Over the past few days I've done a lot of work fixing up warts relating to generating non-English voice files.
Things have progressed to the point where all the build infrastructure
can now automatically generate (and present for download) multiple voice
files, with each voice having its own language, tts engine, and tts
options specified.
(Why multiple voices per language? Consider different voice genders or
regional pronunciations, or in a more extreme example, China has
multiple distinct spoken languages that share a common script!)
So, I just have to flip a switch and all languages with >= 95% string
coverage [1] will get voices automagically generated every evening,
using the espeak-ng engine's default voice for that language.
There is probably no downside to turning this stuff on, but before I do
so I would appreciate it if folks who speak/understand these languages
can do a quick sanity-check to make sure the voices are usable and
useful.
Additionally, this change also includes a switch the nightly build's
English voice from Festival to espeak-ng, mostly due to the latter being
much faster at runtime. I can revert this back to festival if anyone
feels strongly enough about it.
Feedback appreciated!
Meanwhile, back to the bit mines...
[1] English, Serbian, Polish, American English, Russian, Greek, Slovak,
Norgeigan, Italian, and German
- Solomon
--
Solomon Peachy pizza at shaftnet dot org (email&xmpp)
@pizza:shaftnet dot org (matrix)
High Springs, FL speachy (freenode)
signature.asc
Description: PGP signature
