Bends are out of tune
Type
Functional
Frequency
Few
Severity
S4 - Minor
Reproducibility
Always
Status
active
Regression
Yes
Workaround
No
Project
Sometimes, bends produces wrong pitches. The issue is more noticeable with short note (in beat) and with fast pitch variation inside the bend . Play attachment and look at bend proprety for more details.
Attachment | Size |
---|---|
out of tune pitch bend.mscz | 5.29 KB |
Comments
Confirmed on latest master (c51fa84).
Came up again in //musescore.org/en/node/228106#comment-741401 .
And see #139621: Faulty playback of bend acciaccaturas.
https://github.com/musescore/MuseScore/pull/4538
Fixed in branch master, commit 94caa9fd9f
fix #68456: fixes pitch bends
Fixed in branch master, commit 1d5ae8afbb
_Merge pull request #4538 from MichaelFroelich/68456_fixesPitchBends
fix #68456: fixes pitch bends_
In reply to Git webhook message by Git Message
Sorry for your effort, but i tested the development build 1d5ae8a and it introduce a new problem instead fixing the old one. In the old problem, pitch bend tend to sound bad when time subdivision increase. Now, bends that was correct before have wrong pitch. For example, a full bend will be played as a halftone bend.
In reply to Sorry for your effort, but i… by Yruama_Lairba
Thank you for your feedback, I am aware that this changes the amount of pitch bend, the original issue was that the amount was exactly half of what it should be. I actually used a DAW to test this and ensure the starting and ending pitch was correct.
Thank you also for posting a musescore file for an example, I'll get around to fixing this as soon as possible.
Fixed by increasing the number of pitch change events, I might need some time to come up with a better solution but so far this should be fixed.
see https://github.com/musescore/MuseScore/pull/4796
Related to #286108: [EPIC] Bends issues.
"Bend" already wrong on latest nightlies.
It plays at half the value it should to be. eg: 2 half tones = 1 half tone
like: interval = interval/2
And an extremely out-of-tune (detoned) situation occurs.
And I can't say it was better than before. OK no polite: more bad actually
Edit: and Beta1 too: OS: Windows 10 (10.0), Arch.: x86_64, MuseScore version (64-bit): 3.1.0.6437, revision: 06f0ddd
Edit: and Beta1b too: OS: Windows 10 (10.0), Arch.: x86_64, MuseScore version (64-bit): 3.1.0.6721, revision: 1982daf
I found a perhaps extra bug:
The old code did this:
This is almost a division-with-rounding. Were it one, it would have looked like this:
Note that the extra is half the divisor, not half the scaling factor, and appended before dividing. I don’t know if that was deliberate.
However, the new code does:
Note how the first 16384 was halved but the last 8192 wasn’t. This should perhaps look either like this:
Or even like this:
Where are MIDI pitch bends defined? (So we can read which calculation is correct.)
In reply to "Bend" already wrong on… by Ziya Mete Demircan
If nothing has been done, please bring back the old one.
Hearing a little error at the end of the bend is better than hearing the guitar bends is continuously out of tune.
"If nothing has been done, please bring back the old one.
Hearing a little error at the end of the bend is better than hearing the guitar bends is continuously out of tune."
+1
Bump (for 3.1 RC)
https://github.com/musescore/MuseScore/pull/4538 got reverted, waiting for https://github.com/musescore/MuseScore/pull/4796 to get fixed (I think)
I reverted https://github.com/musescore/MuseScore/pull/4538, because the changes from https://github.com/musescore/MuseScore/pull/4796 are not supposed to be clearly understood and thoroughly tested.
Any updates on this?