Is this transcription skill really free to use?

Yes. Because it runs the Whisper model locally on your Mac's hardware, there are no per-minute fees or subscription costs associated with the transcription.

Does this skill work on Windows or Intel-based Macs?

No, this specific implementation is optimized for Apple Silicon (M1/M2/M3/M4 chips) and requires macOS.

Can I translate audio from other languages into English?

Yes, the skill supports a translation flag that allows you to transcribe and translate non-English audio directly into English text.

How fast is the local transcription compared to cloud APIs?

After the initial model load into memory, transcription is extremely fast, typically taking about 1 second per voice message.

How much storage space does the Whisper model require?

The first time you run the skill, it will download a model approximately 1.5GB in size to your local machine.

Local Whisper Transcription

Name: Local Whisper Transcription
Author: sundial-org

bysundial-org

•

150

•

生産性とワークフロー

Transcribes voice messages locally on Apple Silicon using MLX Whisper to eliminate API costs and ensure data privacy.

Local Whisper Transcription is a specialized skill for OpenClaw that enables high-quality, local speech-to-text processing optimized for Apple Silicon Macs. By utilizing the MLX framework, it transcribes audio from platforms like Telegram and WhatsApp directly on your machine, bypassing the need for expensive third-party APIs like OpenAI, Groq, or AssemblyAI. This tool is ideal for users who handle large volumes of voice messages and want to maintain complete data sovereignty without incurring recurring usage fees.

主な機能

01Zero-cost local transcription using MLX-optimized Whisper models

02Built-in translation capabilities from any language to English

03Enhanced data privacy with 100% offline audio processing

04Seamless integration with Telegram and WhatsApp voice message workflows

05Hardware-accelerated performance on Apple Silicon (M1, M2, M3, M4)

06150 GitHub stars

ユースケース

01Transcribing high volumes of Telegram voice notes without incurring per-minute API charges

02Processing sensitive or confidential audio recordings in a secure, offline environment

03Automating transcription workflows within the OpenClaw gateway on macOS

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add sundial-org/awesome-openclaw-skills whisper-mlx-local

For use in Claude.ai and ChatGPT

Download Skill