Show HN: Overtone – Transcribe, correct and cut captions for any screen shape
19 points by rvanderlinden
Overtone is a tool developed by Lont that automates the transcription and correction of video captions by utilizing multiple audio recognition models, including gpt-transcribe, Azure STT, and Elevenlabs Scribe. It addresses common issues such as misheard words and layout problems by re-evaluating conflicting transcriptions and using a language model to determine the most accurate output. This process reduces manual editing time and improves caption quality.