Improved playback interpretation of dynamics in solo piano pieces


First of all, let me say that coming to Dorico from Sibelius and Musescore has been an absolute joy! The fact that Dorico seems to “understands” music, and isn’t just about putting symbols and text on a page, in my opinion really sets it apart.

One issue I’ve been running into:

I’ve been working on some solo piano pieces and have run into a minor frustration in how the playback of dynamics works. In the piano music of Mozart, Beethoven, Chopin etc., we typically see one dynamic marking between the staves, for example:

However, the performer is not expected to play both hands at the exact same dynamic - rather, to oversimplify, the melody should typically be “brought out” to the foreground and the accompaniment should typically stay in the background.

By default in Dorico, the exact same dynamic ranges are assigned to all voices in both staves. In fact, if the melody is finishing a phrase while the accompaniment is beginning a phrase, the first beat emphasis may make the accompaniment even louder than the melody.

This is not an ideal situation for playback, and I have been looking into ways to fix this in my own projects. Some ideas:

  1. Voice-specific dynamics (

This strategy works fine in principle for simple pieces (where we could just hide the lower dynamics for the left hand, for example), but would be very tedious to use for longer, more complex works. As an example, imagine mirroring complex changing dynamics between multiple voices, with one voice being a dynamic level (or more) louder at all times.

Additionally, with the limitations described here, this is even harder:

  1. Enabling independent voice playback (

This strategy, with which I am currently experimenting, seems better than (1). Theoretically, I can separate the different voices for playback and use the mixer to make accompaniments lower. This way, we can utilize a single set of notated dynamics, and bring a whole voice up or down “in the mix” to accomplish the desired dynamic balance.

The only drawback to this, as I can see so far, is that we’ll potentially need to artificially create some new voices if, for example, the melody passes from the right hand to the left hand. But this is not insurmountable.

  1. Feature proposal: Ability to “annotate” sets of notes as melody / accompaniment (or some other, more general terminology). This would accomplish something similar to (2), but would be slightly more direct and optimized.

Anyway, I would love feedback on whether or not I’m missing something in terms of what is currently available, and also whether or not the feature proposal in (3) is indeed something that could / should be implemented.


The most obvious thing you’re missing is the MIDI note velocity editor in Play mode. Select a bunch of notes you want to bring out, in Write mode, then switch to Play mode, select the velocity icon (it’s the bar graph to the left of the forte sign), and pull up the velocity of those notes (or change the value on the left side of the track).

Hi Pianoleo,

Thank you for the response and this is a good suggestion! The one issue I’m running into with it, though is that the MIDI velocity adjustments seem to set each selected note to be the same resulting value. In other words, if I have a passage that goes through dynamic changes, if I select all of the notes and try to move them down 10%, for example, all of them will result in 10% reduction of the first selected note.

I think this would make this type of change for a more complex piece similarly tedious and hard to maintain as in my proposed strategy (1) above (aka the many hidden voice-only dynamics strategy).

One issue that I’m running into with (2) above is that it seems these voices / Play settings work across flows in a project. I would expect mixing, voices, etc. to be contained by one flow, not shared by multiple. Am I missing something?

David, Many of us are looking forward to more sophisticated midi editing in Dorico of the type you describe. I think it will appear, but none of us know when…maybe v.4 will add some additional functionality in this area. It is indeed tricky to simulate realistic piano playback simply by plugging in the notes and adding dynamics, even to different voices. If you are a pianist, I would suggest you record it in, maybe a few bars at a time, in order to get more realism. Of course a recorded piano part does not often look like you want the score to look, so in these circumstances I will have two piano grand staves next to each other in the score. One staff can be for the score and the other for the recorded version with performed dynamics. Once everything is complete you can hide the messy recording staff, (it will still play back) and of course you need to Suppress Playback for the piano part in the score. You can also edit your recorded version to perfect your performance!

Good suggestion, thanks for the response.