What is the benefit of the multi-stage pipeline?

Stage 1 is instant and free using your local dictionary, while Stage 2 uses AI for context-aware fixes. Stage 3 combines both to ensure maximum accuracy and performance.

Do I need a specific AI API key to use this skill?

Yes, the AI correction stages require a GLM API key from BigModel.cn. However, basic dictionary-based corrections can be performed locally without an API.

How does the pattern learning feature work?

The skill monitors AI-driven corrections; if a specific error is corrected the same way multiple times with high confidence, it suggests adding that rule to your permanent local dictionary.

Can I use this for specialized technical fields?

Yes, you can create custom domains (such as 'Medical', 'Finance', or 'Engineering') to manage technical terminology specific to your industry or project.

Transcript Fixer & ASR Refiner

Name: Transcript Fixer & ASR Refiner
Author: daymade

bydaymade

•

108

生産性とワークフロー

Corrects speech-to-text transcription errors and refines ASR output using a hybrid dictionary and AI-driven workflow.

概要

This skill provides a specialized pipeline for cleaning up meeting notes, interviews, and lectures marred by Automatic Speech Recognition (ASR) errors. By combining lightning-fast local dictionary lookups with sophisticated AI correction via the GLM API, it eliminates homophones, technical terminology mistakes, and language-mixing issues. The skill uniquely features a learning mechanism that detects recurring errors and prompts users to update their personal knowledge base, ensuring the correction accuracy improves over time with every processed transcript.

主な機能

Personalized SQLite-backed knowledge base for persistent correction rules
Multi-stage correction pipeline utilizing dictionary, AI, or hybrid workflows
Visual HTML diff generation for side-by-side review of applied changes
Automatic pattern detection that learns from recurring transcription errors
108 GitHub stars
Support for mixed Chinese/English content and domain-specific terminology

ユースケース

Cleaning up ASR errors in corporate meeting minutes and technical briefings
Refining lecture notes and academic interviews containing specialized terminology
Building a collaborative, domain-specific dictionary for team-wide transcription accuracy

概要

主な機能

Personalized SQLite-backed knowledge base for persistent correction rules
Multi-stage correction pipeline utilizing dictionary, AI, or hybrid workflows
Visual HTML diff generation for side-by-side review of applied changes
Automatic pattern detection that learns from recurring transcription errors
108 GitHub stars
Support for mixed Chinese/English content and domain-specific terminology

ユースケース

Cleaning up ASR errors in corporate meeting minutes and technical briefings
Refining lecture notes and academic interviews containing specialized terminology
Building a collaborative, domain-specific dictionary for team-wide transcription accuracy