Earlier this year, the musician and producer S. Christian Collins updated his soundFont, GeneralUser GS. He also blogged about the current state of SoundFont technology and his considerations for the format’s future. He wondered whether this still matters for many people. My personal answer is a resounding Yes!
A little background: I once studied music (in vocational school) and I now have a renewed interest in musical practice. When I began reading about music production last year, I clarified my inner picture of the music tech landscape by first learning that MIDI is a) still totally relevant today and that b) the MIDI protocol is not at all tied to a certain style of (presumably cheesy) sounds. This is an old conflation that many people from my generation have made because, back in the day, when MIDI soundtracks were common – not only in games but in many kinds of multimedia software – the most commonly heard sounds were those of the low-quality licensed GS synths present in Windows and Mac OS. When I realized how little those sounds have in common with modern sample-based synthesis, it opened my eyes to what I would have loved to know much earlier: I love sample-based synthesis and want to work with it in as many contexts as possible.
I had always known, more or less consciously, that there existed adequate technical means to the end of producing realistic orchestrations without ever having to record a live musician, but I had never realized what sampling and sound programming were truly about. For example, I am a Nintendo fanboy who loves listening to the original soundtracks of SNES, Game Boy Advance and Nintendo 64 games, but learning where the sounds of those games were sourced from made me recover a sense of connection with music that I had forgotten about after many years of working in other fields. I love the sounds of General MIDI modules and keyboard workstations, not inasmuch as they imitate acoustic instruments, but as tone generators that happen to be based on multisamples! Judged on their own merits, they have as much legitimacy as any other tone generation techique.
Collins compares current applications for SoundFonts and finds that very few implement the entire specification properly. When I read this and saw that Fluidsynth was definitely the way to go for me, I realized how deep the analogy runs between SoundFonts for synthesis and Adobe Type 1 fonts for advanced typography. In both cases, an ageing technology gets abandoned by its original specifier but remains relevant for an important subset of users and creators. In both cases, one of the best maintained implementations is a piece of free software (respectively Fluidsynth for SoundFonts and the pdfTeX system for Type 1 fonts). Again in both cases, the technology still has immense pedagogical potential, but also much potential in professional use. This is compounded by what I gradually learned about the different synthesis types by reading such publications as Sound on Sound and Keyboard Magazine: their series of in-depth articles demonstrate that a skilled synthesist can conjure a multitude of beautiful, workable sounds with a sample-based synthesizer, if enough modulation options are available. That is the case when a well-programmed SoundFont is used with a synthesizer implementing SoundFont ‘modulators’.
By returning to the field of music, I find myself bringing over my preference for free software, especially in pedagogical settings. (I find commercial audio software much more legitimate than commercial word processing software, but that is a discussion for another day.) Because of that preference and the high availability of free, modern music production software, I profess that I will encourage any newcomers and potential students to the crafts of instrumentation and synthesis to try out free software that implements good sample-based synthesis, along with some aspects of subtractive, additive and other types of synthesis.
Returning to the analogy between SoundFonts and text fonts, I am not alone in capitalising on it. One need only read excerpts from this enticing review of an upcoming typography book:
Several good metaphors are used throughout the book, illuminating its main themes. [The author] compares the fonts to classic musical instruments like cellos and pianos. This metaphor becomes especially apt when he notes that it is difficult to judge some variants of Palatino because they were not actually used—as we cannot judge an instrument which was never played by a skilled musician.
Second, if a font is a musical instrument, then a discussion of it is not complete without a sketch of musicians who played it and the pieces where it shone brightest.
Do give GeneralUser a chance, as well as Fluidsynth’s included SoundFont, FluidR3. Find some good quality MIDI files or, if you can, play an old game with a MIDI soundtrack and use a SoundFont player to render it! You’ll see what I mean. Youtube is full of excellent examples.
With that, I shall return to my experimentations with pulse-width modulated string sounds.