the point of this project is to easily translate srt files when i need to because my gf doesn't understand english.
subtitles are translated in batches of 50 by default, use the -b flag to change it.
the default model is gpt-3.5-turbo but you can always use -m gpt-4.
make sure to export the OPENAI_API_KEY variable.
better result could certainly be achieved by using a sliding context and discarding the begining each time or more advanced techniques;
however, another objective is to have it being simple enough to fit in a single file of code and this will be good enough in 99% of cases anyway.
this could probably be used along some text to speech software such as whisper for some pretty cool results.
export OPENAI_API_KEY=(your key here...)
./main.sh -m gpt-3.5-turbo -l English -b 35 ./path/to/your.srtOr get help:
./main.sh -husage: main.py [-h] [-l LANGUAGE] [-b BATCH_SIZE] [-m MODEL] [-v] [-w WORKERS] files [files ...]
Translate srt files
positional arguments:
files File pattern to match
options:
-h, --help show this help message and exit
-l LANGUAGE, --language LANGUAGE
Specify the language
-b BATCH_SIZE, --batch_size BATCH_SIZE
Specify the batch size
-m MODEL, --model MODEL
openai's model to use
-v, --verbose display errors
-w WORKERS, --workers WORKERS
number of workers for processing long file concurrently