Topic: NY Steinway D Custom Preset, Lessons Learned In 2023

Early AM 1-7-2024, I'm excited to have uploaded a preset I've been working on for some time for NY Steinway D in 96kHz with increased dynamics and low/no effects https://forum.modartt.com/file/8enx43p4. Please, if you can, check out the fxp and post here if you'd like to comment. Honestly, I deleted an mp3 I posted because something made it sound boomy and dull. Feedback (and even freeform commenting) is welcome, absolutely. But also, if you have a specific thing that might be controversial or something, just ask before getting deep into it here? Thank you.

But First, My Preset-Making Techniques, In (Very) Brief

A simplified version of my process for making a Pianoteq preset might go something like: 1. Pick something I like that's already partly implemented in a preset. 2. Try to make a new preset that celebrates that thing I like. (Step 2 is where this post is focused.)

Step 1 choices evolve with experience. What I hear and try for now is more nuanced than what I tried for a year ago. Step 2 incorporates the experience needed for Step 1 and applies it to shaping the timbre. Timbre is definable in terms of combined sounds, both over time and in layers, and it can vary with performance. Since shaping timbre is complicated by complex interaction of various preset settings, both multi-step and simpler single-setting techniques are used.

After selecting an adjustment to make, chose one or more settings to adjust. For example, I might want to increase the perceived loudness of the thunk at forte note attack (attack thunk). (Goals can be more or less specific than that, but less specific goals, say for example, a full-sounding timbre, may require complex testing of many settings.) To continue the example, the attack-thunk loudness is perhaps most directly tied to Hammer noise, Hammer hardness at the level of loudness that you are playing, Volume, and Direct sound duration. Then, having selected a group of settings, change them until roughly at goal (or beyond it) and then check the result. After checking if the settings have been taken beyond the goal, just document the results and try again. An attempt may only be complete after trying the settings two or more different ways, or by expanding the list of settings used. Then, if other things were negatively affected by the intended changes, consider expanding the goal to involve fixing up those things.

"Tuning"/"Voicing"/changing "Design"/changing "Action" involves setting individual note level changes ("note edits") based on comparisons from at least a few keys away (outside-in), comparing neighbor notes (inside-out), as well as important melodic and chord intervals at various loudnesses with and without pedaling. Note editing at the note and in patterns of notes involves checking for adequate similarity and difference, and including as transitions between neighbor notes (and transitions of loudness, articulation, etc).

Sometimes, for certain settings, tuning may involve changing global- or key-range levels. This may involve sliders or using multiple "control notes" (blue squares) in note edit windows to restrict changes to a particular note range. Generally, after bulk editing note levels, check and correct each individual note as needed.

Last edited by bani223 (07-01-2024 08:27)
Soundblaster ZXR, ASIO4ALL. 96khz, ~2ms buffer. Little to no pop/crackle on Realtime priority.
I have posted several times about tweaking Pianoteq

Re: NY Steinway D Custom Preset, Lessons Learned In 2023

Updated to version am.5 on 1/13/2023. Decreasing "fxgain," the preset is more compatible with compression (e.g. mp3) as well as lower dynamic range media. Quick spectrum profile changes to e3 and f3 for a subtle-on-speakers ringing harmonic. https://forum.modartt.com/file/4miso82n.

Sharing another page from my website below:

How to analyze tones for directed preset editing/creation

Timbre describes the textural qualities of a musical sound. All aspects (or sound-parts) of perceived timbre are tied to both (1) the piano's mechanisms and (2) the environment in which sound waves are reflected, resonated and transferred. Pianoteq, modelling a physical piano's tuning and repair, calculates waveforms that would result from hammer noise and string vibrations, the sound board and other piano parts, room/environment, mics and their mixing, and effects processing. As the models address the physical world in ways to affect perception, users of the software may be inclined to think of the settings more or less directly linked to specific perceptual counterparts, e.g. Volume to loudness, Hammer hardness to hardness of sound, etc. Since, however, many of these things interact in complex and sometimes contradictory ways when producing the timbre, a shorthand for the sound's parts (sound-parts) themselves is helpful for virtual (and real-world) piano tuners. Particularly for seeking specific piano sounds (troubleshooting, goals), an optimized schema may best define sound-parts based on how they are perceived rather than how they are created, specifically in terms of: (A) noise (and noisy tone spectrum variations) or its absence, as noises are both the most obvious problems and the most important source of liveliness in a piano, (B) the dynamics with which the note is played (e.g. forte), which is a primary aspect of a player's control over the piano in real time, and (C) time during the sounding of the note when its audible, note duration (and sound masking and emphasis by playing other notes) being another primary aspect of the player's control. (D) The results of pedaling and other manipulations is added for completeness.

A. Noises
- Shaping and regulating applies to both noise and tone (non-noise) sounds.
- Incorporated/regulated noises include "lively" and/or "harsh" noises, such as rings, whines, buzzes, rattles, and clunks. 
-- Each audible sound-component, can be thought of as having a greater or lesser additive effect on loudness, in the form of sound wave amplitudes, but this is particularly true of noises. Even a subtle noise can make a sound seem loud.
-- Noises are often carefully controlled by piano tuners.
-- However, at ff it is allowable, and perhaps ideal, to have some random-enough and audible-enough noises, to make this extreme loudness sound lively.
-- At other loudnesses, low- to mid-pitched notes (and some high-pitched notes) played at other loudnesses may have a brightening or subtle ring or buzz related to the nature of metal string components.

B. Dynamics
- Dynamic response -- playable difference in loudnesses and tone quality -- is typically regulated across the full piano keboard, with some exceptions.
-- Loudness (and especially the Volume setting) is perhaps most important when it comes to regulating response to dynamic play
-- When a note is played at forte levels, often tone- and noise-components defined by both forte and mezzo-- and some piano -- hammer hardness settings may be audible.
-- Subtle Bass note edits to piano hammer hardness can sharpen duller tones when playing softly.

C. Time And The Stages Of Evolution (Of Noise And Tone at Various Loudnesses).
- Components of timbre (sound-parts) can be said to apply at one or more of 5 rough temporal divisions of the note.

  1. Attack, immediately following MIDI note-on/key press, which may have a rounded or metallic tone, and/or either a thunk or a subtle noise.

  2. A resonant/reverberant/bloom-like kind of response to the attack and the tone, that may be bassy, boomy, glassy, rounded, or subtle. (I do not usually adjust Pianoteq's Blooming settings-- these words are used to descriptive the sound itself, not to refer to the settings.)

  3. The first part of the sustained tone, when the tone sounds fullest, including beats (sound wave overlaps that introduce new tones) as well as and other noises, which will vary depending on unison and, for chords, temperament tuning.

  4. The second part of the sustain, where the first part of the sustained tone fades and another, simpler, part remains, as well as certain beats and noises.

  5. Decay, following the MIDI note-off/key release, in which the note is dampened and reverb fades.

D. Pedaling and Results of Other Manipulations
- Additional manipulations, such as use of the sustain pedal or effects, may emphasize new sound parts issues and deemphasize sounds mentioned in A, B, or C.

(See also my list of noises and sounds to shape and to avoid).

Last edited by bani223 (13-01-2024 18:33)
Soundblaster ZXR, ASIO4ALL. 96khz, ~2ms buffer. Little to no pop/crackle on Realtime priority.
I have posted several times about tweaking Pianoteq

Re: NY Steinway D Custom Preset, Lessons Learned In 2023

11 Selected Sound Issues To Avoid -- And How To Fix Them:

Although I wrote this working on custom presets for the NY SD in 8.1.3 and we are currently on 8.2, in which noise (especially buzz and rattle type noises) is more controlled on the back end, these tips are I hope general enough to still be useful.

Outline:
1. Sound is too loud in one of the stereo channels
2. A prominent hammer thunk, clunk, or thump, often corky, sometimes with very subtle high frequency bits.
3. A hollow glassy, possibly boomy, sound at attack and continuing through sustain.
4. A noisy metallic/tinny resonant often-swelling shimmer at forte with a subtle rattle as sustain begins.
5. A simple single frequency is emphasized instead of the full spectrum of the tone. Sometimes 2 or 3 frequencies may be involved.
6. A buzz or rattle is emphasized instead of the tone.
7. Tone is too metallic and thin, or tone is irregular with metallic or thin aspects
8. High note with a slow prominent beat without buzz. May be echoey.
9. Tone is boomy and bassy, or tone is irregular with boomy/bassy aspects
10. A reverberant sound that persists after note release, especially at forte staccato. It may only last briefly after note release and may be metallic.
11. Loudness of a sequence of key presses is too similar



Imbalance issues audible at attack and throughout

1. Sound is too loud in one of the stereo channels
-- What's not to like? Two things: 1. If listening in stereo, it will sound like one of your speakers is weaker than the other. 2. You may be missing or over-emphasizing critical components of the sound field, like bass tones.
-- Why? If you've created custom microphone placements, mics may be missing the right or left side of the piano. Lower Stereo Width may mask this issue, or an left-right-unbalanced pair of headphones or speakers. Alternatively, it may be because one side of the piano was set to be louder than the other. Personally, I most recently encountered this problem because I had cut down on bass tones. My headphones were not very good for very low tones, so for headphone configurations I would increase the bass by placing a microphone under the front left of the piano. On my speakers, when trying to make a "neutral" sound uncolored by the imperfections of my headphones, I was actually still dealing with bass imbalance. The subwoofer was too close to an unpadded wall, and therefore many left-oriented piano sounds auditioned louder than they should have been. After correcting the subwoofer configuration and changing microphone positions and levels, I went through spectrum profiles and was able to fill out the tone for many notes, not just the bass notes. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? Although in some cases you may want to pan your sound to one side or another, I think a more versatile output would have sounds from both sides represented roughly equally.


Noises starting at attack

2. A prominent hammer thunk, clunk, or thump, often corky, sometimes with very subtle high frequency bits. It can sometimes reverberate into the sustained tone.
-- What's not to like? It can sound a too rough for traditional classical performance and many other styles. An exception might be certain acoustic rock performances, or perhaps some old-timey styles.
-- Why? "Hammer noise" or a related setting probably too high, perhaps Forte or Mezzo Hammer hardness. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- What to look for? Sometimes, enough of that already subtle sound will come through (or maybe just be imitated) without much help from the Hammer noise setting.
-- Ideal result? For me a subtle hammer noise that does not reverberate is usually best.

3. A hollow glassy, possibly boomy, sound at attack and continuing through sustain.
-- What's not to like? If it's not to your taste, then it might be because (1) the attack is too prominent and/or (2) the echo-y reflected sound during sustain is more prominent than the than the sound off the soundboard or wherever you'd normally hear it coming from.
-- Why? Volume or one or more related settings (especially things like Direct sound duration, but also Impedance, Damper duration and anything set high or low enough to significantly contribute to loudness) is probably too prominent for that note. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? It's subjective, but a more-subtle attack and a more-subtle reverb during the sustain (and after) are probably preferred.


Noises often first audible immediately after attack

4. A noisy metallic/tinny resonant often-swelling shimmer at forte with a subtle rattle as sustain begins.
-- What's not to like? If it's not to your taste, then it might be because the sustain is not smooth enough, or has too much thin string resonance-like sound, or it resembles the beginning of a feedback loop.
-- Why? Various and often multiple reasons, but when starting immediately after attack, or even during, consider Mezzo Hammer Hardness played Forte. Metallic sound can also be the result of a subtle brightness made by layers of sounds and noises, as well as (a) loudness- (b) resonance- and/or (c) noise-related settings. Consider temporarily erring on the side of too much of this kind of noise while exploring the sound range of an instrument. Then, if desired, decrease overly bright harmonics-- or if you went the other way, restore brightness. Consider doing fixes in spectrum profile among other settings. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? It's subjective, but noises during the sustain should usually be subtle, yet complex and irregular -- i.e. sometimes at forte or fortissimo there should be clearly audible metallic noise, but a metallic resonance during sustain should usually be a minimally-perceptible independent sound. May be needed for fullness and brightness.


Noises often first audible during the first of two sustain periods (if roughly dividing the tone).

5. A simple single frequency is emphasized instead of the full spectrum of the tone. Sometimes 2 or 3 frequencies may be involved.
-- What's not to like? Because the fundamental and/or typical overtone series are partly masked by noises like this, it gives the tone either a weakness or a loud distraction.
-- Why? Spectrum profile is the micro-surgical tool for issues like this. You may be able to remove a tone like this by finding the right frequency within that note's tone spectrum. Unfortunately, literally any other setting that might over emphasize that frequency can also be involved. I try to avoid EQs and effects other than reverb, which may affect tone brightness in several ways.
-- Ideal result? No prominent single frequency distracts from the sound.

6. A buzz or rattle is emphasized instead of the tone.
-- What's not to like? Noises, while an integrated part of many mechanical sounds, are not always welcome, or distract from the tone, either making it sound weak, or making the whole thing sound loud and rude. You can't make friends out of salad.
-- Why? Possibly the result of extremes of settings both modelled within Pianoteq, in the audio production from your computer and speakers, or in reflection and resonance in your environment. Since an extreme is a contextual thing, it might be a level difference as small as 1.05 versus 1.00 that makes the difference, depending on the particular setting you are adjusting and where other settings need to be. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? No prominent buzz or rattle distracts from the desired sound.

7. Tone is too metallic and thin, or tone is irregular with metallic or thin aspects
-- What's not to like? Tinny sounds are subjectively "off" because they sound like you are using a very low-bandwidth data stream or have speakers like an AM radio. More objectively, tinny sounds are often "off" because they emphasize mid-high tones while ignoring bass and/or overtones.
-- Why? Many things can cause this and the solution may be either complicated or straightforward (or a bit of both, as I'll explain briefly). Causes, for example, include higher-end consumer headphones which selectively emphasize some high frequencies tones in sounds and may not fully represent bass tones, which can lead to over-compensation, in the form of increasing bass loudness and dampening normal, desired noises. Solutions, unfortunately, can cross the gamut of settings, but for me, a focused process probably includes balancing Unison width, Direct sound duration, Piano, Mezzo, and Forte Hammer Hardness, Spectrum Profile, Hammer Noise, Impedance, Sympathetic Resonance, Duplex Scale, Velocity Response, Microphones type(s), placements, and levels, Volume, Damper Duration, Damper Noise, Key Release Noise, Pedal Noise, Reverb Mix, Duration, Room size and Tone, and, outside of Pianoteq, loudness and balance of audio output and speakers and headphones. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? A full tone that represents the vibration of the string plus accompanying mechanical noises, without significant interference from over-emphasized thunks or clicks at attack, boominess, evolving whines or simple beats in the body of the tone, or echoes, rattles, or buzzes during or after. There should be subtle but audible transitions between each part of the tone, from attack to the first part of the sustain, to the second part, to note release and after, and the evolution of the tone should be a flow, rather than a mosaic.


Noises often first audible during the second of two periods of the tone

8. High note with a slow prominent beat without buzz. May be echoey.
-- What's not to like? Because the fundamental and/or typical overtone series are partly masked by noises like this, it gives the tone a weakness. With a good attack (a strong kick), a high note can sound icy, which can be good, but with the waivering, it's like ice melting.
-- Why? Sympathetic resonance for that note may be too high. Sympathetic resonance-related items, like Duplex Scale, Damper Duration or Impedance may be appropriate to check after decreasing Sympathetic resonance. Other settings may be involved (i.e. anything set high or low enough to significantly contribute to reflected or reverberant sound may contribute). I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? Typically, subtle beats are normal and desired, and for low to mid notes, a subtle buzz is desired.

9. Tone is boomy and bassy, or tone is irregular with boomy/bassy aspects
-- What's not to like? Because the fundamental and/or typical overtone series are partly masked by noises like this, it gives the tone a weakness, leaving me wanting more fullness. With a good attack (a strong kick), a bass note can sound boomy, which can be good, but in excess or out of place it will sound like something is broken or otherwise muting high frequencies.
-- Why: a fundamental that is emphasized over the other harmonics, even subtly may cause this, and for bass notes, consider the first 2 or 3 harmonics. In general it's anything that can emphasize the first harmonic, though, and perhaps not surprisingly, this involves all my favorite settings: Unison width, Direct sound duration, Piano, Mezzo, and Forte Hammer Hardness, Spectrum Profile, Hammer Noise, Impedance, Sympathetic Resonance, Duplex Scale, Velocity Response, Microphones type(s), placements, and levels, Volume, Damper Duration, Damper Noise, Key Release Noise, Pedal Noise, Reverb Mix, Duration, Room size and Tone, and, outside of Pianoteq, loudness and balance of audio output and speakers and headphones. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? A full tone that represents the vibration of the string plus accompanying mechanical noises, without significant interference from over-emphasized thunks or clicks at attack, boominess, evolving whines or simple beats in the body of the tone, or echoes, rattles, or buzzes during or after. There should be subtle but audible transitions between each part of the tone, from attack to the first part of the sustain, to the second part, to note release and after, and the evolution of the tone should be a flow, rather than a mosaic.


Noises often first audible after key release

10. A reverberant sound that persists after note release, especially at forte staccato. It may only last briefly after note release and may be metallic.
-- What's not to like? If it's not to your taste, then it might be because it persists too long, or it's not long enough, or because the tone is too metallic, makes the next note sound too muddy, or is itself not smooth enough.
-- Why? Various and often multiple reasons, but after note release, consider Damper Duration and Damper Noise, as well as Impedance and Sympathetic Resonance and Duplex Scale. Like loudness and fundamental (zeroth overtone/first harmonic) issues, resonant/reverberant sound can also be the result of (and/or masked by) any setting that is high or low enough. I try to avoid EQs and effects other than reverb, which may also affect tone brightness in several ways.
-- Ideal result? It's subjective, but sounds following key release should generally be subtle (except for undamped high notes) and consistent in length, but responsive to loudness and varied in texture. The tail should only be longer or prominent if trying to emphasize reverberation, e.g. in a large space, and/or resonance. It should not be so short that it sounds stopped short when not playing short notes or staccato.


11. Loudness of a sequence of key presses is too similar
-- What's not to like? Play sounds overly controlled if velocities are repeated too often.
-- Why? The MIDI standard allows for 127 levels of velocity precision. If a velocity curve is too horizontal, or if more than a few notes are played together or in a row they may end up being played at the exact same velocity, which can sound oddly robotic.
-- Ideal result? Although a completely horizontal velocity curve segment can cause play at different loudness levels to sound the same, unfortunately, the primary fix here is to change MIDI controller equipment settings, which, again unfortunately, are not available on all controllers. To create a more lively sound, I use High resolution (14 bit) note on velocity, i.e. continuous controller 88 (CC#88) as velocity prefix to the standard Note On velocity signal, which is available on Casio stage pianos and some Yamaha console keyboards and I believe several other controllers from other manufacturers. It has been somewhat maligned (unfairly if you ask me) in the Pianoteq forums, and it may not be available on your MIDI controller. Alternatives include Yamaha's Disklavier XP and, perhaps soon, Midi 2.0 (though I'm not sure what equipment supports this now). You can get similar sounding but less accurate results in DAWs by adding a filter to alter CC#88 randomly while playing.

Last edited by bani223 (21-01-2024 18:24)
Soundblaster ZXR, ASIO4ALL. 96khz, ~2ms buffer. Little to no pop/crackle on Realtime priority.
I have posted several times about tweaking Pianoteq