Transcript Polisher

•
872
•

Transforms raw voice-to-text transcripts into polished, readable articles while preserving the speaker's original tone and vocabulary.

Transcript Polisher is a specialized capability for Claude Code designed to refine messy transcripts from interviews, podcasts, speeches, and meetings into high-quality written content. Unlike standard summarization tools, it acts as a professional editor that meticulously removes filler words and stuttering while strictly maintaining the speaker's unique voice, metaphors, and specific phrasing. It supports SRT and VTT subtitle formats, automatically corrects common homophone errors, identifies multiple speakers in dialogue, and applies 'semantic breath' paragraphing to ensure the final output flows naturally without losing the depth of the original source.

주요 기능

01872 GitHub stars
02Speaker identification and dialogue structuring for multi-person interviews and podcasts.
03Context-aware homophone correction and grammatical refinement for high-frequency verbal errors.
04Automatic SRT and VTT preprocessing to strip timestamps and formatting for a clean text flow.
05Semantic paragraphing based on logical shifts and perspective changes rather than mechanical sentence counts.
06Intelligent noise reduction that removes fillers like 'uh' and 'um' while preserving semantic connectors.

사용 사례

01Converting raw podcast transcripts into readable blog posts or articles without losing the speaker's personality.
02Refining video interview subtitles into professional written records for archives or publications.
03Polishing messy meeting notes and speech-to-text records into structured, easy-to-read documentation.