synthesizer dynamics

• Sep 1, 2020 - 17:20

I am trying to get a piano piece to sound more expressive. I press Advanced settings and press Switch all sounds to expressive, but I don't hear a difference. Also, what are all the other settings on this Dynamics tab?

Also, under Text when I chose Expression, can this have a playback effect?

I am trying to get the piano sound to sound less robotic and more legato.


Unfortunately single note dynamics (the purpose of that tab) does not affect pianos since they cannot crescendo on a sustained note. The expressive channels (found on wind and bowed string instruments and a few others) use the default synthesizer settings for dynamic changes on single notes.

I've seen people put a lot of work into making piano music sound astounding on but I don't normally write for piano so I'm not the person to help you with that part.

In reply to by mike320

I would love to figure this out. If you come across such a great sounding score, please send it my way. I am trying to orchestra in Musescore, rather than in a SAW, and I want to see how far I can take it. would want to do the reverse of quantizing which is used in a DAW, add a human touch, maybe even randomization.

In reply to by Marc Sabatella

Is velocity the same as volume (amplitude?) Because I have the score notated with ppp, pp, p, mp, etc, but the piano roll shows a flat relative Velocity. Is there a way to draw a line across the piano roll that curves up and down in waves, rather than note by note Velocity adjustment? When I adjust the velocity on the piano roll, I see the adjusted note move up or down in the velocity chart, but it seems the piano roll is not receiving the dynamic markings I inputted into the score.

In reply to by Shirly Lyubomirsky

Well, velocity is literally how fast you strike a key on a keyboard, but indeed this translates directly into how loud the note comes out. Unless your score was imported from MIDI, the default velocities of the notes are determines by the dynamic markings in your score, although you can override these with the Inspector or the Piano Roll Editor.

The fact what you need to tweak things more or less a note at a time is why I said it would be pretty painstaking work, and why a plugin to automate the process would probably be more useful for your purposes.

Adding slurs will make the notes more legato - they play at only 95% of the nominal duration otherwise. But I think you'll find that difference is almost imperceptible, and the real opportunity for expression comes from fine tuning the individual volume of each note, pushing and pulling tempo here and there, and other more detailed manual tweaks.

"Expressive" sound is about single note dynamics only - crescendo during the course of a single note. Not applicable to piano, as the instrument is incapable of this.

Expression text is where you type text to tell a human performer what you want, but if you you also want the computer-generated playback to do something different, the controls available are in the Inspector and/or Piano Roll Editor (right-click the staff to see this option). You will be able to tweak the start/stop time and the velocity (volume) of notes all you want.

In reply to by Marc Sabatella

I'm confused about how to select the dynamics and hairpin for crescendo and decrescendo so I can hear the effect in the playback. I guess I need to figure it for the piano first.
For the dynamics:
1. I choose to start note of the phrase: I intend the first note to start at p and last note to end on mp.
2. I have a crescendo hairpin in between p and mp.
3. I click on the dynamic p and go into the inspector and I chose the "velocity". say 35
4. Do I leave "velocity change" alone, because the single notes of the piano don't change?)
5. What is the "Change speed ?". Is it only related to single note change of speed? Does it determine how quickly a single note velocity change takes place (for say a trumpet)? So "slow" "change speed" will result in more time spent on the softer dynamics than in the ultimate mp dynamic?
6. What do I chose for "Style" - since I am tweaking the dynamic, I just say dynamic, or do I say hairpin? Why am I asked this question?

Then for the hairpin
1. I choose "Type" Crescendo Hairpin
2. "velocity change" - not clear if this is the button that will determine how fast I rise in dynamics from p to mp. Kind of rate of change? Or maybe its the peak dynamic of the hairpin? I noticed a high-Velocity Change causes the dynamic to rise above mp and then settle on mp. So, If p is 40 and mp is 50, when "velocity of the hairpin" is 60 I hear a bell curve effect, with peak determined by the value of "velocity change when the peak is reached is determined by "linear" or "exponential" etc settings.
3. what is the purpose of "single note dynamics" when it comes to setting the hairpin?

During a hairpin, can I give my piano-voice_2 its own dynamics, or is the dynamics connected to the staff?

Thanks so much

In reply to by Shirly Lyubomirsky

For most use, leave the settings for velocity alone and let MuseScore handle it. Here are the details if you decide to change the defaults.

Piano single notes cannot crescendo so there is nothing you can do to make a single note of any length get louder. You can control the decrescendo some but this is based upon how quickly the note fades. The pedal affects this in both MuseScore and a real piano. MuseScore actually does a decent job of doing dynamics on for the piano.

In general, For dynamics (p, mf, sfz etc.) the velocity is how loud you want to change the volume to at the start of the note. the velocity change is used for dynamics like sfz, fp and other dynamics that indicate a change in volume after the initial attack. If you enter a negative number here, the volume will get softer. A positive number will make it louder (like for a pf dynamic).

For hairpins. If you enter a hairpin and do nothing to modify it, it will start at the volume of the current volume. If you enter a dynamic, a hairpin (also includes cresc. and dim. lines) and then a dynamic that will make the hairpin move in the correct direction, the volume will change over the duration of the hairpin. So if you have

p cresc. f

the crescendo will grow louder. If you have

f cresc. p

the crescendo will do nothing.

One thing to keep in mind, MuseScore uses the last velocity for the start velocity of a hairpin and the dynamics (like f and p in these examples) can be as many measures as you like as long as nothing changes the volume in the mean time. If you have two or more hairpins in a row with no intervening dynamic, only the first one will do anything.

This is where the velocity change field in the inspector comes into play. Always enter positive numbers in this field. MuseScore will do the right thing with it. When you change the velocity change field the hairpin will start at the last volume no matter how it got set then change in the appropriate direction by the amount in the velocity change field. So you have

p cresc. (CV=40) dim. (CV=20)

CV is the Change Velocity field in the inspector, you will end up with velocity of 69 (49 + 40 - 20 = 69).

Here is the link to an info page with a slightly different explanation:

In reply to by mike320

What about a single cres hairpin lsandwitched between p and mp like this: p resc. (CV=40) mp?
Will the final velocity be p+40 or just mp?

As far as velocity change of a single note, for version 3.5 do I need to create a separate voice, say voice 4, and put rests in there and have the hairpin affect the rests, inorder to allow the previous note in voice 1 to change velocity?

In reply to by Shirly Lyubomirsky

A hairpin will always use the velocity in effect when it starts. For example

p cresc. dim.(CV30) mf

the cresc. will increase to mf because there is no CV on the the cresc so it looked at the next dynamic which is the mf. The dim starts at mf (because that's where the cresc ended) then the velocity at the end of the dim is mf-30. When the score finally reaches the mf the velocity will be set to mf.

You can take advantage of this if you want but its advisable to put something telling the cresc. in this example how much you want it to crescendo either a CV or even a dynamic (like mp) you make invisible.

BTW, I've only used CV in this discussion, it's not a normal abbreviation.

Some things to consider:

I normally might have two scores for any piece I'm working on. One marked the way I want for handing to real players. And one I mark the heck out of to get MuseScore to playback the way I want. Some pieces will never be for real players, so that cuts out that score.
I don't get the whole "randomization" thing. When I play music, there is nothing random about it. Some people like little random mistakes. Mistakes do not make music human, the expression humans put into the music are what make music human. Hence, your problem. And piano is particularly tough.

Let's take some 8 measure phrase. How would a real player play this? They might star tout a little softer and build slightly in volume and tempo, then ease up on both towards the end of the phrase. A real player needs no instruction to do this. It's how we play music. But a computer, smart as it is, isn't very musical. You have to tell it all that stuff.

My advice would be to keep it simple as you can. Perhaps don't worry about velocity for now. A lot can be done with dynamics and hairpins. Consider the TempoChanges plugin that will add a rit. and accel. type playback. If it fits the music, you can add an ever so slight pause at the end of a phrase. No need to over do any of these. Use just enough to make music. Not just notes on a page.

Some people what software that will do the humanizing for them. I think the human needs to do that. Otherwise you end up with someone else's version, not yours.

In reply to by Shirly Lyubomirsky

I know what you mean. And I know this is a big deal for DAW users. There are so many more things we can do to make playback more musical than to try to imitate human mistakes. I don't suspect that the 5th chair,2nd violin meant to come in late on that 64th note run? I think that they would have preferred to play it perfectly? I don't think that the performance would be somehow worse if they played it perfectly? Because we are using a computer and recorded sounds, we've already cut out real players.
Just my own take. Few agree with me.

In reply to by Shirly Lyubomirsky

Again, I do understand the concept. I only question the quest for that kind of realistic sound. Consider being at a live concert. You are sitting in the front row but on the extreme left side seats. The sound from the right side of the orchestra reaches you at a different time (and not at the intended volume) than the near side. This is realistic sound, but not one that I would try to emulate. DAW libraries are recorded such that there is distance from the mic, close and/or far, to give depth to the finish playback. This is so that the playback is realistic. I get it. Randomizing is another method. I get it. But none of it is real. It's all recorded.
Because we are not working with reality, I would rather be able to concentrate on things that make the recording much more than realistic. We would have the ability to make them musical. To breathe into them the life of a musical phrase, or the ebb and flow of a melody. Not based on some algorithm, but the ability to do the hard work of making music. Live musicians do this. It would be fun to be able to give it a shot on a computer. Oh, and not have to spend thousands of dollars to do so. Or have a steep learning curve. See how unrealistic I am :)

In reply to by Shirly Lyubomirsky

Shirly, don't let them try to convince you that the natural sound of musicians playing is in any way a "mistake". You know it, and I know it. I've fought that argument here too many times, and have been shouted down at every turn. My best wishes for you, but what you are so reasonably suggesting here is never going to happen under the current development team.

In reply to by toffle

What she whats to do is already possible. Just not as some kind of random algorithm. It's the random aspect that makes no sense to me. I'm not interested in shouting anyone down. I don't care if such a capability exists or not. I'm trying to understand how this kind of random system can possibly replicate the natural sound of humans playing. Especially since playback is not the natural sound of musicians playing (for a host of other more important reasons) to begin with. Why are we trying to replicate "realism" when we can do better. Each of us has a different idea of "real". A recording of a real orchestra is still a recording and not real itself. Any more than playback of recorded instruments in MS. Phrasing, tempo change, dynamics, and articulation, to name but a few, are what we need more of. These make music more musical than randomness.
This what I ask when these topics come up. It's got nothing to do with management or the development team.

In reply to by toffle

Hmm, I’m not sure what you mean. I don’t recall anyone ever expressing disapproval of the general idea of someday adding some sort of humanization. The time might not have been right 10 or 5 years ago, but the current team has laid out pretty ambitious plans for MuseScore 4, and this seems totally in keeping.

In reply to by Shirly Lyubomirsky

Part usually is a staff. Exceptions are grand staff instruments like piano and Harp. In the symphonic templates Horns are also place on a grand staff. I normally delete a staff and create two instruments since the horns do things like change keys independently. I recommend against making the staff the default in most cases.

What you can do for these exceptions is edit an existing dynamic then use ctrl+shift while you drag it to a palette (I'd name it something like Staff dynamics) and use these.

In reply to by mike320

I write for small orchestra. Usually only one horn. If I do have more than one, I put them each in their own staff. Though not because of this, " horns do things like change keys independently.", which I'm not sure I understand.

In reply to by bobjp

"horns do things like change keys independently"

I've seen all of the horns start in one key (say f) then only horns 3&4 change to something like a C horn at a certain point in the score. I transcribe mostly 19th century music so I see things in these old scores that I wouldn't usually expect in a modern score. Horns today are normally written for horns in F with tuning crooks not commonly used.

In reply to by mike320

I thought you were seeing this in modern music. Years ago, I played a reproduction of a Baroque trumpet. It was built in D with a crook to C. The same company made horns and came with a rack of tuning crooks. They also offered sackbuts, and a keyed bugle. Lots of fun.

In reply to by Shirly Lyubomirsky

To be clear: the default is for each instrument to get its own dynamics. That's exactly what the default of "Part" does. You only need to change it when doing something unusual, like trying to make one dynamic to apply to all staves, or trying to have independent dynamics for the two staves of a piano part (not that unusual, but not the norm either).

In reply to by Marc Sabatella

Thanks. That would save me lots of time.

I have a suggestion, i made years ago. When a person right clicks on a note, all the options for that note (dynamics, accents, velocity,etc) should pop up right there for the user to chose. No need to go to palette for each and then go to inspector.

Also, once a dynamic is chosen in the score, arrow up should change it to higher velocity, arrow down, to lower velocity. If you lower it past its specified articulation (say u lowered velocity from
120 to 40), then Musecore should automatically replace the current written dynamic with the one appropriate for the new velocity selected.

Alternatively, for dynamics playback, there should be curved velocity line that allows a person to shape it as desired, then the hight of the curve will translate to dynamics written automatically in the score. This way, in one shot, composers chose their playback sound by ear, and the written form (dynamics and hairpins) follows inteligently. The curve could have an amplitude coordinate labeled with a velocity scale.

Do you still have an unanswered question? Please log in first to post your question.