Closed sipherh closed 2 months ago
This is supposed to be fixed in https://github.com/Cerlancism/chatgpt-subtitle-translator/commit/ce63d7af4f89a04deb1a91b441cf3ce8d9b23a3f
Are you on CLI? Are you on the latest version? Do you use --no-prefix-number
?
Can you provide some sample? But I know it could be hard to reproduce.
Yes, I am on the CLI and using --no-prefix-number
with the latest v2.9.0 version.
I was able to remove \N when explicitly telling GPT not to.
Here are the parameters used in my environment. One other thing, the source subtitle file was transcribed through Whisper in Mandarin, without manual auditing; maybe the quality will affect the result.... --model %gpt-version% --batch-sizes [20] --no-prefix-number --temperature %tempvalue% --%system-instruction% --input %source.srt% --output %source_en.srt%
Thank you for your project, it's great. I have a question: sometimes the original subtitles contain line breaks within the same timeline, and after translation, the subtitles show \N. Is there a parameter to request GPT not to include line breaks, or should I specify this in the prompt? Thanks