savbell / whisper-writer

💬📝 A small dictation app using OpenAI's Whisper speech recognition model.
GNU General Public License v3.0
327 stars 52 forks source link

Doesn't seem to support German special characters (Umlaute) like ä, ü, ö, ß #26

Closed RaffCode-Personal closed 4 months ago

RaffCode-Personal commented 8 months ago

I attached a voice recording that should transcribe to "Es wäre schön, wenn das Programm auch Umlaute unterstützen würde."

https://www.dropbox.com/scl/fi/5jez2tuqyjzxxpemtmz12/Umlaute.mp3?rlkey=mure8ckqmc801dgmfsox6f9bq&dl=0

Instead the output is this: "Es wre schn, wenn das Programm auch Umlaute untersttzen wrde."

thfrei commented 8 months ago

I forked it and did some rework. Umlaute work for me. Maybe it is because I changed something with keyboard input? https://github.com/thfrei/whisper-writer

thfrei commented 8 months ago

Transcription: an jemandem Maß nehmen.

an jemandem MaS nehmen.

Transcription: Kopf bis Fuß.

Kopf bis FuS.

Ok, "scharf-S" funktioniert noch nicht.

Edit: ß funktioniert bei mir nicht, weil ich Schweizer Tastaturlayout verwende und gar keine Taste für ß habe. Mit Deutschem Tastaturlayout geht's vermutlich.

savbell commented 8 months ago

Hi, thanks for the bug report! :)

When I tried running the mp3 with the local model, I got "Es wäre schön, wenn das Programm auch Umlaute unterstützen würde." With the API, I got "Es wäre schön, wenn das Programm auch Umlaut zu unterstützen würde."

@RaffCode-Personal, are you running from the latest commit? Special character transcription was an issue with an older version and was fixed when we switched to using pynput instead of pyautogui: https://github.com/savbell/whisper-writer/issues/9

CsehAbel commented 6 months ago

I can confirm that it works. It works perfectly with accented characters, starting from the commit that replaced pyinput with pyautogui for typing the characters.

savbell commented 4 months ago

Closing the issue as it seems to be resolved. Thanks everyone!