OK so having spat the dummy at this being a $4000 synthesiser and pointed out how you could make a substitute for a lot less, the real thing arrived in my lap for about the ‘lot less’. Few of my toys are wildly sought after, so I felt responsible for a clear description of what you get for too much money. I then sold it again because it really doesn’t have bang for buck
Executive overview.
The Formant synthesis (“make vowel sounds”) aspect of the FS1r is the least interesting part of it, unless you are interested in Vocaloid style lyrics, in which case just get a Vocaloid. The built in demo phrases are 99% shit (including Chinese “ching chong” sounds!), and you may only store 6 of your own by giving up half your patch memory. The Radias has a vocoder section that works the same way but with much less pain involved. So you are really looking at buying an FM synthesiser, based on how much you think things have changed since the previous batch.
There are many DX7 patches pre-installed in the FS1r ROM and that along with the square circuit board and the display, point to this being one of the MU series which has been opened up to editing late in the development. This was a cheap machine like the MU’s and it sounds it. Not as bad as the TX81z, but a reminder. The terrible manual shows signs of this late change of design – it’s the least helpful manual I’ve seen from Yamaha, mostly internal notes with little elaboration.
There are 8 ‘voiced’ operators in each algorithm. You might think you’re going to do a lot more with 8 operators, but as the FS1r is 4 FM synthesisers in a stack it’s much easier to just layer two or more FM voices than to do anything with the extra ops. When you import a DX7 patch it is placed backwards on operators 8 and down, leaving 1 and 2 spare.
The operators have extra waveforms, like the TX181z. There is the sine, two kinds of ‘saw’ waves, wide and thin pulse width, two kinds of ‘square’ waves, wide and thin pulse width, resonant waves which are very quickly shrill, and the formant wave. You can adjust the ‘skirt’ of the each which rounds it back to a sine. When I say saw and square I mean the approximations that Yamaha makes out of harmonics.
The formant waves are simply band limited waveforms, with adjustable filtration by frequency, width and skirt. They’re not that uncommon (you can also find them on the Radias) but in the context of FM synthesis they become complex. They’re designed for Formant Sequencing to make voice sounds (which is the real reason for 8 of them), but when modulated some nasty results come about, which is finally the interest of the FS1r.
Not as interesting are the 8 ‘unvoiced’ operators, which are just white noise sources through formant filters, and really crap digital white noise at that. They are specifically for consonants in the voice synthesis. You will not be excited by the various whooshes you can make.
You now have pitch envelopes for each operator, and so you can go all Chowning on it.
There’s a subtractive filter that can be whacked over the whole lot, but beware it halves the voice count. It sound very similar to the DX200 filter. Like that box the FS1r has knobs which can tweak the FM system in real time. Even though one is labelled FM and the other is Formant, you can patch them how you like. All the DX7 patches have been set up so that one controls the timbre and the other the midpoint of the voice – vaguely the vocal range.
Let’s sum it up for now – you can get better, richer, FM sounds out of the TG77. It generally sounds like the DX7 with some new tricks. It compares to the DX200, but with the extra layered voices. Formant synthesis has moved on since 1998 and there’s not much here to inspire. It’s good but the high price is about collecting, and not music making.