Clean up SRT captions and fix homophones
Speech-to-text constantly swaps homophones — the right sound, the wrong word. Clean your captions so the text matches what you actually said.
Homophones are the signature failure of automatic captioning: the recognizer hears the sound correctly but picks the wrong spelling, so the line is grammatically fine and completely wrong. These slip past a quick glance because nothing looks broken.
This tool reads captions in context and corrects the homophones that machine transcription gets wrong, alongside fixing punctuation and line breaks. Context is what makes the difference — the right correction depends on the sentence around the word, not the word alone.
It is built for the cleanup stage of subtitling, where retyping by hand is the slow, error-prone alternative. Run it, then verify proper nouns and technical terms yourself — those are where any tool is least certain.
The tool for this
📝Subtitle Cleanup
Clean auto-generated captions — homophones, punctuation, line breaks.
Frequently asked questions
Why does speech-to-text get the wrong word? +
Homophones — it hears the sound right but picks the wrong spelling. The tool uses sentence context to correct these along with punctuation and line breaks.
Why can’t I just glance over the captions? +
Homophone errors are grammatically fine, so they slip past a quick read. Context-aware cleanup catches what the eye misses.
What should I still check by hand? +
Proper nouns, brand names, and technical terms — those are where any automatic tool is least certain.
Related tools
🎚️Voiceover Script Polish
Tighten a spoken script — remove filler, fix flow, keep your meaning and voice.
🎬Short Video Script
A complete short-video script: hook, shot-by-shot voiceover + visuals, and CTA.
🎯Video Performance Review
Diagnose a video's performance across the funnel and get next actions.
Browse the full tools directory, or see all Panshi services.