Strange noise added while playback on specific notes (E3, F3 and F#3)
Hello,
I have the following problem: when I play a song (playback), or play the note on the piano keyboard with an E3, F3 or F#3 a strange noise is added to the playback.
1) restart doesn't help, it happens 100% of the time
2) If i export the music to for example .flac, the bug ist still there, meaing that the strange noise is added to the playback.
3) I've recently updated my Musescore, and the bug is still there. It seems like its indepented from the version.
Now, I know what I just said is pretty vague, so I've made a video about it. (please ignore the clicking sound, which is my mouse, but if you listen carefully you can hear what I mean): https://youtu.be/kvIrmx5Y5KY
Does anyone else has this probme? Is this a know isse (couldn't find any on the bug forum so I started this thread).
Thank you very much in advance!
-Gleip
Comments
I hear, you must add your version of MS and your soundfound
In reply to I hear, you must add your… by Raymond Wicquart
Hello
The MS Version is: OS: Windows 10 (10.0), Arch.: x86_64, MuseScore version (32-bit): 3.3.4.9066, revision: 7684abe
The Soundfont Version: I am not sure how to check this, but I've never changed it or added anything. The path is in /Documents/MuseScore3/SoundFonts. I've checked the path and it seems to be empty.
In my opinion, this is due to the sample used.
Female voices are participated slightly with male voices from an octave top.
Sample name: Ahh Choir F#4
Used in: #52-#54 (e3, f3, f#3)
In reply to In my opinion, this is due… by Ziya Mete Demircan
Okay, thanks.
Is this a known bug? Switching to a different sound seems to help, but i actually like the choir sound. its usually "cleaner" than instrument.
Is there anything I can do to get it work with the choir sounds?
In reply to Okay, thanks. Is this a… by Gleipdonir
Soundfont's maintainer may be able to get some sort of solution by expanding the coverage of the previous and next samples.
If you wish, you can report it in the issues section. (it may also be useful to specify the address of this topic.)
In reply to Okay, thanks. Is this a… by Gleipdonir
You seem to be using the default sound font. That font is not stored in the Sound Font folder. Do you mean that it sounds like the singers are gargling? I have W10, newest MuesScore, and the default font. I don't hear that problem on my system.
Other than that I might ask you why you have the 32 bit version of MuseScore loaded on your 64 bit system? It works, but the 64 bit version will use your computer better. I can't say it's causing your problem, though.
In reply to You seem to be using the… by bobjp
Thank you for your input! I will probably try different soundFonts and maybe change to the 64-bit version as well. ABout the sound: gargling isn't the right word I think. It seems like another sound is added to the vocals i.e. you play one note but you hear 2 different sounds. I think the YT Video in the first post shows the problem better than any description.
The choir sample that is used for E3-F#3 appears to have a bit of a note sounding briefly an octave above pitch. You're right, that is annoying. I don't know if I'll be able to filter it out or fix it via some other means, but I'll put it on my list of issues to look into when I'm able to resume work on the SoundFont.
In reply to The sample that is used for… by s.chriscollins
First of all: thank you very much for all the help and suggestions. I really appreciate it.
Second: Should I open a Topic in the Issue Forum aswell?
In reply to First of all: thank you very… by Gleipdonir
"Should I open a Topic in the Issue Forum as well?"
Yes, definitely. Opening a new Issue is the only way to ensure that your bug appears in the official list of "issues requiring action". Just check for any duplicate issues, in case someone else reported it first...
In reply to First of all: thank you very… by Gleipdonir
Thanks for #299366: Choir sample on E3-F#3 adds octave pitch
In reply to The sample that is used for… by s.chriscollins
(I'm experiencing this too.)
If you send me the sample as an individual - preferrably not lossy compressed - sound file (WAV), I could try to remove that annoying overtone, and send it back to you for packaging.
In reply to (I'm experiencing this too.)… by HuBandiT
I have attached the offending WAV file here. If you're able to remove the bad frequencies from the audio file, I can have the fix added to the next SoundFont update.
In reply to I have attached the… by s.chriscollins
Thank you, please find attached.
How do you guys like it?
In reply to Thank you, please find… by HuBandiT
This is a nice improvement, HuBandiT! I will add this fix to the next SoundFont update, which I will post here when it's ready for testing.
In reply to This is a nice improvement,… by s.chriscollins
I have finished version 0.1.9 of the SoundFont, featuring HuBandiT's fix for this issue. You can download the SoundFont here. Here is the HQ version (better ensemble strings).
In reply to Thank you, please find… by HuBandiT
I'm curious, HuBandiT, what software/technique did you use to clean the trouble frequencies? Whatever you did worked quite well.
In reply to I'm curious, HuBandiT, what… by s.chriscollins
Yes, indeed. Very nicely done. Could you do it for the other 2 frequencies aswell?
And..I was thinking about some kind of Fourier Transformation, then getting rid of the vocal pitch and encode the pure signal back. Does it work like that?
In reply to Yes, indeed. Very nicely… by Gleipdonir
Thank you. This single sample is used for the whole third, all the 3 semitones with which you had the problem. So fixing this single sample will fix all three keyboard notes.
And yes, that is what happened: I did a spectrum analysis, and found the objectionable sinusoid, and subtracted it.
In reply to Thank you. This single… by HuBandiT
Did you use Audacity's spectral editing for this or something else?
In reply to Did you use Audacity's… by s.chriscollins
Yes.
Oh, and where do I request separate S, A, T, B choirs? When listening to SATB scores, it is very annoying when a high note on a male staff ends up sounding female, and vice versa. Similarly, between S<->A and T<->B. It would be best to have each of the four voice types sampled separately, singing their own full scales. Enable me to hear a clear distinction between them even in an ensemble, and even when they are singing the same (or similar) notes.
In fact, I would be happy even with solo voices, just give me clearly distinct S, A, T, B.
In reply to Yes. by HuBandiT
Not available in GM
In reply to Not available in GM by Jojo-Schmitz
What limits MuseScore to GM? Is it a technological issue? Usability for users (getting confused about what is where)? Project management issue? (As in "where to draw the line about what to include so that everyone is happy"?)
In reply to What limits MuseScore to GM?… by HuBandiT
If it wasn't for GM compatibility:
You couldn't easily use another soundfonts (The instruments were need to readjust one by one in the mixer).
When you export as a MIDI file, the instruments did not match.
The setup of the drum instruments was incompatible.
and such that ...
However, additional banks can be added if necessary (separate S, A, T, B) sounds are found. (But it won't match another soundfont)
In reply to What limits MuseScore to GM?… by HuBandiT
GM allows for 128 banks each containing 128 presets. It is possible to have separate S A T B choirs sharing the same preset, but using different bank numbers. For example, in the MuseScore SoundFont, we could do something like this:
The GM standard features a preset fallback mechanism whereby the preset on bank 0 will be played if a preset using a higher bank number isn't available. So, if your score uses Bank 3, Preset 52: Tenor Choir, but you have changed the SoundFont to a different GM font that lacks a preset on Bank 3, Preset 52, you will hear Bank 0, Preset 52 (Choir Aahs) instead.
We already do this with the individual violins, violas, celli and basses in the HQ version of the SoundFont, not to mention all of the expressive presets for use with single note dynamics.
In reply to GM allows for 128 banks each… by s.chriscollins
And probably do the same for the solo "Voice Oohs"?
In reply to And probably do the same for… by Jojo-Schmitz
Yeah, I'd probably put all of the solo voices on preset 53, and all of the chorus (even choir Oohs) on preset 52, since that would provide the best fallback behavior.
In reply to GM allows for 128 banks each… by s.chriscollins
(Indeed, this is what I meant to use for this. This was introduced in Roland GS (1991), and then later (1999) adopted into the GM2 standard - but it is not in the original GM (also 1991). This prompted my earlier question of what limits us to GM.)
It is a different question whether and how expressive musical performances could be conveyed through MIDI (past or current versions). KTH did some good research on rule based expressive performances in the 90s (on their custom, expressive synths, probably not based on samples). Now that single-note dynamics is possible, I have a hunch their parameters - loudness, fine tuning, musical intensity/expression, etc. - could, after all, be mapped through MIDI, which means we could in theory implement such a rule-based "performance/artist" layer between the score and the MIDI output, resulting in a "humanize" feature that I see was requested before. In fact exposing the parameters or the rules for the user to amend, composers could finely prescribe the intended performance beyond what is possible with musical notation; with that we could open the door on collaboratively working out rules for expressive musical performance in the spirit of free (libre) software. Which could be a great tool for teaching music as well as helping the process of composing become more rewarding by better playback.
Eventually we could move away from sample-based sounds to something more real, more musical, which for example would allow melismas to be played back as such, instead of individual notes, and help with other phrasing. (This is the topic I brought up with you, Chris about a year ago, in email.)
In reply to (Indeed, this is what I… by HuBandiT
Also, this: https://musescore.org/en/node/290272
In reply to Also, this: #290272 by HuBandiT
No doubt there are more advanced technologies out there for creating realistic playback, but until somebody with the knowledge and time to implement such technologies comes along, we will have to continue working with the currently-supported sampler formats. That being said, there are still things that can be done with just MIDI control + the SoundFont format to improve playback realism. We've already added single-note dynamics, which is a huge step forward in expressiveness beyond what we had before, and I have ideas for how to implement a legato mode that will adjust note attacks + sample offset when notes are slurred. This would also help reduce the perceived late note attacks on blown/bowed instruments. It's just a matter of finding the time right now, but I would love to make this happen.
In reply to No doubt there are more… by s.chriscollins
Ok, at least let's aim for https://musescore.org/en/node/290272 in the short term. Let's involve https://musescore.org/en/user/1675965 as he has worked with the synthesizer when adding single-note dynamics and said he might have some ideas on how to do this; furthermore, let's also look at whether it makes sense to catch up to mainstream FluidSynth - maybe contributing our improvements back along the way. Browsing the forums I saw there was some movement in that direction a few years ago, but it sizzled out? I assume due to lack of developer time? Please direct me a to a better venue for discussion.