How do I handle Out of Memory (OOM) errors on my GPU?

If your GPU is shared with other services, wait 30-60 seconds for VRAM to clear or consult the included troubleshooting guide for managing GPU resource unloading.

What are the hardware requirements for this Whisper skill?

The skill is optimized for NVIDIA GPUs with CUDA support. Using the large-v3 model requires approximately 6GB of VRAM, making it compatible with cards like the RTX 3090.

Which audio formats does the Whisper server support?

The server accepts common formats including WAV and MP3, though 16kHz mono WAV files provide the most optimal results for the inference engine.

Does my audio data leave my machine?

No. The transcription is performed entirely on your local machine using a whisper.cpp server, ensuring complete privacy for your audio files.

Whisper Transcription

Name: Whisper Transcription
Author: lawless-m

bylawless-m

•

データサイエンスとML

Provides high-performance local speech-to-text transcription using GPU-accelerated Whisper models.

概要

This skill integrates Claude with a local whisper.cpp server to enable seamless audio transcription directly within development workflows. By leveraging CUDA acceleration and the large-v3 model, it delivers high-accuracy speech-to-text conversion with low latency. It is designed for developers who need to process audio files, automate voice-to-text pipelines, or build voice-controlled scripts while keeping data entirely local for privacy and speed.

主な機能

Systemd service configuration for persistent background operation
Local GPU-accelerated transcription using whisper.cpp and CUDA
RESTful API endpoint for easy integration with curl and Python
5 GitHub stars
Support for high-accuracy large-v3 models with 6GB VRAM footprint
Smart VRAM management and troubleshooting for multi-GPU environments

ユースケース

Recording and transcribing live audio for rapid documentation and note-taking
Building automated speech-to-text processing pipelines in Python or Bash
Transcribing voice memos or meeting recordings into text directly from the terminal

概要

主な機能

Systemd service configuration for persistent background operation
Local GPU-accelerated transcription using whisper.cpp and CUDA
RESTful API endpoint for easy integration with curl and Python
5 GitHub stars
Support for high-accuracy large-v3 models with 6GB VRAM footprint
Smart VRAM management and troubleshooting for multi-GPU environments

ユースケース

Recording and transcribing live audio for rapid documentation and note-taking
Building automated speech-to-text processing pipelines in Python or Bash
Transcribing voice memos or meeting recordings into text directly from the terminal