The Yamaha PLG100-SG
By Thor Zollinger, writer/musician/engineer at large. 2016
The SG manuals translated into English are linked below.
Download 10.9 MB
I recently acquired one of these rather rare Yamaha PlugIn cards. In order to use it, I translated the manuals into English. The most useful was translating all of the editor menus for XGWorks into English. The synthesizer isn't really very complicated to operate, it's nowhere near as complex as the Yamaha FS1R. The SG card is similar in nature, using 8 Formant filters like the FS1R, but it lacks the entire FM side and all of the envelope controls of the FS1R. The FS1R can't script sentences though, which is what I bought the SG card for.
The SG card has 8 Formants, with two oscillators per Formant filter. One oscillator has harmonic content, the other is a noise source for all the breath noises a person makes when they speak. The filter's center frequency, and the amplitudes of the two oscillators are adjustable. The card also has 10 additional adjustable parameters to control the 'color' of the voice, and 7 insertion filters for special effects (besides the ones the host Synth provides). i.e. It's not very difficult to set up a voice.
The other aspect of the card is the ability to script sentences for it to speak or sing. The card uses Japanese phonemes and syllables which can be entered using either an entry list box, or read in from a text file. Putting together a script isn't as hard as it sounds, all of the data tables in the back of the manual have a column for English pronunciation. The tables are also arranged phonetically, all of the 's' sounds, etc are grouped together. Using the tables, it isn't hard to string together phrases. Find the phoneme you want, copy the Japanese characters from the table, and paste it into your script. Inhalation breaths are entered as a '/'. Actually, I've been using an online Japanese Hiragana typewriter app to string my text together.
The Lyric editor then matches up the phonemes in your script to notes in your midi file. Yes, the song notes have to be written first. There is a section in the manual on fine-tuning pronunciation, but I won't be able to comment on that part until I've had some more time to experiment with the SG card.
I re-edited the manuals extensively to correct the previous versions, I had to fix all the errors Google Translate came up with on round one. I also added in notes from my testing with the card, I added in two SG demo MIDI files I located, and I added in a couple more dialogs that needed to be translated into English to the window graphics bitmaps as well.
I've successfully programmed in seven new voice patches based on the voices of Mel Torme, Harry Connick Jr, Nat King Cole, and Christina Aguilera, all big improvements over the factory patches, but I haven't been able to figure out where the new Registered patches are saved yet. I can't include them in the package until I can figure out where they end up.
Here is the first test of the card singing 'Summertime' by George Gershwin. I haven't tried doing the lyrics yet, this is just the Christina1 voice patch singing an 'oooh' tone to the tune of the song.