How does the pattern learning mechanism work?

The skill tracks AI-driven corrections; when a specific error-to-fix pattern occurs multiple times (typically 3+) with high confidence, it prompts you to add it to your permanent dictionary for future instant corrections.

Can I use this for bulk processing of many files?

Absolutely. The core CLI supports automated stages and input/output directory management, making it suitable for batch processing large volumes of transcription data.

Does it support languages other than English?

Yes, it is specifically optimized for Chinese-English mixed content and technical terminology often found in global technology and engineering environments.

What happens if the AI API is unavailable?

The skill includes a fallback strategy where Claude can analyze the text directly and guide you through manual edits, which are then saved back to the dictionary to prevent future errors.

Why is the 'uv' Python tool required?

'uv' is used to manage dependencies in a fast, isolated environment. This ensures the skill runs consistently without interfering with your system's global Python installation.

Transcript Fixer

Name: Transcript Fixer
Author: nguyendinhquocx

bynguyendinhquocx

0•

생산성 및 워크플로

Refines speech-to-text transcriptions by correcting ASR errors and homophones through dictionary rules and AI-powered pattern recognition.

Transcript Fixer is a specialized tool for polishing raw speech-to-text output from meetings, lectures, and interviews. It utilizes a sophisticated three-stage pipeline that combines deterministic dictionary-based corrections with advanced AI models to resolve complex homophone errors and technical terminology mismatches, particularly in bilingual Chinese-English contexts. By maintaining a persistent SQLite database, the skill learns from manual corrections to build a personalized knowledge base, ensuring that frequent transcription mistakes are handled automatically and consistently over time.

주요 기능

01Native support for mixed Chinese-English technical terminology and homophone correction

020 GitHub stars

03Domain-specific dictionaries for specialized fields like finance, medical, or AI engineering

04Personalized learning engine that detects and suggests new correction patterns automatically

05Multi-stage correction pipeline combining dictionary rules and AI-powered analysis

06Visual word-level diff generation for easy review of changes in a browser-friendly HTML format

사용 사례

01Building a shared team dictionary of frequently mis-transcribed project names or industry jargon

02Cleaning up automated meeting minutes and interview transcripts for professional documentation

03Correcting technical terminology and ASR errors in recorded educational lectures and webinars

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nguyendinhquocx/code-ai transcript-fixer

For use in Claude.ai and ChatGPT

주요 기능

01Native support for mixed Chinese-English technical terminology and homophone correction

020 GitHub stars

03Domain-specific dictionaries for specialized fields like finance, medical, or AI engineering

04Personalized learning engine that detects and suggests new correction patterns automatically

05Multi-stage correction pipeline combining dictionary rules and AI-powered analysis

06Visual word-level diff generation for easy review of changes in a browser-friendly HTML format

사용 사례

01Building a shared team dictionary of frequently mis-transcribed project names or industry jargon

02Cleaning up automated meeting minutes and interview transcripts for professional documentation

03Correcting technical terminology and ASR errors in recorded educational lectures and webinars