The sounds in Musescore 1 were better

• Aug 28, 2014 - 23:29

The new update is pretty good, but is there a way to load the sounds from version 1? The violin on version 2 has a delay to it, and it's really messing up the song I have (it has 32nd notes). The sound from version one sounds better


A delay? There shouldn't be. Could you post the score you are having trouble with? And maybe do a a "save as to MP3 and post that too so we can hear what is happening> The soundfont used in 2.0 Beta 1 is actually *MUCH* better than the old soundfont, so something must be wrong on your system if you are hearing problems. But FWIW, you can just tell MsueScore to use the soundfont from 1.3 if you really want - go to Display / Synthesizer and load the old TimGM6mb soundfont from whever you have MuseScore 1.3 installed.

In reply to by ChurchOrganist

Have you both (nsj_3.14159 and ChurchOrganist) change the default setting on the soundfont to catch what you wanted in Musescore 1.3 ?
There were somewhere a discussion in which some people had change the setting to match their earings idea, and have to do it again with the new soundfont....

default settings are better I think in Musescore 2.0 (but for now I only tested on "lento" tempo score...)

In reply to by ChurchOrganist

BTW, it's FluidR3, not R6.

I'm not saying FluidR3 is*perfect*, but it is definitiely an order of magnitude better than TimGM6mb for virtually every conceivable instrument. And I've yet to hear a GM-compatible soundfont that is better all around. Some have better sounds for some instruments here of there, but as far as I have been able to tell - and I've done a *lot* of testing - FluidR3 remains the best all-around GM soundfont freely available.

If people have other suggestions of soundfonts that could be considered, that's a good topic for discussion. But I some of the ones that have been discussed lately - GeneralUser, Arachno, Tears of Heaven, Sonatina - are very clearly inferior to my ears. Well, Sonatina is somewhat better for some many the instruments it does have, but it is not GM-compatible and is missing far too many instruments to be a consideration.

In reply to by Marc Sabatella

Piano is my instrument, and I have negligible experience with arranging or transcribing for other instruments. The 'Yamaha Grand Piano' in FluidR3 is awful to my ears, and 'Bright Yamaha Grand' is worse. (Some people would argue that 'Bright Yamaha' is redundant!) I actually think that Electric Piano sounds more like an acoustic piano, which was ... unexpected.

In reply to by [DELETED] 448831

"Awful" compared to what though? Surely it is light years better than TimGM6mb. There are indeed specialized piano soundfonts that sound better for that specific instrument, but again, given that we need a GM-compatible soundfont with a full complement of instruments, I still say, show me something you think is better that meets the requirements (free, GM-compatible, reasonable size) and then we can talk.

BTW, "Electric Piano" in FluidR3 sounds to me like a sample from the Yamaha Electric Grand, an instrument that does indeed sound more piano-like than a Rhodes or Wurlitzer or similar brand. But still, not as m,uch like an actual piano as Yamaha Grand Piano. The Bright Yahama Grand sound is, I would agree, pretty harsh.

In reply to by Marc Sabatella

'Relative to what?' is a valid point, and I suppose I would agree it's an improvement over TimGM6mb. In any case it certainly serves its purpose, and I didn't mean to overstate the importance of this topic. (For me, playback is mostly limited to 'proofreading' for errant notes - musical typos.)

In reply to by Marc Sabatella

The sound of the soundfont has become very well! Become feeling creative!And I'm looking forward to more and more.but I feel not clear that built-in synth sound. it cause might synthesizers.

I've tried in some software using FluidR3 GM sf2(I was by myself download for comparison).No effect conditions. but MuseScore was as wet as effects(can I no effect export). and there are also differences in the other. I've prepared the figure and comparison of sound. I belong to the attached zip.

Attachment Size 1.61 MB

The problem isn't that there's a literal delay, but each note crescendos for the first part, and that's causing my 32nd notes to sound quieter. Because of this, the higher notes sound like they are completely dead. I can't upload the song to musescore or save an MP3 file of it. You can go to my profile and listen to the song (Vivaldi's Winter Violin Part) with the 1.3 sounds, and then I will try to upload the version with tremolos (2.0 sounds).

In reply to by nsj_3.14159

Ah, OK, it's a specific different in the attack of that sample that happens to sound worse for this particular track because it relies so heavily on very short notes. That much is plausible. But in general, the violin sounds *infinitely* better - the violin in TimGM6mb was pretty terrible. One of its worst qualities was that it had a very harsh attack, but this I guess worked in your favor on this one particular piece.

In reply to by Marc Sabatella

I have to disagree there, Mark. Almost every single instrument, but especially strings and plucked strings were WAY better in TimGM6mb than in the new glop. I can't use that now because of the pestilential whip cracks in 2.1, but I am actually running one computer on the earlier version to get the sounds I want from TimGM6mb and another on 2.1 to get the lute tabs. If I could get another soundfont that sounded just like TimGM6mb but without the whipcracks and running on 2.1, I'd be there like a shot. The new soundfont just doesn't handle short notes worth a darn, and most of the plucked strings sound like playing a piano with the pedal down.

In reply to by Charric Van de…

Well this is very subjective - different people have different preferences. But the overwhelming majority prefer FluidR3, which is why we changed. But people who prefer the smaller TimGM6mb are welcome to use it.

Have you tried a nightly build of 2.2 to see if the audio glitch you are experiencing is gone? There have been fixes made, things shouldnbe better.

In reply to by Marc Sabatella

You're a Lord, Marc.
I just installed the 2.2 Nightly Build and spent a little time moving over the custom instrument.xml file and all the soundfonts I've been trying. TimGM6mb and no "whip crack!" I can finish the album I've been putting together without the sound suddenly changing to that of a cheap Casio in the middle of the album, now. I put together audio from Elizebethan primary sources for people to listen to what they used to write, and a clean sound like TimGm6mb really helps serve the purpose. I was afraid I was going to have to settle for the least objectionable alternative. Can't wait for the stable versions. I'm going to start using the 2.251685 Nightly Build exclusively until the new one's out, I'll be sure to let you know of any noticeable problems with it.
In case you may be curious, I put all the albums I arrange under my stage name of Charric Van der Vliet at this site:

where most of it was produced using MuseScore. As a side note to that, the most recent, and thus, to my ear the best arrangements, are at the bottom.

 Once again, thanks for all the hard work and patience with us "Who Moved My Cheese?" people.

C. A. Powers
AKA Charric Van der Vliet

In reply to by jim.newton.562

Right, this thread is about the change in default soundfonts. If you are using a non-default soundfont and keep it the same, you shouldn't notice much difference. There will be *some* though - different reverb processing, and I think maybe better handling of the attacks and/or releases which might be noticeable on drum sounds? Seems there was an improvement made there but I don't know the details.

This is likely to be the problem of the Fluid Soundfont. I had that same problem back then when I used the same Soundfont in the old MuseScore. And yes, now that Fluid is being used as the default, I guess there needs a new feature urgently:

- Playing an attack on bowed instruments on notes that are not a part of a slur, and on the first note of a slur.

I've just been looking at this problem in Viena, and can confirm that the problem is the Violin samples being used, which are all of a long sustained bowed note, for which, of course a violinist would use a bowing technique which shaped the sound in a pleasing manner.

The problem is, of course, that different lengths of notes are bowed quite differently from each other, and this is just one example of a number of indications that the original author didn't understand the principles of sound design properly.

It is going to be difficult to rectify this in the soundfont, as currently velocity is the only way of controlling dynamics in MuseScore, so we can't use velocity splits to control which sample is played. GeneralUserGS uses two violin samples played together, the first providing the attack on the string and the second providing the sustain element achieved with bow pressure.

One way forward would be to email Christian Collins and ask for permission to use his attack sample(s), then incorporate it into the Fluid Violin instrument.

In reply to by ChurchOrganist

I'd hate to lose the wonderful legato quality we get now, though. It's one of the things that to me is so much better about the FluidR3 violins than almost any other. Winds too. It's basically impossible to make slurs sound like slurs with any other soundfont - for strings as well as as winds. If having more realistic sounding legato passages means a softer attack on articulated passage and slightly quieter tromolos, I'll take that over harsh attacks that prevent one from ever creating legato passages any day of the week and twice on Tuesdays.

Of course, the real solution is for MuseScore to support true slur playback - so you get the attack by default, but not when you place a slur, the attack is suppressed. I actually tried to implement this using the legato controller but it didn't work, and I came to the (perhaps premature) conclusion that the underlying fluidsynth library simply doesn't support it, except perhaps in Mono mode.

In reply to by Marc Sabatella

According to FluidSynth's MIDI IMplementation chart it currently does not recognise the Legato Controller (#68).

It does, however, support the Expression Controller (#11) so if we can add the ability of MuseScore to send controller #11 messages then we could use velocity splits to control attack.

In the meantime we could add a Violin preset on a different Bank with a faster attack.

I will look into that when I have some time.

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