Ever since I switched TTS to Larynx, while I did get a massive voice quality improvement, I've had one massive bother: the way it pronounces "шесть", Russian for "six". It's one of the ten digits, so it's spoken very often when reading sensor data.
After poking around a bunch, I randomly discovered the issue in gruut-lang-ru/gruut_lang_ru/lexicon.db:
sqlite> select * from word_phonemes where word = "шесть";
530064|шесть|0|ʂ oː s tʲ|
Here's another word with a very similar pronunciation, at least as far as vowels are concerned:
sqlite> select * from word_phonemes where word = "жесть";
123349|жесть|0|ʐ eː s tʲ|
The pronunciation of шесть should probably be ʂ eː s tʲ. Wiktionary (mostly) agrees.
I applied this change to lexicon.db in the Russian profile inside my installation:
update word_phonemes set phonemes = "ʂ eː s tʲ" where word = "шесть";
And it worked!
How can I contribute this change for everyone else to enjoy? Do I just send in a PR with the modified database file?