FS1R was...
...Perhaps too hip for the room... I had the task of demonstrating the FS1R at the AES show back in 1998 and from the initial reaction to it, I could tell it was going to be very misunderstood. It was a combination of FM + FS (Formant Shaping) Synthesis. But rather than accepting what it *could* do, the initial gut response from the customer base was: "I want to program my own".

Here's what I mean... The presentation would start by telling people it was FM+FS synthesis and that there are no samples in this technology - that what they were about to hear represented the state-of-the-art in the evolution of Frequency Modulation synthesis. The FS1R would play "Ladies and Gentlemen, may I have your attention please" in reponse to notes played on the keyboard (via what was called an FSEQ). Now there are no samples in this technology, but people would think they heard it say something or sing something, so therefore they would conclude that it must be a sampler. So I found myself repeating that there are no digital samples in this technology - everything they heard was synthesized. But then they would think they heard it say something else - and they would forget that there are no samples... it was pure synthesis. If you thought it was talking or singing, it was not. The FSEQ was the amplitude envelope outline of something that was spoken or played at one time but that was now used to create the formant content. There were some 90 preset FSEQs in the FS1R... the pitch and playback speed of the FSEQ could be varied over an extremely wide range. You changed the speed via clock and added pitch (from a pair of operators, one voiced and the other unvoiced). There were 8 pairs of operators available so the capability was unbelievable. If regular MIDI tracks are confusing, imagine an 8 track FSEQ, responsible for frequency and level output for 8 stacks of 2 operators!!!

Making your own FSEQs was (and is) beyond the scope of mere mortal musicians and that was the BIG issue.

The FS1R sounded phenomenal. But people could not get out of their heads that they wanted to program it to say what they wanted it to say... which, needless to say, was not going to happen. Not without tons of knowledge about FM synthesis and the mechanics of human speech (and a lot of computational computer muscle). It wasn't enough that the FS1R was a killer FM module with filters and fidelity miles beyond any of the previous FM synth engines.

Those who were lucky enough (and smart enough) to grab one, probably still use it. It was awesome, but very misunderstood. Technology can be great but unless you can bundle it into a compelling product, the result will not necessarily be a success. At the time the software to do the analysis of 128 amplitude stages of the voice to create the FSEQs was not only cost prohibitive but needed a powerful computer to run. And even then it would be far beyond what the typical musician could program. If FM was deep as an ocean, the FM+FS was as deep as outer space!!!