Stuttering Salamander SFZ playback in v2.1
Hi,
Just upgraded to v2.1. Getting very stuttered playback on my piano scores (sample attached) using the Salamander SFZ soundfont.
v2.0.3 handled these fine, so I don't think this is an issue with my PC or set-up.
Seems worse in measures with shorter note durations or tremoloes (which don't seem to play at all), but there's no obvious pattern. Playback is fine when switching sound to FluidR3Mono, but my preference is Salamander. Have checked a number of my scores, they all have the same problem to a greater or lesser extent.
Any suggestions/advice?
Thanks, Phil.
[v2.1, Windows 10 64-bit, SalamanderGrandPianoV3.sfz.]
| Attachment | Size |
|---|---|
| Sonata_in_B_Minor_1._Lento_Assai,_Allegro_Energico.mscz | 101.75 KB |
Comments
I can also reproduce this on openSUSE Tumbleweed with MuseScore 2.1.
I cannot reproduce this issue with some other SFZ files though, like Ivy Audio's 'Piano in 162'.
I suspect this has to do with some changes made to the Zerberus that affects the Salamander soundfont -- I'll take a look and see what I can find. My first guesses would be something to do with #116871 or #116691, but I'm not sure at this stage.
In reply to I can also reproduce this on by Qantas94Heavy
You mean #116871: Zerberus rt_decay or #116691: Zerberus onCC?
In reply to You mean #116871: Zerberus by Jojo-Schmitz
Turns out Zerberus seems to be "out of voices", which is causing this issue. As to why it's running out of voices, I'm not sure, but it's definitely worse following the fix for #116306: Zerberus Trigger::Release.
In reply to I can also reproduce this on by Qantas94Heavy
Thanks. I just spotted the following on the Downloads page re v2.0.3...
"64-bit version: There are only two known reasons why you would want to run this unsupported 64-bit build: Because you want to use a huge SFZ or because your MIDI keyboard driver doesn't support 32-bit."
I know Salamander is *very* large (something like 2GB uncompressed) - is there some reason v2.1 can't handle such large soundfonts? Hoping not as the 2.1 features will be v. handy, I'd rather not roll back.
Phil.
In reply to Thanks. I just spotted the by PhilTA
2.1 should support the file. It has better handling of sfz,
I can also confirm this type of bug.
To recreate it put eighth notes a perfect fifth apart like a and e and repeated them for about two bars and it starts making weird noises.
This could be some sort of feedback in the audio playback, I guess.
It was tested on the latest Linux Appimage on Ubuntu 16.04 also with the Salamander
Same problem here. Sometimes I could 'cure' an individual bar by deleting it or the contents, then re-keying, but that doesn't always work. In the end I reloaded MS 2.0.3 (had to uninstall MS 2.1 first). Not ideal, as I'd like to try MS 2.1 features. Will keep an eye on this thread to see if there's a proper solution some day...
In reply to Same problem here. Sometimes by Mark Sandham
You could of course use another soundfont in Salamander has issues with 2.1.
In reply to Same problem here. Sometimes by Mark Sandham
The perhaps it as a workaround "solution" for you, you can have 2.1 installed for editing, and use the portable version of 2.0.3 for your audio-exports.
In reply to The perhaps it as a by jeetee
Thanks jeetee - that advice (and the discovery of the Acoustic grand piano soundfont for v2.1) have kept v2.1 on my system!
Would love to see a fix for this though as it would greatly reduce the workflow for uploading Salamander scores to musescore.com.
I have this same issue with SalamanderGrandPianoV3.sfz; however I wanted to mention a further wrinkle I've encountered, which seems to suggest the problem may be in how new files are created, not just how they are played back?!
I am running version 2.1.0, revision 871c8e, on Mac OS 10.9.5. The stutter is occurring with a brand new file. However, an older file created with 2.0.3 still plays fine inside 2.1.0. This seems weird to say the least.
I will attach both files for comparison. The older file is a partial transcription of a Jack Dejohnette composition, "Lydia"; the newer file, meanwhile, is an extremely partial (just begun) arrangement of a much simpler tune, "Pancho and Lefty." With this new file, the stutter occurs in the third and fourth bars of the piano part, and goes away once doubled notes are no longer being played (i.e. it doesn't occur at all in bars 5 - 11). This might suggest it is only repeated/doubled notes that are an issue - but I will return to this issue in a moment, because I'm not sure this is so.
For the file created in 2.0.3, "Lydia," there is never a stutter at all. And here is what's interesting: although Estard has mentioned in his comment that he can recreate the problem with repeated doubled eighth notes ("To recreate it put eighth notes a perfect fifth apart like a and e and repeated them for about two bars and it starts making weird noises"), when I do this in the old file - that is, put in repeated and doubled 8th notes for several measures - I still don't get any stutter. The file still plays back fine.
So I have no idea what's going on or how it can be troubleshot. But it would be nice to get Salamander back for playback, as it is still the best piano playback I have found.