Help

Spelling errors are caused by sounds being incorrectly transcribed. To get the fewest spelling errors during transcription, use one of the large models. In files with more than one speaker, turning on the speaker recognition feature also improves transcription quality. You can correct the spelling and writing errors that occur in a transcription with the help of AI. Claude performs best for this kind of editing. When giving the .srt file to the AI, pay attention to the points below and provide the stated instructions.
Points that improve file quality:
• If you are working on a short file, prepare a reference text, or add to the .txt file any parts that are in your audio / video file but missing from the .txt file created by MurText.
• Give the AI the .txt, .srt, and project_directory/fallback/project_name.json files located in the project directory. The .json file contains the start and end times for each word.

Instructions:
• Never touch the timestamps.
• Treat the timestamps as numeric (integer) values. Never convert them to strings.
• First work on the reference text and correct misspelled words, names of foreign origin, and technical terms. Apply a spelling and punctuation check to the file.
• Synchronize the .srt file with the corrected reference text in terms of content.
• If speaker recognition data is available, start a new line at each speaker change, or create a new segment using the data in the .json file. Remember that Hugging Face may make mistakes where speaker changes overlap. Check that each speaker change is consistent with the meaning in the flow of the sentence.
• If the number of words in a segment is inconsistent with the segment's start-end range, warn me. Remember that WhisperX may have skipped something because of background hum or loud music.
• At the end of the process, check whether any segments overlap in time. Report to me the areas that are in the reference text but not in the .srt file, the ranges that may have been skipped, and any inconsistencies you notice.
• Return the .srt file to me under the same name.

Final Steps
Follow the steps below to save the edits you have made. If you do these steps, the .txt file belonging to the transcription will also be edited automatically.
• Open the .srt file you received from the AI. To do this, right-click the file and select 'Edit in Notepad'.
• Select the contents of the .srt file with CTRL+A and cut them with CTRL+X.
• In MurText, trigger the SHIFT+E shortcut or click 'File Operations' > 'Edit' > 'Edit SRT'.
• In the window that opens, press the 'Edit Text' button. After deleting the text in the editing area, paste the new content using the CTRL+V shortcut.
• Press the 'Save Text' button.
• Set your other options and save.

After this process, the content you received from the AI will also be reflected in the other files in your project.