I am transcribing German audio to text using VOSK using a German model. This works quite well, except for a few issues:
- it outputs text in all lower case
- numbers are transcribed as text
- the result has no punctuation
I'm now looking for a software that can help me with the first point of that list: fixing the casing of the text.
Requirements:
- TXT file as input (German, lowercase)
- TXT file as output (German, upper case corrected)
- command line utility (it will be used in an automated script)
- no user interaction, like no interactive spelling correction
- works on Debian 11 or Windows 10
- is gratis, open source preferred
- offline. I don't want anything to be sent to the Internet
The results needn't be 100% correct. 80% correction sounds good enough for me.