Pitch Accent Drop is off by one kana (for generated pitch graph that have no audio)

[emwspeed]

Type:

• bug
• feature
• enhancement
• question

Environment:

• OS: Windows 10
• Migaku Japanese Ver. 1.2.3
• Anki Version 2.1.35

I'm going to open a PR to resolve the issue:

• yes
• no (I don't really know much about coding)

Description:
I hope I'm not misunderstanding pitch accent notation, and sorry if this has already been mentioned somewhere but I could not find it in any other issues or on the add-on page on ankiweb. Also I don't know if this is an issue with the add-on or the database used for pitch accent but I thought I'd share my findings anyway.

I've noticed that for some words with small ゃ、ょ、ゅ the pitch accent with be generated correctly for the sentence but not for the graph that comes with the audio, when there is no audio generated. Some examples:

宇宙ステーション will have become 宇宙ステーション[うちゅうすてーしょん;n5] (I edited it to 宇宙[うちゅう;n5]ステーション because it looks nicer) but in the audio the pitch drop is after ス, as in it doesn't count the ちゅ as one but two.

距離感[きょりかん;n2] is correctly generated as nakadaka, but the audio graph is atamadaka with the drop after きょ

And some more without written explanation:

This is only a small sample, but I noticed that all of these originally came with ［ア］ as in they did not have an audio file (most in the screenshots no longer have that, as I've replaced it with audio files from somewhere else, like Forvo).

Further, here are some examples that did not come with ［ア］because they have audio generated with them, and they are correct!