MuseScore_General.sf3 & MuseScore_General.sf2

• Jun 16, 2020 - 13:57
  1. What is the difference between MuseScore_General.sf3 38959kb and MuseScore_General.sf2
    210525kb are the sounds the same quality and that the .sf2 is uncompressed resulting in faster access?

  2. GeneralUser GS MuseScore v1.442.sf2 30545kb what is this like quality wise compared to above

  3. i see some samples are stereo and mono, how do you hear the difference. as once you have say 7, 8 + tracks
    its hard to hear.


Comments

  1. SF3 is using OGG compressed samples. The sounds should be the same, except for possible compression artifacts. The SF2 one loads much faster on MuseScore start, but of course tkes longer to download and needs more space on disk

In reply to by Jojo-Schmitz

ah ha, i was getting artifacts, easy to hear, but difficult to pinpoint,
because i would export from Musescore in WAV or Flac 48000K, then open in audacity
in some places the artifacts were in the same place then on other recordings same piece, in other places.

I then reduced overall volume in Musecore before exporting this did seem to help ( i think)
then. Then used the .sf2 GM which has helped.

  1. With much difficulty.

I'm not convinced of the usefulness of stereo soundfonts, (since most instruments are close to being mono sources from a typical listening position), but it can make sense to pan mono sources to create a stereo mix. Perhaps there are special situations where a stereo font is useful.

In reply to by beanroti

No, I said that instruments are mono. :-)

It makes sense to mix a set of mono tracks into stereo, e.g. to sound like a band spread across a stage but I'm not sure why a single instrument track would be useful in stereo. I find that a stereo mix of instruments sounds clearer than a mono mix.

In reply to by yonah_ag

My humble opinion is about a typical listener, sitting at more than 5 meters out the central orchestra director stage point...

Is that people listen all the orchestra instruments sound with a real wide spacing stereo sound? ??? NOT!!!

What about the, sometimes, excessive reverberation of the room? ???

Can the listener really know and distinguish what is the direct sound (from the instruments) and what is the "ghost" bounced sound (from the room walls)? ???

If you think about those questions... A soundfont file with stereo sounds is... I don't know what is the right word, here... My idea is: Not useful at all!!!

Yes, stereo sound is a very good technology. It is a very nice to test our pair of "microphones" (our ears). But... In the real world... The things don't sound as the Phase 4 Stereo old vinyl records from 1960's decade.

In other words...

In a real stage, only the director can hear the full orchestral stereo sound effect.

In a real stage, only the pianist can hear the full piano box stereo sound effect.

In a real stage, only the organist can hear the full pipe organ structure stereo sound effect.

So...

In reply to by beanroti

I have tried quite a few free soundfonts in the pursuit of some nice guitar sounds and these have all been SF2. Soundfonts can take a few moments to load in MuseScore but once they are ready I haven't noticed any sluggishness, even on my ancient computer. I've had up to 8 soundfonts loaded in the synthesizer with no problems - although you do end up with a very long selection list in the mixer!

In reply to by yonah_ag

yes thats my take as well, but i sluggishness only if you have say Steinway pianos 300mb, + orchestral 300mb
etc.

Funny enough once loaded Musecore 2.3 Win 7 64b 4gb Ram ran fine, perhaps a little slow on exporting a .wav audio of +/- 25mb about 30 seconds but it was ok.

All said 3.5 Beta is fast all round. except if using .sf3 once loaded ok, but when listening to Playback of a .wav or .flac you can hear little distortions or artifacts. so that as Jojo said is the .sf3 being uncompressed.

MuseScore_General.sf2
210525kb i am happy with it.

but is there a way to reduce the number of instruments ie take some out. as adding other soundfonts like i do say 8 or 9 the instrument list gets very long, and its a pain continuous scrolling & tweaking.

I did look at Vienna polyphone program but i could not get any where with it.

In reply to by bottrop

I downloaded the free 64 bit Windows 10 Polyphone app, opened it, opened a soundfont I wanted edited, and ...
there is no "Tools>Global>Remove unwanted elements menus. I don't know how to do what you say. could you reply with a screenshot perhaps? Thanks

In reply to by fsgregs

i still use Polyphone 1.9, because it works better for me, newer versions look like they are designed for telephones.
i had a look at the screenshot of the newest version at the Polyphone website and i guess the Tools are now behind that toolbox picture (apparantly a service for people who cannot read)
for deleting Presets you can also use the Delete button on your keyboard.
regards bottrop

  1. The soundfont creator took care that the presets where it's most important e.g. the string sections are in stereo. But inside Musescore you can bring out the best of your music by tweaking the reverb and chorus effects. These have a huge impact on the stereo image.

In reply to by Daniel Sabados

Daniel, are you just talking about MIDI files? The Musescore User guide says that reverb and chorus adjustments have no effect on playback within the MuseScore synthesizer. Since I do not have any other way to play back a score except using the built-in synth, do reverb and chorus change what I hear when using the Synth, or is the guide correct, and they only affect playback using an outside MIDI device?

In reply to by fsgregs

Inside the today MuseScore synthesizer THERE IS a good reverberation control, when we use the "Zita1" virtual reverb device ("See" menu, "Synthesizer" option, "Master Effects" tab).

The controls inside the MuseScore Mixer are absolutely useless. They don't work.

And... The great problem is about the MIDI messages we need to put INSIDE the standard MIDI format files to control reverberation and other parameters. MuseScore doesn't let us to directly manipulate these kind of things, yet.

We have to wait to the near future version 4.x.

Maybe...

In reply to by jotape1960

Thanks. That answered it. Too bad the Master effects menu does not have a chorus effect. I am trying to fill my music with 10 - 15 violins as would be heard in a full orchestra, but when using the strings" soundfonts available, they at best sound like 2 or 3 strings, not 10 or 15 as would be present in a full strings section. Chorus effects would have been nice.

In reply to by fsgregs

Hmm, that shouldn't be, the "stirngs" samples in most GM soundfonts should be recorded from a larger group than that. Are you sure your ears aren't playing tricks on you? The default soundfont seems to be a smallish section to me, but surely other GM soundfonts use larger sections in their samples - one should need to resort to hacks like chorus, which really won't sound like more violins as much as it will sound like they have intonation problems. Which soundfonts did you try?

In reply to by Marc Sabatella

Marc: I have been trying a LOT of soundfonts to find a really beautiful strings font. I've installed all of the Orchestral soundfonts in the MuseScore user guide, plus at least three others from the web. Every strings section I try and play back in the MuseScore synth sounds tinny and sparse in voices, maybe 3 or 4 violins at most. Maybe I am expecting too much from the synth or the soundfont, but I assumed that if microphones recorded 10 or 15 violins in a strings section playing and put them into a computer file, that the produced sound should sound like that ... a set of tones that brought a smile to your face at how melodic they were ... even when playing a single sustained note. To mimic a real symphony, do I need to put 3 or 4 strings tracks into MuseScore, all playing the same note ... to hear that sound, or should a single note from a 10 violin strings soundfont sound that beautiful?

In reply to by fsgregs

Well, if the person creating the soundfont recorded 15 violins, then that's what you'd get. But, it would be really noticeably louder than any single isntrument, and probably most people have therefore then artificially turned down the volume causing it to not mix as you'd expect. But in theory it's possible, I'd just keep doing web searches maybe looking specifically for that. You only need one sound, basically, not a whole GM soundfont.

But, to mimic a symphony, certainly you should have 1st & 2nd violins separately, also violas, cellos, and contrabasses, each as full sections.

In reply to by Marc Sabatella

Ok, I guess. I wish there was some spectacular soundfont that everyone is raving about that really sings its strings section, preferably a free font, but I would even pay for a really good one. Do you use one in particular?

Two quick questions:
1. Is the MuseScore Synth just as good as any other, or is it just a basic player that I should consider upgrading to something else? It would not be worthwhile to buy a top quality soundfont only to discover that the Muse player doesn't do it justice.

  1. I realize you would not want to favor any one person or score, but on the MuseScore site, has anyone uploaded an orchestral "strings" song that would really teach me how to create a blended strings section, with 1st and 2nd violins,violas, cellos, and contrabases? I have no idea at all what to do. I don't need a fancy song, just something to show me if I should include only single notes or chords, how to combine the voices, etc. I checked the site but don't really know what I am examining.

Thanks as always. It is comforting to be able to communicate with you. I don't feel so alone in learning the complex topic of music theory, and MuseScore particulars.

In reply to by fsgregs

Music is really moving away from SF2 soundfont technology so I wouldn't invest any money. It started life back in the early 1990's so it's had a good run but is now getting replaced with VSTi and DAW technologies.

Check out this thread for some impressive violins:
https://musescore.org/en/node/312341
Apparently these mp3's started life in MuseScore 3 and were then finished in Reaper. There are indications that Musescore 4 will go this way – which should be good.

If you can't wait then you could consider making your own violin soundfont with an app like Polyphone and some good samples.

https://www.polyphone-soundfonts.com/
https://philharmonia.co.uk/resources/sound-samples/

In reply to by fsgregs

I am satisfied with the default soundfont for my purposes, so I rarely investigate others. But I do plan to do a "soundfont shootout" as part of my MuseScore Cafe series next week (https://www.youtube.com/playlist?list=PLpx1s2WkyujZEH_XzG_3LE8jVNzttkJ9c).

For playing SoundFonts, MuseScore is as good as any program I suppose, but as mentioned, for more realism you need to look beyond this technology. Using DAW software with VST instruments etc is the sort of path people take who want to really push this to the next level. MuseScore can interface with these programs via JACK or MIDI Out, but a lot fo the work going on towards MuseScore is in incorporating more of this type of technology directly. So hopefully within the next year there will be improvements.

As for how to write effectively for strings or other ensembles, well, that's a huge topic ("orchestration"), the subject of college textbooks etc. There are tons of great orchestral works on musescore.com, but why not start with some of the very best:

https://musescore.com/openscore/scores/5733014
https://musescore.com/openscore/scores/4145221
https://musescore.com/openscore/scores/5271023

In reply to by Marc Sabatella

Thanks. I understand. By the way, I watched your 1 hour YouTube MuseScore Cafe chat-cast on writing for string instruments. Very informative. Much appreciated. If I grasp what you said, if i want my song to sound more like a full symphonic string section, I need to add perhaps 3 or 4 separate staves of violin strings, 2 or 3 staves of violin 2 strings, 2 or 3 staves of viola, and a few staves of cello and contrabase. I can do that, but do I need to de-tune each a few staves a bit (say to 439 hz or 441 hz) so they sound like separate instruments, or should all of them be exactly smack-on 440 hz?

In reply to by fsgregs

Actually not that complicated - the standard is 2 violin parts, single viola part, a single cello part, and a single contrabass part (which is very often the same as the cello but sounding an octave lower). So it's really just SATB writing - violins 1 & 2 are S & A, viola is T, cello & contrabass are B. So then you write for that grouping as for any other section or for choir or for piano, trying to create good voicings and voice leading etc.

Just be sure to make each instruments the "plural" version - eg, don't add "violin", but "vioilns", etc. So there will be "violins" instruments, one "violas" instrument, etc. Thus all the first violins are playing the same note, all the second violins are playing the same note that is different from the firsts, etc. No detuning or anything like that. That just makes things sound out of tune. If you really think the balance is off, feel free to tweak it in the mixer, but in my experience it's pretty accurate by default.

In reply to by fsgregs

Please!

Remember this is a machine.

There was not, there is not and there will be not a better sound than the human players orchestra group.

To make a very high quality soundfont file, some guy needs, at least, three very difficult things:

1) Better real instruments and human players (real musician). Where in the universe any of that is for free?

2) Very high quality audio studio (microphones, audio interfaces, audio recording software). With the very noticeable exception of the audio recording software (Audacity, as sample), microphones and interfaces are still at a high prices level.

3) A very high amount of time, hours, and hours, and hours of experimentation.

Maybe, some day we will have a very near "natural" soundfont file, FOR FREE, to these machines.

For now... It is what it is.

In reply to by jotape1960

Jotape: I totally understand and sympathize. I am willing to buy a beautiful, realistic soundfont with a section of strings and violins, provided it does not break the bank. Could you recommend one for under $100, if possible?

Do I also need to get a different synthesizer, or is the one in MuseScore able to reproduce those digital files as well as any other? Marc mentioned something about DAW software and VST instruments. Are they that much better? If so, do I need to purchase all of that software as well? Sorry, but I am ignorant of this whole topic, but am willing to learn. Thanks.

In reply to by fsgregs

Some very good soundfonts are available here, but maybe you have already tried these:
https://sites.google.com/site/soundfonts4u/

I don't think that a different synthesizer is an easy option unless you want to customise MS. In any case, I have compared soundfont .wav samples with the output from MS's Fluid synthesizer and they sound pretty much identical to my ear.

Try before you buy with any soundfont.

VST
https://en.m.wikipedia.org/wiki/Virtual_Studio_Technology

In reply to by fsgregs

I have just checked Polyphone and you can probably build a multi-violin sound quite easily using your preferred soundfont as a starting point.

You could build a single preset made by repeating the same violin instrument but vary the chorus, pan and volume envelope to achieve a 'larger' sounding violin section. This way you would then be able to add the preset to a score as a single sound in the mixer.

In reply to by yonah_ag

Whoa! I suspect I am a far way from trying to build a soundfont. I know what chorus, pan and volumes are, and I suspect I could work on building a rich strings section in Polyphone, if I could ever learn its workings, but I really would love to find one already done. I am taxing my intellectual limits as it is. I will keep looking for one already done, but if I can't seem to find one, I can try to edit the ones I've got. Thanks for the suggestion.

In reply to by fsgregs

I only suggested building a single preset in an existing soundfont, using that soundfont's existing instruments and existing samples. Intellectual ability comes into play in considering: how and why does a group of violins sound different to a single violin. Loudness? Left-Right spread? Front to back spread - does this affect volume and is there a subtle but detectable time difference between further way violins reaching your ear. Probably other factors too. There are controls in a soundfont which would help you to model these attributes.

You already know what chorus, pan and volume are so this is a good start.

Did you try any of the GM soundfonts on Soundfonts4U?

In reply to by yonah_ag

Yes and no. I downloaded and installed several Soundfonts4U fonts, and tried them out in MuseScore. I was shocked that a few of them, labeled as great fonts, sounded actually rather poor in the MuseScore synth. They were either tinny or harsh or sparse. So far, I have found only 1 grand piano soundfont (Salamander) that rings well, and am using it (although it has one annoying note in my score which refuses ... refuses ... to hold a sustain??? So far, I have not found any beautifully melodic violins or strings. I will keep looking and listening.

To edit one of these fonts to manipulate the strings to create a wider, more symphonic sound, may be possible in Polyphone, but I have to learn how to do it ... what tabs to click, what to delete, what to drag, how to save, how to copy and paste, how to use the pan, chorus and volumes, etc. Of course it is doable, but it scares me. The learning curve for Polyphone is not small, and it will take me quite an effort to do it. It can't be that easy or someone else would have already done it. What I have been hoping for is that someone who already understands music and fonts, has demanded the fidelity and already done it. So far, no luck!

Given your expertise and demanding ear, what has been your most beloved strings and violin solo soundfont? I found no such separate fonts on Soundfonts4U involving strings. Are there any I missed?

In reply to by fsgregs

I only play solo guitar, (nylon), and I like:

• "Acoustic Guitars JNv2.4" - Soundfonts4U
• "Classical Guitar 'Ichiyanagi'"- Hedsound
• "FS-Gibson Les Paul I5" - HedSound

Sorry that I can't help with orchestral strings.

Odd that one note refuses to sustain. It could be something score related, (in which case one of the MS experts should be able to help if you start a new thread and upload a sample score which shows the problem), or it could be in the soundfont. This latter seems unlikely but you can change the sustain in Polyphone.

I'd make a very short score and post a new query.

I have the guitar samples from Philharmonia.co.uk and will make a font when I get some time.

In reply to by yonah_ag

The sustain is a pedal sustain using the MuseScore piano pedal marks. It is only occurring on one note (a C3), and only when using this particular font. All the other notes played by that soundfont so far, are sustaining and sounding fine. When I use the MuseScore grand piano font, the sustain works fine, so it is not the score.

In reply to by fsgregs

A problem inherent with most window and string instruments is that they are capable of such a wide variety of sounds. So you might have an absolutely perfect sample of a violin playing a long sustained note that works perfect when playing long sustained notes, but sounds ridiculous when playing short accented notes, and vice versa. Also, the first note of a bowed passage will sound different from the rest, and so will the last. So really, there is not going to be one violin sound that does exactly what you want all the time - getting something that sounds realistic in any musically meaningfully context will requires lots of different sounds that you switch between as needed. It's a big undertaking that you would normally need to accomplish using DAW software to do those sorts of edits. Realistically, depending on the project, hiring an orchestra right be an easier way to go!

Regarding the non-sustaining note in your piano part could be an issue with the soundfont, but just as likely an issue with your score, like the same pitch occurs in any other voice or staff and cuts it off. if you post the score we can understand and assist better.

In reply to by Marc Sabatella

Marc and Yonah: As always, thanks for the reply. The non-sustaining note only occurs when using a downloaded soundfont called "SalamanderGrandPianoV3.sfz", and so far, it has only occurred on one particular note ... a C3 note playing on a treble clef. All the other piano notes in the same range play fine, as does the resident Grand piano on the MuseScore .sf3 font. So, it is not due to the score. If you are willing to download the sfz file to try to duplicate the problem, it is one of the piano fonts referenced in the MuseScore user guide. I really like the sound of the piano. It is rich and very melodic, so I don't want to abandon the font. You might like it also.

Regarding the strings fonts, I understand I cannot duplicate a real symphony, but I've been hoping that the violin solos I've found, sound really close to a good violin. Problem is, when I play them in MuseScore, including the decent violin in the default MuseScore font, the biggest problem I have is tremolo coming in immediately, and a tone that needs some mellowing out. Maybe I'm comparing them too unfairly to a real high quality violin, but my dream has always been to outfit my computer as a kind of music studio, with it playing multiple instruments for me. I am now disabled healthwise, and this has been my hope, but so far, the sound has just not been what I'd hoped. I know that I can't expect such performance from a free software program and free soundfonts for pete's sake, but ... In that regard, is there a date when DAW software may become resident in MuseScore?

I will continue to explore strings fonts. There are probably 20 more out there I have not tried, either alone or as part of a GM font. I'll keep working on it. Further, if I can figure out how to edit the tremolo timing and change the tone a bit via an equalizer in Polyphone, maybe I will be able to use it to edit the MuseScore violins and strings. Hell, with COVID keeping me glued to my chair, I guess I have the time.

In reply to by fsgregs

FINALLY, FINALLY ... I found a beautiful violin sound. I've been searching for weeks. I sought a violin that is expressive, very melodic, not harsh or shrill and really sounds like someone playing the instrument live. I found one.

It is called "Violin (all around)", and is one of the instruments in the free "Essential Keys - sforzando-v9.5.sf2 soundfont, available at https://sites.google.com/site/soundfonts4u. The strings section is not the best I've discovered, but it is reasonable. I have not checked out all the other instruments, but the grand piano is quite good. For a very nice strings section (multiple strings), I have discovered a .sf2 strings font called **"Strings_dsx_super_orchestra.sf2"++. it has four strings fonts, each of which are very nice. It is available at: https://www.polyphone-soundfonts.com/documents/17-bowed-strings/314-str…

In reply to by fsgregs

:-)

So a project for you in the future is to assemble your own GM soundfont from your favourite instruments across several other soundfonts. This is just a copy/paste/delete exercise in Polyphone. It's not necessary to do this for MS but it keeps the sounds drop-down list shorter.

In reply to by yonah_ag

OK, I posted on the Polyphone forum to get some basic directions on how to open a large symphonic .sf2 soundfont in Polyphone, then copy and paste only the few instrument I wish into a "new" .sf2 file, and save it. Apparently, it must be easy to do in Polyphone, but damn if I can find the directions on how to do it. So far, no one on the Polyphone site has explained what I must do. Can you point me to the step-by-step directions on how to copy a sound (instrument, preset???) from an .sf2 file and paste it into a new one? is it in the manual, or on YouTube???

HELP!

In reply to by fsgregs

Here's the thing:
It has been pointed out that the only thing that sounds like a real player is....a real player. Anything else is a recording of a real player. So the best we software users can hope for is a file that sounds like a recording of a real player. As such that file is subject to all the short comings of a recording. The recording will sound different based on what the source is. Different headphones sound different. Different speakers sound different. When I listen in my car, it sounds different. The school sound system sounds different. So which one is correct? I suspect that because it is recorded sound, all of them are correct. And just because a font sounds "bad" in my headphones, doesn't at all mean it will sound bad in other sources.

It also makes a difference what the font is trying to reproduce. Expecting an orchestra font to reproduce uncharacteristic music, isn't going to work.

What makes a live performance sound melodic and expressive? The player, of course. Can any font reproduce that? Not by itself. Can notation software reproduce it? Not by design. Together, along with many hours of work on most every note, some acceptable results might happen. But in general, you need an expensive DAW, and a few thousand dollars worth of instrument libraries to make a good recording that sounds like a recording of real players.

I believe that we have to know the instruments we are writing for. If I write for piano, what brings that music to life? Well, a good player can make poorly written music less painful to listen to. But even the best font won't save what I've written. It becomes easy for me to blame the font. I compose for orchestra. I know something about how to do that. But there is more to it than that. I have to pan my orchestra correctly to get the full sound I want. My EQ has to be just right. What I'm listening through has to be the best it can. I have to take the time doing what I can to adjust playback to get what I want.

I have tried several different fonts in MuseScore. They all fall short. Consider that Virtual Playing Orchestra is less than 700 MB. Some are bigger. Most are way smaller. Now, consider that paid notation software libraries start at 35 GB and go up from there.

But personally, I use the default General HQ font in MuseScore. I've added a solo horn and solo trumpet. This combination works best for my composing. I don't always like the 1st violin section sound, so I sometimes use the 2nd sound for both 1st and 2nd violins. It just depends.

In reply to by bobjp

I'm so glad that I only arrange for solo guitar. No panning, no orchestral soundfonts to consider, no instrument levels to mix, etc. just a few decent guitars sounds needed – and I found some in the public domain. This is enough to produce acceptable score playback but, as you rightly say, it takes real performers to bring a score to life.

In reply to by yonah_ag

I play what used to be called Folk style. Some strumming and finger picking. Not classical. MuseScore doesn't have very good strum pattern possibilities. Or sounds. BTW, I built from scratch the steel string guitar that I play. Kind of fun.
If you were to write a trio or duet, panning could be fun.

In reply to by bobjp

Nice one, building your own guitar. I wouldn't know where to start. I've lowered the saddle on nylon string guitars but nothing more complex.

Panning is actually a non-starter for me as I only hear in one ear so I have no directional hearing. Sound doesn't come from any direction, it just sounds! If someone calls my name then I have to do a 360 look-around to try to locate them. I have no concept of surround sound and I need a mono signal to feed headphones, otherwise the mix can sound odd.

Strumming patterns and MuseScore sounds like hard work. Maybe MS4 will help. I'm an ex-Guitar Pro user and came over to MS because of the amount of control it gives me over score styling - even just for TAB.

In reply to by yonah_ag

yonah_ag

Bummer about your hearing. I am slowly loosing mine, altogether.

Lowering a saddle is nothing to sneeze at. Too easy to make a mess of it.

As for building one? It is as difficult or easy as you make it. My first guitar didn't look great, but sounded good enough. The real value is the realization that I'm "making music" in most every sense of the word. I'm not a great player or builder. But that's not the point.

In reply to by yonah_ag

I would never consider myself to be highly skilled. Let's put it this way. The book I learned from started out talking about how to setup the workbench I would need. Workbench? I didn't have a workbench. I would have to build one. Never built one before. And so it went for the entire project.

In reply to by bobjp

WHEW! I am not alone then in wondering why so many soundfonts don't sound very realistic. I thought it was just me. Thanks for explaining. BTW, I had no idea that a realistic font can exceed 35 GB. No wonder the free ones fall rather short in sound.

In reply to by fsgregs

See also my previous explanation. Even with a 35 GB soundfont, you'd still struggle with all the issues I mentioned., The soundfont might have a beautiful recording of a violinist playing a single A4 as a whole note at mf, another at f, etc. Great, now if your music consist of nothing but A4 whole notes, you're set :-). But, what if you actually want that note to be played marcato? This will not simply be that same note played shorter or louder - the bow attack and speed will be different, etc. Or, how about a passage of four quarter notes? If the sample was of each one being bowed with a new downbow stroke, it will sound equally unnatural to hear them simply played sequentially - a real violinist would likely play all four notes on a single bow stroke, thus changing the tone quality very noticeably. None of this is likely to be captured in the soundfont, and even if it were - by having a dozen different violin instruments for each type of sound the violin can make - it would then be up to you to manage switching from one sound to another at the appropriate moment, by adding staff texts and channel switches, using an external DAW, or whatever.

By way of comparison, consider Siri, or Alexa, or Google Assistant. Each was recorded from a real human being speaking real words and real syllables. If all you need is one word, Siri saying that one word will be as convincing as any other recording of one person saying one word - hard to screw that up really. But assemble a bunch of these individually spoken words into a sentence, and it's glaringly obvious it's not real. That's because. a. real. person. doesn't. speak. by. saying. each. word. one. at. a. time. Not even by using AI to take individually-recorded words and attempting to combine them cleverly to sound more natural. Real speech sounds qualitatively different than synthesized speech, and although AI technology gets better and better, it's still not there, and that's with the biggest companies in the world spending millions and millions of dollars on the effort.

Needless to say, no one at MuseScore is spending millions and millions of dollars on the AI technology necessary to get individually recorded notes to sound natural when combined. I don't care how good the original recordings of the individual notes are, it's still going to suffer when you try to assemble those individual notes to make music. Improvements come incrementally, but if you think about how far Siri is from sounding like a real person, don't expect MuseScore to come closer to sounding like a real orchestra.

In reply to by Marc Sabatella

And you know what? That’s okay.

MuseScore is a musical notation software. If it can also play sounds, okay, that’s expected. If it has a great “general use” soundfont shipped that’s compact enough to work on even somewhat older devices, etc. wonderful, that’s all I need. Sure, some of the samples could use tweaking, but it’s good in the soundfont system. (Even the dreaded Choir Aarghs got improved some by the magician soundfont master.)

And if I want a multi-million dollar emerging technology, there’s probably offers for that, too; I read about NotePerformer somewhere, that’s supposedly a start. (For my vocal scores, I don’t, though.)

But that’s not MuseScore, and the beauty is when these things can work together. (I read about NotePerformer not being able to work directly with MuseScore, just per MIDI, but that that may change. But this is not about that particular piece of software, more the generic concepts.)

Thanks Marc for explaining this so well and in detail!

In reply to by mirabilos

You're welcome :-)

Of course, some will say I'm overstating my case, and that better results than what MuseScore currently produces are possible. I agree - but again, the real gains won't come from finding a better soundfont, but in going beyond that technology. I'm almost completely ignorant about NotePerformer, but people definitely seem impressed, and if MuseScore 4 ends up being compatible with it, I'm sure that will make a lot of people happy.

In reply to by Marc Sabatella

NotePerformer...meh. It is a sound library and a set of instructions for the software to playback a score in a more realistic way. Whatever that means. I've had someone use NP on one of my scores. I've also spent some time on their website. All I can say is that their idea of more realistic is not necessarily the same as mine. Yes, some people really like it. I haven't been impressed so far

In reply to by fsgregs

Just in case...

Inside Ubuntu (Linux) there is a command called: "sf2extract". This command is used inside a "Terminal" (something like the old MS-DOS panel, before MS-Windows).

That command gives you the option to extract all the sound samples from a soundfont file to individual audio (wav) files, sample by sample..

The only one problem is about you will get only the samples, without the internal links to some instrument preset. So... You will have to manually rebuild those association presets.

Maybe, this command will be useful to get samples from different soundfont files and to build a new one soundfont with just your most loved sounds... maybe.

In reply to by jotape1960

Jotape1960: You hit on a possible solution. I don't have Linux and doubt if I will install it on my Windows 10 machine, but ... I suspect i can do something like this in Polyphone. If the "loud" wav files in my violin font can be replaced with something less harsh, I may try to do that BUT ... I would have to learn how to edit lines and commands within the .sf2 text file to call up different files, find better violin wav files and substitute ... it would be a LONG learning curve that I probably will attempt ... someday. Maybe in the next few months.

Until then, I will just try to keep my violin notes below a velocity of 90 in MuseScore, in relation to the other instruments.

In reply to by yonah_ag

Absolutely, but only up to a volume/velocity in the MuseScore synthesizer of 89. As soon as a velocity of 90 or more is placed into the score ("f", "ff" or manual entry), the samples in the font change to a harsher, more grating sound, and the sustain time of each note changes to end abruptly. The "beauty" of the sound goes away. To fix this, either I have to keep the volume of the song below 90 throughout, or I have to find different sample wav files in the .sf2 soundfont that do not have those flaws.

In reply to by fsgregs

A soundfont can use different samples for different MIDI volumes. I'm guessing that velocity 90+ use a "harsher" set of samples. You could confirm this with Polyphone and, if it proves to be the case, you could use the other samples for all velocities; or even use a mix of samples. They probably aren't flaws but are by design.

I have done this with a guitar soundfont where I stopped using the samples from the bass E string and instead used the bass A string samples. The samples get adjuststed for pitch and I managed to take a metallic edge off the lowest bass notes.

In reply to by fsgregs

So the preset Violin (all around) is made up of 4 instruments:

poly01.png

Instrument P is used for all pitches and for velocity range 0-89
Instrument F is used for all pitches and for velocity range 90-116
Instrument X is used for all pitches and for velocity range 117-127

I don't know what instrument Rel X is for yet.

If you look at the Instruments section for each of the above you can see, (and play), the associated samples. Sounds like P is Piano, F is Forte and X is Extra Forte. With F and X you are hearing extra bite - presumably because the player has to bow more strongly.

You could take a copy of this soundfont to play with and set P to use range 0-127 and remove instruments F and X from. I would be tempted to copy the violin to a new soundfont on its own, (at least initially), as you are then not working with a 1 GB file.

In reply to by fsgregs

Here's my copied violin in its own soundfont with the smooth samples covering the whole velocity range:

poly03.png

Just use copy/paste on a preset and it will take all linked instruments and samples with it. There is no right-click paste function so just use Ctrl-V

In reply to by yonah_ag

In my own experiments with this font, here is what I came up with:

I deleted the preset and all but the P instrument. Then I re-associated presets with P. Saved as a new font.

Loaded it into MuseScore.

Set up score of several measures of F4 half notes at 45bpm. Mixer set to P font.

Set the dynamic of the first measure to p. Next measure at mp, and so forth up to ff.

I set up the synthesizer so I could watch the volume meter and the score at the same time.

I noted that the meter registered higher every measure. But not as much as I might have thought.

As a control, I did the same thing with the default HQ expr. instrument. To my surprise, the meter readings where basically identical to P. P is deceptive because notes tend to swell and recede. So it may not sound louder at ff, but it really is louder, over all. And it stays smooth.

In reply to by fsgregs

I didn't edit the velocities. I just cleared out stuff you don't need. What I ended up with is the font I sent you.
P works just as it is. It is very expressive. Perhaps deceptively so.

On the left hand side, right click the "all around" preset and delete. Right click and delete each of the instruments except P. Select P. In the menu bar select the two 8th notes (Add a Preset). In the popup window select OK. Export. That's what I sent you. Try the full test I out lined above.

In reply to by yonah_ag

HA! It worked. I opened the big font in Polyphone, created a new file, returned to the big violin font and copied the "violin all around" presets. Then pasted them into the new file. Then deleted the F and X harsher presets, but left the P and RelX presets. Then I changed the velocity of the P to 0 - 127, and saved the soundfont. Then, I loaded it into my song, set the mixer to call it up on the violin track, and added dynamics of both f and ff, AND THEY WORKED. The beautiful violin got louder but stayed soft and melodic. You are a genius. Thanks

BTW, this is such a pretty violin that you too may want to keep it in your soundfonts folder.

In reply to by fsgregs

As bobjp pointed out, you can clear the velocity range and then the whole range applies. I used 0-127 for clarification. There is no advantage or disadvantage either way since the whole MIDI velocity range is 0-127.

If you ever want to use the F or X instruments and their samples then you still have them available in your violin font due to the copy/paste. Maybe you would find the F instrument useful for fff dynamic. It's up to you.

In reply to by bottrop

The version that I made is not layered and fsgregs took the same route so his shouldn't be layered either. I checked out some of the samples with Audacity but felt that they could not benefit from any changes so I left them alone.

Perhaps you mean something else by layered? I don't remember seeing this term in Polyphone.

In reply to by yonah_ag

The All Around font does change to the louder, more aggressive instrument for fff. It also changes to it for an accent. But it is too loud.
I can't imagine a fff dynamic for solo violin. Trombone? Sure. Why?
Just for fun let's talk about Vivaldi for a moment. In Baroque times there were no diesel trains screeming by. Or 747's roaring over head. No Rock concerts. It was a quieter, slower time. Instruments were quieter. Even pip organs. They were powered by low pressure foot pumps. In modern times we insist on playing baroque music in a loud bombastic manner. Every Stradivarius played in modern orchestras has been rebuilt with a modern bass bar, longer neck and different neck angle. Partially so that they will be louder. As a guitar player, you know how changes like that alter The sound of an instrument.
Vivaldi may have performed Winter with a 8 piece orchestra or just a soloist and a keyboard and bassoon. Many concertos come down to us as a solo line over figured bass. Players just knew what to play.
For comparison, I just listened to several versions of Winter. Most played very aggressively. Well suited to the louder all around loud font. Only Perlman played more smoothly. Probably more authentic. Although we really don't know for sure. I like loud an bombastic as much as the next guy. It just depends.

In reply to by bobjp

Interesting, indeed. I haven't heard Vivaldi's Four Seasons for over 40 years, (having abandoned classical music), but I recall that the violins on "Winter" definitely had more bite than the other seasons. (I am finally rediscovering classical music, with some help from Bachmeister BSG).

Does the All Round smooth instrument still sound realistic when used at f ? (I am clearly in no position to make such a judgement).

In reply to by bobjp

and is that smoothness realistic for a violin at ff or would a real violin have more bite? Perhaps it is not simply dynamic related but the player has control over smoothness vs. bite at any dynamic? I have no idea but I'm sure that any violinist would be able to answer that.

In reply to by bobjp

Bob: When I played the all around violin unedited, it called up the F and X wav samples in the .sf2 presets whenever an f or ff dynamic was used. It used the "P" sample for all the quieter, more melodic music. Frankly, when the notes got harsher, they also changed in sustain as well, cutting off the sustain too early. Perhaps that is what happens when a violin player presses harder with his/her bow to get a louder sound, but there was no way to keep the violin sustained from one note to another, if the soundfont changed from the "P" samples to the"F" or X" samples. If there was a way, I might be willing to tolerate a stronger bite to the note.

I suspect we will have to use the all around for faster louder violin music, and the edited file with just the "P" samples when we want quieter more melodic or haunting violin. Marc is probably correct when he says there is no way a single violin soundfont can be used for ALL styles and speeds and volumes of play. There will always be a tradeoff.

In any event, I really really want to thank you and Yonah for all the work you've put into this. Hopefully, you will be able to use the All around violin for some of your own music. Its P samples are just beautiful and so expressive.

In reply to by fsgregs

It is absolutely possible to play loud sustained notes on a violin. Just like it is possible to play aggressive attacks at lower volumes. What bothers me about the all around font is that it makes some assumptions about what you are after. That's why I stripped out all the stuff I don't want. I'm still trying to figure out if I like the font to begin with. Some of it is a bit much. It doesn't work for all solo music.

In reply to by bobjp

I suspected that might be the case. Clever folks these violinists.

It's becoming apparent just how difficult it must be to make an 'all round' GM SF2 since every instrument must have its own set of nuances that you can't possibly allow for with a single preset. Presumably this is where VST comes to the rescue - but maybe at the cost of a lot of manual effort.

In reply to by bobjp

Bob: As you detailed, if you strip out the three other samples in the font and leave only the "P" samples, and change the velocity in the preset to 0-127, then yes, I can get the P notes to sustain even when loud. However, all the notes are the P samples and have no bite or harshness at any volume. I wish I was a nuanced musician and was able to tell how realistic it is to use only the P samples for all volumes, or if we really need some harsher tones when the note gets loud. I have yet to find another violin soundfont that plays the melodic rich tones the P samples give me at lower volumes, yet also changes to a rich but different tone with more bite when louder. As Marc said, that may be impossible to find. If you do find one, please list its download site here. Thanks

In reply to by fsgregs

If you have the patience then you could load your samples into Audacity and make a new mixed set of samples from P and F in any volume proportion, e.g. 50/50, 70/30 etc. Save your preferred mix as a PF sample. This would give a smooth/bite blend which you could then assign to midi velocity. You might need to reduce the overall amplitude of the mixed result - I'm not sure, so this is one to play by ear.

I have not used this feature so you'll need to read the manual at:
https://manual.audacityteam.org/man/mixing.html

However, as bobjp has indicated, violinists can play low volume with bite and high volume with smoothness!

In reply to by yonah_ag

I feel I can use Audacity, but it is the sustain (release???) that I am worried about. I can combine the mild, soft P wav sound with a more bite F sound in audacity to form a 70/30 type of combined wav, but when the velocity of a note exceeds 90, the sustain portion of the soundfont changes. If I want the sustain to stay the same as the P note but just add a bit more bite to the louder notes, what value in Polyphone do I change to increase the sustain time?

In reply to by fsgregs

Here is my guess.
This font has been around for a few years. In that time, who knows what has been done with it. Who knows how the various parts of it have been dismembered and recombined for what purpose.
There are parts of it in Essential Keys.
There is a version in Nice strings plus orchestra.
There is the version you found Polyphone. I'm not interested enough to register to down load it, but my guess is that it is yet another version.
I have spent the time I have on this because of the learning experience in Polyphone. But even that is of limited value to me. I am mostly interested in extracting instruments. And even at that, they are of limited value in MuseScore because of lack of SND compatibility. This font might be useful to me in certain solo situations. It is a bit overdone for my taste. And I can't use it for a violin duet because the vibratos line up and sound weird.

In reply to by fsgregs

Every instrumentalist has their own way of playing their instrument. How they start and stop or connect notes. is going to be different in any given situation. As well as vastly different from any other player in the same situation. No two players will play the same the same melody the same way. Even if player "A" plays a melody, then player "B" plays the same melody on the same instrument, the sound will be different. This is how it should be. Playing a string or wind instrument involves vibrato. The speed and intensity of each musician's vibrato is different. And so it is with the all around font. In my experiments with different fonts, I have a short bit of orchestra music that I wrote. To test the all around font I assigned it to the first violin staff. Then I assigned it to the second violin staff and played them together as a duet. The player who recorded this font plays a certain way. he has a consistent vibrato speed. this is normal. In my test I listened as the two lines played a recording of the same player playing two different lines the same way. As is the nature of recordings, the matching pulse of the vibrato started to cancel various frequencies out. To a lessor extent, this is a problem with all computer playback, but really bad in the case of my test.
When I changed the second violin part to the default solo violin in the MuseScore HQ font, (which isn't all that bad, BTW) things were much better. Two different sounding players. Which is what I want. Except that the all around font does not work with single note dynamics. The Musescore violin will crescendo on a whole note (for example) if I put a hairpin under it. The all around font will do what ever it normally does on that whole note whether I want it to or not.

MuseScore does not play well with many fonts. Such is life. I use MuseScore to compose for the purpose of play back. Right away, I am at a disadvantage. But I don't really care how the score looks. So I can mark it up any way I want to get the sound I want.

In reply to by bobjp

Wow! I understand. I have so much to learn about music. I have not yet tried to play any piece as a duet, so I have not yet tried to play the "P" violin as a duet. I did, however, play it as part of a 3-part score with it playing the main melody, a 2nd "strings" voice playing a 2nd track of the same notes, and the MuseScore violin playing a 2nd violin of different notes. Together, they all sounded quite nice and full. I will restrict my use of the P violin to solo melodic, plaintive scores.

FYI, I have recorded the score I created using this violin into an MP3. It is a powerful Hans Zimmer piece. here is a link to it. Listen for the solo violin, then listen as it is joined by 3 other strings. One of them is playing a duet with the violin track. https://1drv.ms/u/s!AlxY_utkDRE6nHrgaIF5hbl3aPJe?e=JVotPY

In reply to by fsgregs

mp3 vs wav.
Simply put wav is the better sounding format of the two. Wav is brighter, punchier, has a wider frequency response, and is a crisper, cleaner sound over all. All of which makes it easier to pick out the second trumpet part. As a result, a wav file is much bigger because there is more information in it.

Mp3 is more popular for downloading because the files are much smaller than the same wav version. The wav file can be 9 or 10 times the size of the same mp3. Most people can't tell the difference, nor does it matter that much to most.

The point is that in order for software to produce the best possible result, you have to start with the best possible sources.

In reply to by fsgregs

It’s actually a Microsoft/IBM thing, released in 1991; WAV is just the file extension (and internal marker) of the RIFF container format when used to hold audio.

RIFF can hold many things: audio (WAV), video (AVI), MIDI (RMI), and Google’s new WebP picture format also uses RIFF as container. You can imagine this container like a tar or zip file, which also contain data on the inside, and some meta information about what these data are.

WAV files are RIFF audio containers; WAV has support for codecs, so it’s possible to put MP3-encoded audio into a WAV file (I’ve done that back in Windows 9x times so that the built-in Sound Recorder could play them; WAV4MP3 is a freeware tool that can add the WAV “container” or remove it to make it a plain MP3 file).

However, normally, WAV files contain PCM data (linear pulse code modulation, mostly 11'025/22'050/44'100/48'000 Hz, mono or stereo, 8 or 16 bit per sample). This is what you want, because it’s lossless (and, in fact, not even compressed).

There’s another lossless audio format out there, FLAC, which does compress; FLAC files are still quite a lot larger than MP3 or Ogg Vorbis files, but massively smaller than uncompressed PCM WAV files. If you want to store or upload/download largish amounts of audio, I’d suggest converting to FLAC; converting back to PCM WAV will produce the exact same sample data points the original had (probably even the exact same bits and bytes).

MP3 and Vorbis (the audio codec most often used with the Ogg container format) are both lossy compression algorithms, which means they introduce artefacts into the audio stream in order to be able to compress better. They try hard (using research into how the ear and brain works) to make that inaudible, but they are early lossy formats, and so they’re not all too good at that at usual bitrates (I’m told 320kbps MP3 is okay, and 256kbps Vorbis probably).

However, when you have a soundfont, the samples are processed a lot by the synthesiser: preamplifying, something with envelopes, panning, hall, chorus, other effects, pitch bending, volume change, mixing with other samples when you have polyphonic music, etc. to then produce output (MuseScore can produce PCM WAV, MP3 and Ogg Vorbis). You can imagine that a small error (from lossy compression) in the initial sample is multiplied many times during that process.

Then, the final output from MuseScore is most often MP3 or Ogg Vorbis. Taking a lossily compressed sample, decompressing it, then recompressing it again is not safe. Unless you use the exact same parameters, codec, compression software, etc. (but sometimes even then), you have a compression that introduces loss at different places, so this is also some kind of error amplifying. Here, you have a lot of changes (volume, mixing, etc.) in between, so you’re basically guaranteed to have bad results when starting from a lossy sample.

Then, there’s the soundfont to consider. SF2 stores the samples as PCM internally (omitting the WAV container because its information is already present in the SF2), but SF3 uses Ogg containers with Vorbis compression, which is also lossy. Now if you take a sample from MP3 (lossy), decompress it to PCM then compress it with Vorbis to put it into an SF3 format soundfont, you’ve again got error amplification because you use multiple different lossy compression formats.

This is, by the way, also the reason why, when editing soundfonts, you will always want the SF2, not the SF3. If you need an SF3, convert the SF2 to it after editing, but keep the SF2 in case you, or someone else, wants to make further edits. (Decompressing an Ogg Vorbis sample, doing changes to it, then recompressing, will almost always also introduce audible artefacts; hah, what an alliteration!)

In reply to by mirabilos

The individual note samples from Philharmonia are surprisingly good. I wasn't going to bother with them when they weren't WAV files but I checked them out and made them into a soundfont. This actually sounds better than several other nylon guitar presets in other soundfonts but not as good as Tyros Nylon.

The samples have a clarity in the mid to high frequencies which is particularly good. I converted them directly to wav and then maintained this format, i.e. no re-compression.

So, whilst I would agree in theory that MP3 samples should yield a poor soundfont, in practise it doesn't have to be the case.

In reply to by bottrop

I still don't have the foggiest idea what you are on about.

I have incorporated the guitar samples into a soundfont and, in the process, checked each sample. A few were at an inconsistent volume level, one had a 'click' sound in it and another had some extraneous noise at the end. The rest were high quality and, since there were samples for nearly every pitch, it was easy to work around these issues.

I assumed that the violin samples are of equal quality but maybe they are not. What bells and piano have to do with any of this eludes me.

In reply to by bottrop

Thanks for the comment. Since Chevaliers is supposed to be a rich orchestral piece, I did not want to highlight the violin or strings over any other instrument. I intentionally wanted it all to blend together, particularly as I added a new instrument with every stanza. However, by using the "P" preset violin in Violin all around, I really loved the melodic smoothness of its sound within this score.

Maybe, just maybe, the final great solution would be MuseScore had the capability to directly use the samples one by one. So, we could decide which sample use to which instrument.

Of course, I'm talking just because I have a great mouth. I don't know if there is something like that somewhere, hihihi.

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