clips-machine

What This Skill Does

The Clips Machine skill is designed to automate the process of transforming long video content into engaging, short-form clips suitable for platforms like TikTok, Instagram Reels, and YouTube Shorts. It begins by accepting a wide range of video sources, including YouTube URLs, local video files, and even streams from platforms like Twitch. The skill then utilizes Whisper, a free and locally run transcription tool, to generate accurate transcripts with precise timestamps.

Leveraging AI analysis, Clips Machine identifies potentially viral moments within the transcribed content. These moments are identified based on several factors, including strong opening hooks, emotional peaks, quotable lines, controversial statements, surprising facts, and actionable advice. Each detected moment is assigned a 'virality score.' The skill then automatically selects and cuts the best 30-60 second segments based on these scores.

To enhance the clips, the skill adds trendy, animated captions. Users can choose from several styles, including the popular 'hormozi' style (word-by-word highlighting), a clean 'minimal' style, 'karaoke' style, a 'news' lower third, or a 'meme' style with Impact font. Finally, the skill exports these captioned clips in a vertical 9:16 aspect ratio, ready for immediate upload to short-form video platforms. The entire process is designed to be self-contained, using 100% free tools that run locally on your machine, meaning no external modules or paid services are required.

Installation

To install the Clips Machine skill, you will need to have a few dependencies set up on your system.

Dependencies:

• FFmpeg: A powerful multimedia framework. On macOS, you can install it using Homebrew: brew install ffmpeg. On Linux, you can typically install it via your package manager: sudo apt install ffmpeg.
• yt-dlp: A command-line program to download videos from YouTube. On macOS: brew install yt-dlp. On Linux, you can install it using pip: pip install yt-dlp.
• Whisper.cpp: A high-performance inference engine for OpenAI's Whisper speech-to-text model. On macOS: brew install whisper-cpp. On Linux, you will need to build it from source.

Skill Installation Command: Once the dependencies are met, you can install the skill using the following command:

clawhub install openclaw/skills/skills/mayank8290/clips-machine

Use Cases

1. Content Creators: Quickly repurpose long-form content (podcasts, interviews, lectures, livestreams) into multiple short, engaging clips for social media promotion, driving traffic back to the original content.
2. Marketers: Extract key marketing messages, testimonials, or product highlights from longer videos to create easily digestible social media ads or campaign assets.
3. Educators: Turn lectures or educational videos into bite-sized learning modules or study aids for students.
4. Journalists/Researchers: Identify and extract pivotal moments or quotes from interviews or news segments for quick dissemination.
5. Personal Use: Create entertaining highlights from personal videos or vlogs to share with friends and family on platforms like TikTok.

Example Prompts

1. Turn this podcast into viral clips: https://youtube.com/watch?v=EXAMPLE_ID
2. Extract the 5 best moments from my-interview.mp4 and add 'hormozi' style captions.
3. Create short clips from this Twitch VOD and export them with karaoke captions.

Tips & Limitations

Tips:

• Experiment with Styles: The hormozi style is often highlighted as the most viral. Try it first, but explore other styles like minimal or karaoke to see what best fits your content.
• Specify Clip Count: If you have a specific number of clips in mind, use the --clips argument (e.g., --clips 10) to guide the AI.
• High-Quality Input: The quality of the output clips is heavily dependent on the quality of the input video and audio. Clear audio and good video will yield better transcriptions and more accurate moment detection.
• Review Output: While the AI is powerful, it's always a good idea to review the generated clips and the summary markdown file to ensure they accurately represent the intended message or highlights.

Limitations:

• Local Processing Power: Since the skill runs entirely locally, the processing time and resource usage (CPU, RAM, disk space) can be significant, especially for very long or high-resolution videos.
• Transcription Accuracy: The accuracy of transcription depends on the audio quality and the complexity of the language or accents. Errors in transcription can lead to less accurate moment detection.
• AI Interpretation: The AI's definition of 'viral' is based on learned patterns. It may not always perfectly align with human subjective judgment of what makes a clip engaging for a specific niche audience.
• Dependency Management: Users are responsible for installing and maintaining the required external dependencies (FFmpeg, yt-dlp, Whisper.cpp). Issues with these dependencies can prevent the skill from functioning correctly.
• No External API for Video Fetching: While it supports various sources, it relies on tools like yt-dlp for fetching. Complex or region-locked content might pose challenges.