Sfz player and Sonatina soundfont
I know when the subject of high end soundfonts comes up, the Sonatina soundfont often comes up, but it is only available in SFZ format. While the demo on Mattias' site is impressive, I have my doubts about how well it would actually work for basic playback - without the enormous amount of hand tweaking that presumably went into that demo. At one point I tried looking into tools to convert it to SF2 with some lmited success, then I gave up, as it seemed there would be a lot of work involved in really making it happen.
Now, it occurs to me that if there were a program that could play MIDI files using SFZ soundfonts, I could create a MIDI using MuseScore and hear for myself how it sounds with automated playback. Hearing that might change how I feel about thus whole idea of growing our own soundfont, one way or another. But while I know the VLC player can play any MIDI file using an SF2 soundfont, if there is something analogous for SFZ, I haven't found it. I found the free program from Cakewalk, but I think it perhaps is just meant to be used as VSTi plugin or something, because if there is a way to just select a MIDI file and play it, I have't figured it out. Also, it appears you can only load individual samoles, no way to load the whole library that I can figure out.
Any ideas? I have Sonatina downloaded and installed and would really like to hear it on some of my files.
Comments
From what I can gather from a brief digging spell on Google the Cakewalk SFZ player is designed to be linked to a sequencer by VSTi.
So you would need toload the MIDI file to your sequencer, install the player as a VSTi and then play back the file from the sequencer.
I don't know whether VLC supports VST but if so you could use VLC instead of the player, or maybe link VLC to the player by means of MIDIYoke or other MIDI routing software.
HTH
Michael
In reply to From what I can gather from a by ChurchOrganist
I've been using Sonatina for some time now (and my students too). Under windows, I use Cakewalk sfz player as vsti on Reaper (caution: there is 32 bits and 64 bits, to avoid some problems), but it only loads an instrument at a time. Under linux (there is also mac and windows version) there is Linuxsampler (http://www.linuxsampler.org/), wich is an independent host for giga, sf2 and sfz soundfiles, as well as lv2 and vsti pluguin. I have loaded up to 20 instruments simultaneously, combining formats. I think this is the best option.
In fact, the open source DAW Open Octave (http://www.openoctave.org/oomidi_2011) has linuxsampler and sonatina preloaded in it. Maybe Musescore can use a similar approach
I have now had time to look at the Sonatina website and listen to the mp3 demo.
From what I have seen and heard, I am not impressed.
The demo reveals that the samples have no release envelope shaping, and so consequently sound dry and unlifelike.
I am also concerned about the lack of velocity layering and additional articulations, as some of the comments we have been getting on the forums show that some composers are very keen to have these included.
So, as far as I'm concerned Sonatina is a dead end - OK it is an improvement on TimGM6MB and the GS synth supplied with Windows, but then so is the Fluid R6 soundfont, and at 440MB it is a big download.
Also the sfz format, whilst acknowledged as very powerful, does not seem to have the breadth of support in the form of editors and players that you find with Creative's format.
In reply to I have now had time to look by ChurchOrganist
Yeah, lack of velocity layering kind of shocked me too, although I guess I was more impressed by the demo than you - it does sound better from that one example than FluidR3. But again, that coukd the a lot of hand-tweaking in DAW. One thing I'm trying to get a handle on is just how big we are talking about needing to get before you have something that sounds better out of the box than FluidR3. Not that this changes anything - and the ability of MuseScore 2.0 to load multiple soundfonts is going to be a huge help here. But also, to what extent the soundfont really is the limiting factor, versus the fact that MuseScore doesn't take full advantage of playback possibilities like using legato controller for slurs, or various aspects of the "human playback feature" in Finale.
Again, I'm not really in a position to help directly in any way I can see right now. Still, I'd like to understand what and where the potential is here. Which is why I was hoping for a quick and dirty proof of concept - "this is what my score could sound like played with MuseScore as is but with a better soundfont", and then also figure out a way to model how it might sound with FluidR3 but taking better advantage of what's already there.
In reply to Yeah, lack of velocity by Marc Sabatella
Sadly I think the answer is that we have to wait for the dev team to implement more playback features.
A certain amount can be done with creating special soundfonts, and that will help a lot with playback once we have 2.0 with it's multiple soundfont support,
Unless someone takes up your idea of an itermediate application that will read a MuseScore or MusicXML file and render it into a good MIDI performance.
In reply to Sadly I think the answer is by ChurchOrganist
Since the main goal right now is in usabilty and writing, not sound, I'm not shure a more complex audio engine in order to get better sound would be desirable. Maybe users prefer a simple playback solution like what we have now, and not the need to spend time tweaking the audio engine (I do).
Maybe another approach is desirable: to make it easier to conect Musesc. to external audio-engines (linuxsampler, or kontakt, for instance), so interested users can take advantage of it, but Musesc. remains simple and fast as it is now.
In reply to Since the main goal right now by pdro74
Playback has never been an issue for me - I am old enough to have had to imagine how my scores would sound, consequently I mentally hear missing articulations when playing back from MuseScore.
It is evident, however, from the amount of traffic in the forums on this subject that there are many users who find the limited playback facilities a barrier to their use of the application.
Even building in the support you suggest is going to take the development team away from the important job of ensuring MuseScore's notation display is rock solid, and even if that support is implemented, many users will not have the knowhow on how to achieve their playback desires with external samplers.
I feel the way forward for the time being is for custom soundfonts which have various articulations built in, so users will be able to choose Violins-Marcato for example, or Brass - Sforzando attack.
One thing I DO feel that is badly missing from the audio engine, however, is the ability to use the Expression controller for dynamics - velocity is only valid as a dynamics controller on pianos, fretted instruments and tuned percussion. In the world of strings, brass and woodwind, velocity is the amount of attack a note gets, and dynamic expression is a continuously variable property during the duration of a note.
I shall, therefore, be pushing for this change in the next major version of MuseScore.
In reply to Playback has never been an by ChurchOrganist
My feeling is that I would like to see the built in playback facility remain "simlple" - I don't see the need for complex comtrols within MuseScore. A separate program that provides that makes sense to me. But still, I think there are something things that can and should be done within MuseScore to improve the quality of automatic playback. Mostly a matter of making sure hay all markings have playback effect. As mentioned, I think making better use of controllers is part of this. Volume to allow crescendo/descrescdo within notes, legato for slurs, pitch bend for scoops and falls, etc. This doesn't If something in instruments.xml allowed you to customize so e aspect of this (eg, the length of staccato), that's more than good enough for customization within MuseScore, I think, at least for the foreseeable future.
The question in my mind is prioritization. To what extent does improving soundfont make the buggest difference, and to what extent are these other enhancements more important. I'm not too keen on having articulations and so forth built in to the soindfont as separate samples unless MuseScore can automatically switch between them as appropriate.
In reply to Playback has never been an by ChurchOrganist
My feeling is that I would like to see the built in playback facility remain "simlple" - I don't see the need for complex comtrols within MuseScore. A separate program that provides that makes sense to me. But still, I think there are something things that can and should be done within MuseScore to improve the quality of automatic playback. Mostly a matter of making sure hay all markings have playback effect. As mentioned, I think making better use of controllers is part of this. Volume to allow crescendo/descrescdo within notes, legato for slurs, pitch bend for scoops and falls, etc. This doesn't If something in instruments.xml allowed you to customize so e aspect of this (eg, the length of staccato), that's more than good enough for customization within MuseScore, I think, at least for the foreseeable future.
The question in my mind is prioritization. To what extent does improving soundfont make the buggest difference, and to what extent are these other enhancements more important. I'm not too keen on having articulations and so forth built in to the soindfont as separate samples unless MuseScore can automatically switch between them as appropriate.
I don't know if this is still of any interest, but here's a link to a sf2 version of sonatina:
http://the-filmmusic-group.deviantart.com/journal/Soundfonts-of-Sonatin…
In reply to Link to sf2 by heuchi
Cool! It's kind of a drag that they did this as separate sf2 files for each instrument since that means you can't use it with MuseScore 1.2 (except one instrument at a time). I just made a quick attempt to load a few of the individual sf2 files into MuseScore 2.0, which does support multiple soundfonts. Unfortunately, it's not stable enough for me to get more than little tastes of how it might sound. Initial impressions: promising, but not obviously better than FluidR3. Whatever one hears in the demo to make one think it might be better, I still think that's the result of heavy hand tweaking and processing.
I did go on another hunt recently for other free soundfonts and found several that came highly recommended, but none that actually seemed an improvement over FluidR3 when played by MuseScore. Although one I did find intriguing was ChoriumRevA. which does a surprisingly good job with some of its sounds in a soundfont that is under 30MB in size. An alternative to the GeneralUserGS soundfont, I'd say, with different strengths and weaknesses.