Name: Faster Whisper Gpu
Author: skills

What breaks without openclaw faster whisper gpu skill

API costs for every audio file. CUDA version mismatches crashing installs. VRAM limits blocking large models.

→

Local GPU-speed transcription at zero API cost × 350-star proven skill ÷ 30–60 minutes ÷ CUDA GPU only = private high-volume transcription.

openclaw faster whisper gpu skill — what it actually does

Transcribes audio locally at GPU speed with Faster-Whisper

Sends zero data to external APIs — fully private

Supports tiny through large-v3 models via WHISPER_MODEL

Requires NVIDIA GPU with CUDA 11.8+ for acceleration

Uses base model on GPUs with less than 8GB VRAM

Security check — openclaw faster whisper gpu skill

Privacy score: 7/10 — accesses connected platform APIs only. Lock it: review OAuth scopes before install, confirm Linux; NVIDIA GPU with CUDA ≥11.8; OpenClaw ≥1.0 compatibility.

Quick start — openclaw faster whisper gpu skill in 30–60 minutes

Setup time: 30–60 minutes

You need:

OpenClaw core
NVIDIA GPU with CUDA
Faster-Whisper installed

Install the package:

clawhub install skills/faster-whisper-gpu
# Requires: pip install faster-whisper --break-system-packages

Install faster-whisper with CUDA support

Install skill

Set WHISPER_MODEL (tiny/base/large-v3)

Run /whisper-gpu audio.mp3

Troubleshooting openclaw faster whisper gpu skill

1. CUDA version must match PyTorch build — mismatched CUDA causes cryptic failures

2. large-v3 requires ≥8GB VRAM — use base for low-memory GPUs

Compatibility & status

Works with: Linux; NVIDIA GPU with CUDA ≥11.8; OpenClaw ≥1.0 advanced Last updated: Oct 2025 ★ 350 on GitHub MIT

Official docs →

View on GitHub →

Related — more like openclaw faster whisper gpu skill

More by skills

Homeassistant

Integrations

Excalidraw Diagram Generator

Skills

Camoufox Stealth Browser

Skills

Linear Webhook

Integrations

Every cloud transcription call is a cost that local GPU eliminates. Install before the next high-volume audio batch starts.

Get it on GitHub →