[25] L'lasons de Aether ver Beinags #7

So, this week I've finally gotten my hands on an Exprish TrueType font. I spent at least a day researching on making my own, until my friend just gave me his own version.

Under closer inspection, the font apparently has a few quirks to it.

Non-capitalized letters are not mapped in any way, meaning that you have to use capitals to write in basic Exprish.

Also, you might've noticed a lot of glyphs in the above picture that aren't even present in Exprish's phonetics. These glyphs represent things like @,~,", et cetera.

Some of what even those extended symbols don't include

however, are numbers. This is likely due to the fact that Exprish's numeral system isn't that well developed to begin with.

It's still in conceptual limbo, but it will likely be tied to ancient Greek numerals in some way.

Pictured left is the extent of the numeral system's documentation I actually possess currently. (Yes, this is all I have.)

I might just make my own number system anyway, to remove the burden of making a text formatter in a numerical system I don't know all of the rules to.

While it is true that there are no actual numeric symbols mapped to the number keys, it's also true that a few of them are mapped to other stuff.

Pictured left are the only glyphs that the numbers are mapped to, which are 0, 1, and 2 respectively.

The latter two of these I know are affixes.

The "-AE" suffix would mean a present-tense verb. "KA-" would signify an adjective manifestation of a noun, if used as a prefix. If both are used in the same word, that word becomes an adverb.

There are two affixes unaccounted for here. The "-AH" suffix is used for past-tense verbs, whereas "-ON" simply makes for manifesting nouns.

These will likely be added in a new version.

What I mean by manifesting is that, by default, words in Exprish have abstract and symbolic meaning. This is mostly to make spells more concise, given its history of being used in scenarios that would use some form of magic. Specifically, Strykae-like magic, but I might hold off on that tangent for another time.

Next, an update on the whole speech synthesis thing.

To make this process easier, Instead of source-filter synthesis, I could use concatenative synthesis. The basis for this is relatively simple, where you concatenate different phonemes together. You'd also need to interpolate them to reduce clicks and whatnot when splicing.

I remembered this term when I was reminded of how the "Uncle" character from Jalopy talks.

This audio to the left shows all of the 4 voice lines the game uses for this character, all strung together. All 4 of these clips are game file rips.

Usually, the the audio clips used for concatenation are pairs (or more) of syllables, rather than just singular ones. I think I'll try the latter, as it not only makes for fewer clips to make clauses and interactions for, while also making the speech sound less human-sounding by a tiny bit.

To simplify splicing, I thought about making both ends of each syllable in the soundbank neutral. If I do this properly, the resulting audio shouldn't crack and click, as the splice will always occur at a zero crossing. Below is what a zero crossing looks like.

Finally, I started looking into how Unity handles UI elements. I didn't get very far this week, but I at least found a cool option in the inspector for TextMeshPro text.

This gradient gives the text a bit more character; I like it.

That's it for this week.