure' unflavored non-brand-based cardio mics to use instead..):

They're not straight up/down vertical but angled to capture spread of tones without emulated mirror-like reflection straight back at the mic heads..

2 Different mics (in-Pianoteq testing bass vs trebs. plus stereo spread by ear - could this be better? For sure - all another moonshot)..

No 'delays' between L/R so no variant of any kinds of HAAS or other offset spatialization effects, which might give transducers some unnecessary clutter/cancellations)..

Using Cardio instead of Figure 8 (may introduce some built-in stereo issues with transducers?? not sure - what you've got seems nice on video) also not using omni as they pick up around/behind the target.. but thinking cardiods would point at 'roughly where soundboard' bass to treble ranges are - and your real soundboard could generate the 'around' aspect, if that makes sense (but indeed finding some long-distance thinking about this project has not all been as I'd expected - transducers on a piano are an amazing thing, have to say)..

First select mic 1 and change it to..

CMC6MK4

Then select the 5 lines below, copy, and then right-click the mic in the pane (not the selector box), click paste in the popup.

Click the mic's selector box, click "unlink mic" - (and I'd also try both proximity effect on/off when all done).

X = +0.688
Y = +2.058
Z = 1.329
Angle = +135.0
Vertical Angle = +317.4

Next, same with mic 2...

DPA4011

X = +1.058
Y = +1.058
Z = 1.616
Angle = +131.0
Vertical Angle = +290.1

Absolutely could be wrong - but that may hopefully be good to start on mics. (There's no way to share mic presets, like FXP Corner yet.)

///

Your Q2:

If this were using the U4, I'd suggest putting mics on the other side of it - but putting mics under the grands doesn't keep as much of your position vs a vis the soundboard.. the grand piano vs. upright..

I'm imagining the U4 player preset might be interesting to use for a non-Steinway B piano sound.

///

Your Q3: - specific to this quote..

Agree with dv - since noticing your early velocity curve was quite truncated at high velocity - it's prob the 1st thing I'd consider to limit high forces at higher velocities. But it may indeed cut into the tones, and if you like them now, that may change with a changed velocity curve too much.. but maybe not - so worth some experimentation.

But, what you described in the quote is a good description of a problem which might be suited to applying dynamic EQ - (if you've decided on using a DAW).

Something like F6 (Waves) for example - or now the included Pro EQ in Studio One is also excellent (I'm sure free ones and paid of all kinds exist - but F6 and Pro EQ3 are two I really really like. 

Dynamics-wise, set the EQ any way you like overall, and then for any frequency range, dial a threshold at which point that range can attenuate.

Essentially giving a limit to that range, so it won't get too hot of have too much force compared to surrounding ranges.

Extras to that..

a
Maybe instead of EQ, a multiband compressor could take some force out of target frequency bands well enough. (Again plenty around, the default one in Studio One is quite good).

b
Claro (by Sonoxx) recently hit me with it's "width" variations, and other things.. so you can make only certain frequency ranges "wider" or "narrower"). I think it would be 'the tool' I'd break out to manage stereo width as you describe issues with that inre what you're hearing.

Stereo-width-wise, in Claro, click 'width' and adjust a bass band wider (since your soundboard's bass waves would probably wash across the whole thing with more bloom beyond the left).. or narrower (since you may be able to that way contain much of the bass to the left side of the soundboard).. but in the stereo blend.. that may be also altered and attenuated in the Pianoteq 'mics' pane - I guess every thing like this, by ear comes down to what you hear there.

Beyond that also in Claro, is the ability to right click any dot(s) for any frequency, select 'split' and to individually raise/lower either/both left/right channels independently.. which might be excellent to remove bass or other ranging frequencies from taking up space on the 'wrong side' of the soundboard/keyboard as you hear those in situ.

///

I'm prob commandeered, tasks/partner for the next.. numbers? of days - but whew - so good to see - and all this wonderful help - hats off to all.

Overall David..

What you've come up with sounds remarkable in the video - Happy for you!

I also felt that double-take when you 2nd guessed real? or Pianoteq? there, I had to replay the vid

[

hope to not have forgotten something or mixed up anything in mixing together different text notes from different times - also here are links to EQs and Studio One in case thinking of a DAW.. but for sure, Reaper, Cakewalk and on and on might be just as good of a fit.. (Including Studio One here because its internal new dynamic EQ is competitive with the Waves F6 and because I use it most - love the ease of workflow with the same kinds of results from any old DAW in the past.. it only misses out on actually 2 things I'd like it to do but in those cases I load up Cakewalk - but again.. plenty of good DAWs to see out there.

Maybe this will be overkill for most starting out into DAW land, but I also like Presonus Sphere for a whole lot of reasons, just like I love Pianoteq Studio Bundle. No having to find "Oh, I don't have that tool" - but anyway, there are links to Studio One 'Artist', 'Professional' and 'Sphere' and a comparison pdf if some things are desired - if this DAW seems good to you. I'd add, the Artist version does not have included multi-band compression - but it does have the included multiband Pro EQ

https://www.waves.com/plugins/f6-floati...echain-tip

https://shop.presonus.com/Studio-One-6-Artist

https://shop.presonus.com/Studio-One-6-Professional

https://shop.presonus.com/PreSonus-Sphere

https://pae-web.presonusmusic.com/uploa...rsions.pdf

https://www.sonnox.com/toolbox/claro

].

BIG PROGRESS Today!

Wow, that is really great progress. With my speakers I can't tell the difference between acoustic and PianoTeq!!

Congratulations and thanks for sharing the details. Now, additional thoughts from me.

Have you tried playing some pieces and see if such great results are confirmed in that case?

For high velocities, why don't you simple change the velocity curve, making it less steep and hence not reaching the highest level? If I recall correctly, doing so affects both volume and timbre, but then you can change the dynamics slider (if necessary) which controls only the volume but not the timbre: the combinations of the two should allow you to avoid the problem at the highest velocities, assuming is a timbre problem. From my experiments, I'd say make sure it's not a volume problem, since my exciters really started distorting at higher volume levels. In other words: lower the volume of the amplifier you are using, and see if the problem persists (timbre is the issue) or goes away (volume is the issue), since you need to address them in vastly different ways!

For "it sounds like you are on the wrong side of the piano" have you actually tried to sit on its back when playing a MIDI (e.g. its demo) and it sounds more realist from that point of view?

Cheers!

I will post tonight's Acoustic vs. Pianoteq comparison video first, and then explain how I got to this point next:

     https://photos.app.goo.gl/uK3yfcKDUDayDtuZ8

So this is a series of notes from treble to bass, played back and forth between my acoustic 1875 Steinway upright Model F (which is in need of some serious tuning) as compared to my Pianoteq Model B .FXP that I have been modifying (tonight I also bumped Condition to 0.31 to make it closer to the acoustic in out-of-tuneness).

Several things got improvements from two days ago until today:

1)  I set up an equalizer curve based on white noise balancing:  I used a white noise video from YouTube, and then used a non-ASIO equalizer program that Gilles had suggested ( FXSound ) to make the tracing look (and sound) relatively flat on FrequenSee:

https://photos.app.goo.gl/wUFc9YAScELhrgXAA   As flat as I could get the white noise curve with FXSound.

https://photos.app.goo.gl/DJjYj4LoW5wrXddR7     The equalizer graph after I was done with FXSound.

https://photos.app.goo.gl/8daak4DePJ7AZZeCA    The equalizer curve as essentially copied into Pianoteq.

2)  I made lots of changes with the Pianoteq microphones: 

https://photos.app.goo.gl/m8GTr56E13g7tPbi7

I was hoping to find places over the soundboard to get bass to prevail for the left mic, and treble to prevail for the right mic, but I am frustrated that treble is predominant on the right, but bass is as well.  So I spread the microphones, and kept changing their relative volumes to try to get the cleanest separation possible (as close to hard-panned as possible) since they're just going to get recombined on my piano's soundboard when 'played'.  Also, because I have a smaller lightweight transducer on the treble side, and a heavier transducer for the bass side, I was trying to drive them differently.  I made Output1 to be only from the left mic, and Output2 to be only from the right mic.  I turned off the Level and Delay Compensation (it just seemed easier to get better useful sound differences that way) and turned off the proximity effect for the microphones as the percussive sound of the hammers to the keys seemed to be reproduced better that way.  I also listened to each the bass and the treble, testing all of the microphones, and comparing them to my acoustic piano.  I settled on both microphones being Fig-8 variants, with the C-414 for the left-sided and more bass-driven mic, and the U87 for the right-sided more treble-driven mic.  I also dropped the mics really low to the soundboard, as the percussive hammer and key sound was better, though any lower produced more overwhelming mechanical sounds.

=======================

Overall, I am super-pleased with the progress, and occasionally even fooled myself, especially in the bass, on telling which was acoustic and which was Pianoteq as I went back and forth.  So, where am I now, and what is left to do?

1)  My new transducers are due in soon.  I want to try very small transducers on the treble side.  I wish that I knew how to apply filters to the Pianoteq mics so I could get the left mic to pick up predominantly the bass, and the right mic to predominantly pick up primarily the treble -- any ideas here?

2)  When I am happy with my choice of transducers and their location, I want to move them to the front side of the soundboard.  Currently, playing on the back of the soundboard, they make me feel like I'm sitting on the wrong side of the piano!

3)  Currently the Pianoteq preset is working quite well at low to medium velocities - the sound vastly changes at high velocities, sounding like there's much more bloom.  As far as I know, the EQ curves can't be applied differently for higher versus lower velocities.  If I could make changes with velocity levels, that would be great.

4)  I'm sure that there are many more Pianoteq Soundboard and Note Edit characteristics to change.

5)  I haven't yet even tried to damped the acoustic harp yet.

6)  Also, I'm relatively new at shooting videos like this with my iPhone.  I note that I had it sitting on its microphone at the right edge of the keyboard, so maybe even just putting it upside down would make the sound less colored (since now the microphone is 'trapped' against the body of the piano).  Also, at the end of the video I said "white balance" rather than "white noise" (sorry - can't get the photographer out of me!).

That's all for Thursday night!  Thanks, guys, for all your help with my project.

David

Nick

Yes, I have no recording training or experience.  From my naïve logic, it seems that if you are trying to reproduce the imaged sound of a piano with speakers placed four feet apart, that you want to record it with microphones placed four feet apart (at essentially the same location that the speakers will be in).  Perhaps overly simplistic an analogy, but it you want to see what a piano looks like to a person who will be standing across the room, take a picture with a camera positioned where that person's head would be, and then print the image - then you would see what they would see.

Furthermore, the vibrations produced by the treble strings on the right and the bass strings on the left (temporarily assuming a straight-strung instrument, like an older Pleyel) are 'mixed' throughout the piano by vibrations carried on the soundboard, so why 're-mix' left and right microphones into left and right channels?

When copying a completed recording, then hard L/R panning should be used, recording should I think have some mixing between channels. My knowledge on this subject is quite basic but a friend of mine corrected my view that in recording, hard panning is not the way to do it.

Nick

Anyway, just got home, found DRC Designer, but now need to decide on an unpacker since I don't have anything on this computer to decompress .tar files.

There is a link for a .exe Windows file on the DRC Designer site, the .tar format is for Linux. But yes, I would suggest staying with Pianoteq's EQ3, maybe also using the global Equalizer.

Last night I left the EQ thoughts and distracted myself by playing with microphones.  Two thoughts, and I apologize for not making a recording this morning, but I'm out of time:

1)  I understand that mixing Mic1 and Mic2 (and 3 and 4 and 5) into each channel produces pleasing results, but why do the Pianoteq Player presets also have mics mixed into the outputs? (presuming Output1 is Left channel and Output2 is Right channel)  Since the whole soundboard vibrates, some bass sound is heard on the right and some treble sound its heard on the left anyway, so why should both Mic1 and Mic2 each be combined into Output1 and Output2?  I would think theoretically that to make a realistic Player preset, the Right side of the piano should play into the Right microphone, which should then play out of the Right speaker.  This is not the way that Pianoteq Player presets are constructed.

2)  While I tried to set up the Right => Right => Right conditions as described above in #1, I placed cardioid microphones x 2 over the Model B soundboard, pointed down, just a few inches up from the soundboard.  When auditioning each Right and Left, I played both treble and bass, to find positions for the Right that emphasized treble output and positions for the Left that emphasized bass output.  Curiously, I got more bass output on the Right side of the soundboard, under where the high treble strings are.  Why would that be the case?  I would certainly have expected higher bass signal immediately adjacent to the bass strings.