In both Dorico 4 and 5 manuals, at Notation reference > Dynamics > Dynamics in playback > Playback Options for dynamics, is the statement:
A dynamic curve power of 1 creates a straight line, producing a steady dynamic increase. The difference between pppp and pp is the same as between p and mf.
This is not borne out in either of the two test examples I’ve submitted, one for a sustaining instrument, the other for a non-sustaining instrument, both written in Dorico 4. The difference between p and mp, and between mf and f, are each half of the other differences. I did not find that this had been addressed in the list of Dorico 5 updates. If the Dorico team agrees that this is an error, I’ll leave it at that. If not, I can expound on why it is.
Daniel’s reply in the other thread might be a clue. He said the difference between mf and f is only half a dynamic level, which is an idea I had never heard before. In my decades of music making, all dynamic levels including mezzos are treated as equidistant.
I’ve certainly never heard that either! Are there any modern (post-1900) orchestration, performance practice, or notation books that claim this? I certainly wasn’t taught that, and don’t teach it myself as I’ve never heard of it before in modern practice. What about the serial composers that also serialized dynamics - is there any evidence they treated mf to f as anything other than an entire dynamic level?
Regardless of the theoretical correctness of the notion, it’s correct that Dorico treats mp and mf as “half” dynamic levels. Dynamics are represented on an abstract scale from +6 (ffffff) to -6 (pppppp). mp is -0.5, while mf is 0.5. Even more mysterious is the default dynamic, which is level 0, which is itself halfway between mp and mf. What should we call that?!
Level 0 can be called mh. That’s a contraction of “meh…”
I’m fine with mf having a level of 0.5. But that means f is 1.5, ff is 2.5, p is -1.5, etc. While any span between single dynamic point levels is still an integer, the distance from 0 is necessarily an integer plus a half. So +6 cannot represent ffffff, if the 6 is being used to calculate anything, unless an extra 0.5 is included. Although unpleasingly unsymmetric, defining either mp or mf to be the 0 level allows integer definitions of all single dynamic points. I would also be fine with accepting the current scale where 0 is between mp and mf, but with the understanding that the input text “+6” means the sixth dynamic louder from mf, which would be ffffff. Perhaps all of this can be sidestepped by replacing the + and - numbers with their corresponding dynamic texts.
My concern, however, still stands, as the progression from pp… through ff… is not smooth.
Doesn’t every elementary school band or orchestra director get asked that in week 6 or so? LOL! None of the method books have dynamics until then, and then the kids get f and p, then the next week get mf and mp, followed by the question, “why isn’t there just a medium level?” (That’s a good prompt as an intro to interpretation even if they don’t realize it.)
Shouldn’t a curve power of 1 actually give a linear progression as pictured in the Playback Options diagram though? Even uncommon modern innovations like this one on Read pg 257 still treat mp to mf as a whole dynamic level.
I’d have to research it, but I’d be surprised if any composers using serial dynamics were assuming mp to mf was anything less than a full dynamic level. I’m not sure I’ve ever encountered anything stating otherwise.
If 0 is the default dynamic, which is only played when no dynamic is yet notated, this implies there is no way to specify the dynamic level=0 after any other dynamic has been specified, ie., How is it possible to notate back to the default, level=0 ?
Nothing is simple in the musical universe. And there are no absolutes. Some composers regard mp and mf as the same level, and refuse to use one or other, restricting themselves to one notated dynamic between p and f. For example, the colleague that I engrave modernist scores for:
There certainly are some alternative practices. Tchaikovsky used fff to mean “fortissimo” so that ff could be “più forte”, and the same with pianos. Not to mention certain 20th century composers using up to 8, which I loathe. (If you need to be that specific, you might as well use numerals.)
I too learned the idea of 10 dynamic levels, but from 4 to 4, so including mp. Omitting mp I like to think of as “Beethoven mode”. (Not that it is impossible to find mp in any of his music, but most often he did skip it.)
None of this addresses questions of perceived loudness. Do people feel the smaller increments in the middle of Dorico’s dynamic markings sound linear?
Edit 2: Carl Czerny appears to be defining the dynamic ranges for the piano as: pp, p, mezza voce, f, ff and says that the mastery of the transitional range in piano playing is the distinction of skilled artists (Czerny “Complete Theoretical and Practical Piano Forte School”).
I looked through a few books I have, and was surprised how little this is actually discussed. Peter-Lukas Graf’s great book Interpretation: How to Shape a Melodic Line has an entire chapter on dynamics, discusses how they are performed differently in Baroque, Classical, Romantic, and Modern genres, but never actually discusses the differences in levels of intensity. On pg 85 he gives examples of mp and mf that should be performed with “minute exactitude” but never really addresses what the differences between p, mp, mf, and f actually are.
I stumbled on this proposal that tried to scientifically quantify dynamics calling for a minimum of a 30dB spread over “all six dynamics” from pp to ff, so a 5dB gain moving from one dynamic to the next. Clearly this is assuming the an equidistant range from p to mp to mf to f:
So as it currently stands, when you only use p, mf and f, then between p and mf there’s actually 1.5 steps, while between mf and f it’s only 0.5 steps? Well, this explains quite a lot of what I was experiencing and could not explain…
I just tried a little test, and I’m now convinced this discrepancy really should be noticeable in real life. Playback settings were Curve power: 1.0, Min level: -3, Max level: 3, all Humanization off. VST used was IK Multimedia Hammond B-3X (so something with basically unchanging sustain), all vibrato off, Leslie off (brake). Decibels measured on an iPhone with SPLnFFT app. iPhone was unmoved and placed on a desk triangulated between EVE Audio SC207 stereo speakers. Measurement occurred after the initial attack after the note had stabilized. Test ran 3x and values averaged. Y values in dB as reported by app.
Obviously there’s a way bigger jump in perceived volume between p and mf, than from mf and f. If you are using a system that doesn’t use mp and are expecting mf to be the midpoint between p and f, then it’s not really gonna happen and I would think the imbalance would be audible to most listeners. Personally, I’d like a setting to have p - mp - mf - f actually be linear as a starting point with even increases in intensity, before applying any curve.
This is an interesting discussion.
But it seems that dynamic is being discussed as only indication of volume.
With my 40 year long experience, as piano performer I would like to suggest that dynamic indication have much to do also with the character of sound and balancing of voices. As an example, it is not uncommon that an accompanied melody notated as “piano cantabile” can have a bigger volume than for example eight voices chords notated with “forte” (I mean every one of the eight voices doesn’t have to reach the forte volume: it is the sum of the eight voices that give powerful sounding chord.). Also, the “attack” and deepness (how deep the weight of the arm goes) that the player gives to the note can influence the character of the dynamic (for strings the weight on the bow, for winds the “stützen” of the diaphragm):
Fast, deep and nervous attacked “piano” can sound louder in character as a slow and less deep attacked “forte”. The context of the music material and density of harmony and counterpoint give clues how to perform the dynamics. Also, it’s interesting how composers tend to clarify the character of their dynamics (that are not only a volume indication). Thinking of Brahms “piano dolce”versus “piano espressivo” or pp vs. sottovoce. Brahms uses often pocoF.
I have this thoughts always when I try to mockup something that I play on acoustic piano, into a DAW with virtual instruments. Every library reacts differently and is needed much manual work to let the music regain her natural feeling throught the characters and the “affetti” that the dynamic should convey.
Just some thoughts…
The explanations here of Dorico’s nonlinear response seem to echo what I hear as “improper dynamics” in my piano playback when using The Grand. I have played with the dynamics curve of The Grand quite a bit, the VST has a configurable curve, and it never seemed to match the dynamic output which I thought it was supposed to have from the velocity input. There are a lot of variables (software layers) at work here, including humanization. Maybe now that I am aware of this nonlinearity between p…mp…mf…f then perhaps I can write dynamics which match my intention, purposely using Dorico’s dynamics-half-step.
The response of Dorico’s playback dynamics is important for those pieces which I submit directly as audio tracks, i.e. for film/tv underscoring (vs. engravings intended for live performance/recording where humans will interpret the dynamics). That is, the audio is not simply a mock-up or a demo; it is the real track. It is very time consuming to manually redraw the velocities of piano notes especially when they are overlapping as a harmony; other (single-line) instruments are not as troublesome. So, precise dynamics are needed.
Hopefully the official explanation will be added to the Dorico manual, at least.