Musically sensitive midi output
I find myself more and more fascinated by the task of using Musescore to generate musically sensitive midi output. My current situation does not permit me to have a piano or recording equipment, hence Musescore has become, by default, the only instrument at my disposal. Getting the most out of it is very important.
I've put two more examples of Musescore-generated scores and midi output on my YouTube channel.
Simple Gifts and Quiet Songs, a song arrangement for clarinet, guitar, bass, and strings that really shows off what you can do with a good soundfont and a lot of midi tweaking (in this case, the Merlin orchestra)
http://youtu.be/9XjBparEfgg
Intermezzo, essentially a piano solo with a vocal interlude where the voice part is carried by a cello.
http://youtu.be/uQjeOe45u9o
The music is beautifully engraved, as is to be expected from Musescore, but it's the midi output that really surprises. Although it's still a lot of work, using Musescore to prototype musically-acceptable arrangements is easier and more musically intuitive than working with a midi editor.
Piano scores are the most labour-intensive to realize. Is anyone else out there doing the note-by-note, chord-by-chord, phrase-by-phrase thing with Musescore piano scores? If so, I'd love to hear about your experiences and what you've learned from them.
Comments
I would be curious to know which parameters you are changing and if you think it could be automated?
In reply to I would be curious to know by [DELETED] 5
The parameters are the same ones I would be changing reflexively while playing the piano: dynamic, attack, sustain, decay, and tempo (or, in midi-speak, velocity, ontime offset, offtime offset, and beats-per-minute). I doubt the process could be automated since it goes to the very heart of interpretation. Rubato, for example, is much too context-dependent to be automated, as is the balancing of voices within a chord. I imagine a really good soundfont would take some of the work out of the latter, especially for piano music, but other than that, nothing short of reading a composer/performer's mind would be truly useful. (A "Do What I Mean" option for Musescore—or any other software—would be a nice touch, wouldn't it?)
One thing that would help a bit would be a slightly different approach to the Note Properties dialogue. As things stand now (v1.2), if I select a batch of notes and change the velocity, I lose the individual notes' ontime and offtime offsets. This becomes problematic in phrases where the offsets have already been adjusted and all I want to do is to make the whole phrase a littlle louder or softer. If only the changed parameter were applied to each individual note, while the others remained untouched, it would certainly make life easier.
Another useful feature would be an option to set default velocity/offset types and values for an entire score (probably in the Preferences menu), prior to beginning note input. It gets very tiring having to constantly tell Musescore I want a velocity type of "user" and a value of, say, 60; the default—"auto" and "80"—is nearly never what I want.
In reply to The parameters are the same by Peter Schaffter
Regarding the initial values for velocity, while Ai'm sure that could be done, how the results different from somply placing an appropriate dynamic (eg, "mp") at the beginning of your score? Wouldn't the same MIDI get generated either way?
In reply to Regarding the initial values by Marc Sabatella
Yes and no. Even when, say, "mp", is set such that, contextually, it produces the right velocity value, the velocity type remains "auto", and still has to be changed to "user" in order to shape the dynamics of the mp phrase. It's actually a global velocity-type option that I'd like to see, not a global velocity-value option.
For jazz and popular music, I find I can often stick to the same note velocity throughout entire passages, but for music that demands classical phrasing, it's simply not possible. For example, the basic shape of a cantabile phrase in classical music is < > (ie slight crescendo to middle, diminuendo to end), and requires note-by-note, user-set velocities. While that aspect of a midi "performance" cannot be automated, there is some advtantage to making it a little less awkward in the way I suggested.
All this in recognition of the fact that Musescore's focus is superior music engraving, not midi editing.
In reply to Yes and no. Even when, say, by Peter Schaffter
Ah, you mean MIDI tweaking *within*MuseScore. For some reason I thought you meant you were exporting a MIDI file then tweaking the result in a sequencer. In which case, the distinction notes entered orognally as auto versus user wouldn't have mattered - 60 is 60.
Anyhow, I could certainly imagine a whole interface that sat on top or next to the main MuseScore interface to allow more of this tweaking to be accomplished more easily.
I am with you about the need to produce really useful audio as well as beautiful printed scores. Recently I had to convince a skeptic of the usefulness of audio while developing this improved version of Robin Frost's "Cold Soup." Making the changes without the audio would have been almost impossible.
In reply to Usefulness of Musescore audio by EdwardsRG
While it's all well and good for those of us with years of training and experience to read a score and hear it in our heads, many's the musician who can't. When we wish to share our songs and compositions with preformers who don't have the facility, we need good audio.
I've looked at and listened to quite a few score+midi combos, and frequently find the midi so hamfisted that I have to kill the audio in order to assess the composition's merits. Fine for me, as I said, but what about the talented singer who's looking for new material but can't form an image of a song from just the chart? If s/he has only clunky midi to listen to, the necessary connection to a song may not be made, and a beautiful piece of music may be passed over.
Thus, increasingly, I'm wanting to produce both professional scores/charts and useful midi. Musescore is so very close to being the perfect tool that one truly does long for a more sophisticated midi interface.
Delicious Cold Soup, by the way. :)
I think an automatic scalable velocity from beat position (strong/weak) could do wonders. e.g. assuming 4/4: beat 1 is automatically made louder than beat 2, beat 3 is louder than beat 2 (but not as loud as beat 1), etc.
This could possibly subdivided into levels based on the note duration. Syncopation may not work under this scheme.