Usage

Through the shell:

Running Subtitle Corrector requires three arguments in the following order:

• [original subs file]
• --ofile [name of outputfile]
• --prompt [path to a prompt file] An example command on GNU + Linux would look like:

./SRT-tui.py examplesubs.srt --ofile modified_example_subs.srt --prompt Prompts/mafioso.json

Through the GUI:

You can use the program through the GUI by following these steps:

1. Start the application.
2. Select a prompt from the dropdown menu.
3. Select a subtitle file to process
4. Select the output folder (The modified subtitle file will be created in this directory with the name: corrected_[name of original file].srt
5. Click the "process subtitle file button".
6. Wait till processing is complete. Progress can be seen by watching the terminal.

Terminal output guide:

For each query sent, a new line will be printed displaying the following three values:

1. The amount of tokens sent.
2. The query number. For example, a query with number 11 would be the 11th query sent.
3. The total amount of queries to be sent in order to fully process the file.

Occasionally GPT4 will attempt to merge a sub block. This would cause desynchronisation hence making it so that subtitles will display later or earlier than expected. In order to correct this, the program will also count the amount of sub blocks in a response. If the number of sub blocks sent does not equal the number of sub blocks received, then the query is resent. If this occurs, the program will inform the user with a message consisting of the following elements:

1. Amount of tokens to be sent
2. Query number.
3. Total queries.

In the case that a response has the correct amount of sub blocks and nothing has gone wrong otherwise, the program will display a message informing the user that a response has been received. It will contain the following elements:

1. Query number
2. The amount of seconds it took for the response to be received.