asr

Transcribe audio files into text using local speech recognition models.

2|Updated Apr 6, 2026

Search Tags:#offline #transcription #audio #polish #asr #speech-recognition #coli

Ready to run

npx skills add https://github.com/augustscl/awesome-xiawang-skills --skill asr-augustscl

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: asr
Source: https://github.com/augustscl/awesome-xiawang-skills/tree/main/asr
Command: npx skills add https://github.com/augustscl/awesome-xiawang-skills --skill asr-augustscl

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Converts audio recordings into readable, editable text without sending data to external APIs, preserving privacy and enabling fast local transcription for meetings, interviews, and notes.

Core Features & Use Cases

Offline transcription using coli asr with support for sensevoice and whisper-tiny.en models.
Automatic language detection and optional AI polishing to fix punctuation, remove filler words, and improve readability while preserving meaning.
Exports transcripts with metadata (language, emotion, events, duration) and can save polished output as a Markdown file for archiving or sharing.
Ideal for transcribing meetings, interviews, voice notes, and podcasts where local processing or data privacy is required.

Quick Start

Transcribe the audio file meeting.m4a with polish enabled.

Frequently Asked Questions about asr

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I transcribe audio files to text offline without sending data to external APIs?▼

Offline transcription converts audio files to text locally using speech recognition models like sensevoice and whisper-tiny.en. It processes meetings, interviews, and recordings directly on your machine, preserving privacy without external API calls.

What languages does offline speech recognition support for meeting transcription?▼

Offline speech recognition supports transcription in Chinese, English, Japanese, Korean, and Cantonese. It automatically detects the spoken language and returns metadata including the detected language alongside the transcribed text.

Do I need ffmpeg or coli asr installed to transcribe audio recordings locally?▼

Yes, local audio transcription requires coli asr to run speech recognition models and recommends ffmpeg for audio file processing. These tools enable offline conversion of recordings into text with metadata like duration and emotion.

Can I automatically polish audio transcripts to fix punctuation and remove filler words?▼

Yes, optional AI polishing fixes punctuation, removes filler words, and improves readability while preserving meaning. Polished transcripts can be exported as Markdown files with metadata including language, emotion, event, and duration.

What metadata is included when transcribing audio files with sensevoice or whisper-tiny.en?▼

Audio transcription returns JSON metadata containing the transcribed text, detected language, emotion, event, and duration. This metadata is exported alongside polished Markdown output for archiving or sharing.

Is local speech recognition suitable for transcribing short recordings and voice notes?▼

Local speech recognition is ideal for short recordings, voice notes, meetings, interviews, and podcasts. It processes audio offline using sensevoice or whisper-tiny.en models, ensuring data privacy for sensitive content.