youngcooper626/descript-edit-ai-transcriptor

Descript Pro AI Editor is a video and podcast editing tool described in this README as allowing users to edit recordings by modifying a text transcript rather than manipulating a traditional video timeline. The central idea is that every word in the automatically generated transcript corresponds to a precise moment in the video, so deleting or rearranging words in the transcript automatically cuts or rearranges the corresponding footage. The README describes four main capabilities. The first is transcript-based editing, where the tool generates a word-by-word transcript of uploaded video or audio and lets users make cuts simply by deleting text. The second is a voice cloning feature called Overdub, which the README says can replicate a user's voice after 10 to 15 minutes of training audio, allowing new sentences to be generated in that voice for corrections or additions without re-recording. The third is Studio Sound, a one-click audio cleanup that removes background noise, levels volume, and reduces room echo. The fourth is automatic filler word removal, which detects and deletes words like "um" and "uh" throughout a recording while attempting to preserve intentional pauses. An integrated screen recorder is described that captures video and narration simultaneously and generates a transcript of the narration in real time, removing the need to manually sync voiceover with screen footage after the fact. The tool also includes automatic speaker detection for multi-person recordings. A command-line interface is described for importing files, applying edits, triggering Overdub, and exporting output without using the graphical interface. JSON configuration profiles control settings for filler removal aggressiveness, audio enhancement style, and screen recording resolution. The README lists compatibility with Windows, macOS, Linux, iOS, and Android, with full support on Windows and macOS and partial support elsewhere. The README is structured as a product guide and feature showcase with a download link pointing to a GitHub Pages site. No source code, build instructions, or contribution guidelines appear in the provided text.