Subtitle Refiner
Optimizes SRT subtitle files by removing filler words and correcting speech recognition errors while strictly preserving original timestamps.
Subtitle Refiner is an AI-driven utility designed to polish raw SRT files generated by automatic speech recognition (ASR) systems. It intelligently identifies and removes verbal fillers like 'um' and 'ah,' corrects technical misspellings (e.g., 'XGBT' to 'ChatGPT'), and ensures structural integrity by keeping all timestamps and indices identical. Utilizing the SiliconFlow GLM-4.7 model, it provides context-aware refinement and automatically delivers the optimized results and token reports via Feishu, making it an essential tool for high-quality video post-production.
Key Features
01Strict preservation of all SRT timestamps and sequence indexing
02Detailed token usage tracking and cost reporting for every job
03Intelligent correction of technical ASR errors based on video context
0417 GitHub stars
05Automatic removal of oral filler words like 'um', 'ah', and 'basically'
06Automated delivery of refined files to Feishu (Lark) workspace
Use Cases
01Standardizing brand names and technical terminology across AI-transcribed video content
02Refining interview transcripts by removing repetitive verbal tics and filler phrases
03Cleaning up auto-generated subtitles for technical coding tutorials and courses