Piano notes short-lived during playback

• Jul 15, 2018 - 09:05

I've migrated from Sibelius a few weeks ago and I'm never going back. I have to say, I'm fascinated with Musescore. Every single thing I set myself to figure out how to do in it turns out to have been sorted out in the most clever way I could possibly think of. And every single thing I wonder whether Musescore also lets you do... turns out it does! Every. Single. Time. It's awesome!

There's only one detail I miss, and that is piano playback. Any other instrument plays just fine, but piano notes under a sustain pedal don't overlap like they would in real life.

This is what I've tried so far:

1) Changing the soundfont. I've tried tons, but the best I could achieve was a decent legato, never truly overlapping notes. I've even opened the .wav files to check the durations and they sound for several seconds, Musescore just kills them too soon when another note starts playing (or sometimes even before!).

2) Saving as .mscx and changing the gatetime manually with a text editor. Still the same.

3) Manipulating the piano roll. I can get the notes to look overlapped in the diagram, but they sound exactly the same as before.

4) Instead of employing a sustain pedal, writing each note in a different voice and giving them long enough durations. No difference.

My apologies for the wall of text, but it does reflect all the effort I've invested in working it out. So is there anything else I can try? Has this point been risen a thousand times already? (Sorry if that's the case.)


Comments

In reply to by Shoichi

The reason I asked if this point has been risen before is because upon researching, this article comes up immediately on Google:

http://missives-from-the-edge.blogspot.com/2015/10/improving-playback-f…

It's the author of that very score you referred me to (he seems to be no amateur). He wrote an entire piece 3 years ago on how to work yourself around this shortcoming, so I figured this must be a known bug by now.

My expertise is nowhere near the point of understanding the whole article, but that's where I got the idea of manually changing the gatetime (and the know-how). But notice this is just one of several complex measures he takes in order to solve this. He also mentions resorting to the equalization. I don't know the first thing about what the markings on the "master effects" sleeve in the synthesizer mean, but I did try fiddling with the nobs very insistently and in the end I sort of got something similar to what I was looking for, only it wasn't the notes themselves overlapping but their "echo" (resonance? chorus? "wetness"? IDK)

As to my precise score, it's a work in progress, but I'm uploading a specific problem passage as "problem passage 1.mscz". I also uploaded a MIDI file exported from Sibelius that shows the sound I used to get (even with that horrible artificial piano). I know there is a little reverb involvement, but besides that, I can hear both eighth notes overlapping with their following quarter in each compass, which is basically the point of a passage like this. With Musescore I get legato at best.

[UPDATE: when writing this I set myself to also export a midi from Musescore for comparison, and to my surprise, I GOT THE EXACT SAME AUDIO I DID FROM SIBELIUS! Again, I'm far from an expert, but I figure it's exporting the notes alone and that's behind all this.]

I'm also uploading a second problem passage. The point of that one is for the 3 notes in the second beam of each compass to overlap, which I got with Sibelius. This one passage is especially problematic because depending on the soundfont I'm getting "somewhat legato" at best; and at worst, for some reason, the G sharp in the second beam of the second compass detached from its following F sharp, which is the exact opposite of what I'm looking for!

Anyway, sorry again for the long post but I really am enchanted with this open source software and do want to make it work (I stumbled across one impractical workaround upon writing this, which is, to export as midi every time I want to listen... although I guess I'd have to rely on that dreadful artificial piano forever doing that). Thank you for your time.

In reply to by Boente

The problem for me is not the length of the post but the focal point of the question...
You talk about beam, Beam property palette(?), but maybe you're referring to audio playback. I hope that someone will be able to give you some clarification.

In reply to by Boente

I hear it sustaining, I guess maybe it decays a bit too much and too fast so that the sustained note isn't as apparent as it might be. Try deleting the last two notes of each measure and you'll hear the first half is sustaining (and then delete the pedal mark to hear the difference).

Anyhow, this is an attribute of the soundfont I believe. Feel free to try a difference soundfont as described in the Handbook under "Soundfonts". The one that was used as the default prior to 2.3 - FluidR3 - shows the effect better, I think.

In reply to by Marc Sabatella

> Try deleting the last two notes of each measure and you'll hear the first half is sustaining (and then delete the pedal mark to hear the difference).

I tried that and I do notice a difference, but by doing so it becomes evident that this is not going to be enough to attain the effect I'm looking for.

However, after further experimentation I've come to the conclusion that it's in the soundfont, as you say, and not in how Musescore renders the notes.

1) If I play only one note spanning over several compasses and look for its specific .wav file and compare the sounds, both last just as long.

But for this comparison to be fruitful I need to employ a note from the lower end of the piano.

2) If I lower the entire "problem passage" a couple octaves, I get the EXACT overlapping effect I'm looking for.

So the whole issue boils down to the higher notes on the piano naturally decaying faster than the lower ones. I guess I was used to an artificially long-lasting high end of the piano.

In reply to by Boente

Actually, I agree with you that the higher notes do decay a bit too fast in the current soundfont. Each piano is different, just as each soundfont is different, but it seems the default soundfont is modelling a piano with an unusually short sustain.

I could maybe imagine this was done deliberately because many people don't write enough pedal changes and this keeps the music from sounding too muddy. But I'd rather aim for more realism.

Hopefully it's something that can be addressed in a future update. Meanwhile, though, as mentioned, using another soundfont seems to address it nicely.

In reply to by Marc Sabatella

> Hopefully it's something that can be addressed in a future update.

That would be awesome! I've downloaded Viena hoping to be able to modify this at will, but the whole program seems to be much more complex than I foresaw, I guess there's plenty I'd have to learn. I still might get into that at some point, but it would be great for the default piano to have a longer sustain.

PS: (In case it matters at this point) every time I've said "compass" in previous posts, I meant to say "measure". Pardon my Spanglish, amigo!

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