kdenlive-tweaks

Setup scripts and configuration for Kdenlive on Linux: Contour ShuttleXpress jog controller, GPU-accelerated Whisper speech-to-text, and SAM-2 object detection with CUDA 13.2 support.

Tested on:

• Ubuntu 22.04 / Pop!_OS
• Kdenlive 26.04
• NVIDIA RTX 5000 series (CUDA 13.2, driver 595.84)

---

Contents

shuttlexpress/

Configure the Contour ShuttleXpress as a native Kdenlive JogShuttle device on Linux.

• udev rule setup for device access
• Device detection and event node discovery
• Recommended 5-button mapping for Kdenlive editing
• Input Remapper fallback profile

ai-tools/

Install and configure GPU-accelerated AI features for Kdenlive:

• Whisper speech-to-text (openai-whisper, CUDA-accelerated)
• SAM-2 object detection
• OpenCV and SRT support
• Supports both Kdenlive venv and system packages mode

---

Quick Start

ShuttleXpress

cd shuttlexpress
sudo ./scripts/setup-udev-rule-shuttlexpress.sh
./scripts/check-device.sh

Then in Kdenlive: Settings → Configure Kdenlive → JogShuttle → enable and set device to the detected /dev/input/eventX.

AI Tools (system packages mode)

cd ai-tools
./scripts/install-ai-deps.sh --system

Then in Kdenlive: Settings → Configure Kdenlive → Plugins → set Python to /usr/bin/python3, check Use system packages only under Object Detection.

---

Requirements

• Linux (Debian/Ubuntu/Pop!_OS recommended)
• Python 3.10+
• NVIDIA GPU with CUDA 12.6+ (13.2 recommended for RTX 5000 series)
• Kdenlive 26.04+
• ffmpeg, usbutils, evtest

---

License

MIT