My "end of week 1" impressions of Vocaloids: much to learn and Yamaha has already started throwing some illogic into the mix. Here's a perfect example: in constructing phonetic pronuciations you use "phonemes" (pronounced "fowneems") and a list is provided. Some of the conversion from what you type to phonemes is automatic but editing is usually necessary. There are abbreviations for phoneme symbols, like 't' for the kind of t used in the word "sit". Makes sense huh? Well, Yamaha says the phoneme symbol for the t in "think" is 'T', while the symbol for the t in "top" is 'th' - that's a direct quote from the manual. And the manual further suggests that to make editing easier you commit it's phonemes and symbols to memory, even if they are ass-backwards I guess. Yamaha is still in need of English proofreaders.

I had trouble getting Vocaloids to say "health" - it comes out sounding like "halth" like the 'ha' sound in "has". It appears that I can edit the dictionary and edit the phonemes but there are too many other parameters I want to understand first. Also some vocal samples end funny - like "ooo" has more of an "oooee" sound to it. I have a ways to go before I know if this is a problem with the program, the vocal fonts or me.

The program is deceptively easy to get started with and to enter lyrics and start throwing expression parameters around for instant gratification, but the real effort is "humanizing" - as in combining the many possible parameters to sound lifelike. I still don't know what the actual impact of some of the parameters are even after trying them out. One parameter that does add interesting texture is "noise" which makes the voice sound more airy and human so that'll get used a good bit. My focus now is just trying all of the parameters one at a time until I know what I'm doing.

Becoming an expert with Vocaloids will not be easy. That's good in it's own way - mastering Vocaloids could be a unique skill that won't be commonplace. My attempts so far are very synthetic-sounding and not ready for hearing (I'm afraid it'll turn people off too soon). Maybe next week.