2025-11-05 –, Blue Language: English
Tabbers is a platform made by guitarists, for guitarists, enabling them to teach each other guitar riffs simply by playing.
Our core technology turns guitar-playing videos into rich, instructional content. We use audio and video analysis to generate tablature, color-coded finger visualizations, and other enhancements that make any riff easy to learn.
It is useful whether you’re a tutor teaching a targeted passage, a social media content creator, or an amateur player exploring phrasing and analyzing techniques.
Tabbers aims to reduce the friction between inspiration and understanding.
Our motto: You just play - we do the rest.
In this talk, I’ll walk through the core architecture and design of the current version of our model, a project that uses modern (and less modern) AI techniques to process both video and audio of guitar performances.
The system includes:
• Audio analysis for channel separation and cleaning, and note detection
• Computer vision application to detect finger placement on the fretboard
• Video and audio based technique extraction (bends, slides, hammer-on, etc.)
• Fusion pipeline for aligning visual hand positioning with note transcription
• Tablature generation engine
• Comprehensive web app platform
We’ll dive into key technical challenges like handling polyphonic audio, fretboard tilt, aligning asynchronous audio and video, and distinguishing techniques like slides or bends - and how we solved them using AI-first design thinking.
Tools & Stack:
• Backend in Python (FastAPI, OpenCV, Librosa, Pandas, Numpy, etc.)
• A verity of open-source audio and visual models
• Pyppeteer, ffmpeg
• Next.js frontend with custom SVG rendering for tabs
Who Is This For:
• PyData developers interested in applied ML for video/audio
• Professional and recreational musicians
• Engineers looking to build educational interfaces powered by AI
What Attendees Will Learn:
• Basics of Audio and Video analysis
• Trade-offs between heuristic methods and ML in "free-tier-solopreneur" mode
• How to turn a hobbyist tool into a structured AI product
No previous knowledge expected
Lior Kupfer, creator of Tabbers.
I love playing guitar, data, and innovation - and now I’m trying to mix 'em together.