Clips AI is an open-source Python library that automatically converts longform video into short clips with adjusted aspect ratios, built specifically for audio-centric content like podcasts, interviews, and speeches. Clips are found by analyzing the video’s transcript – generated via WhisperX, which wraps OpenAI’s Whisper model for word-level timing detection – to identify segments worth extracting.
Speaker-focused dynamic reframing is the second AI process doing real work here: Pyannote handles speaker diarization to determine who’s talking at any given moment, and the system uses that to reframe the video around the current speaker as it converts between aspect ratios (16:9 to 9:16, for example). That’s a meaningfully harder problem than a generic center-crop – a naive crop cuts off whoever isn’t positioned in the frame’s center, while speaker-aware reframing follows the conversation.
As an open-source library rather than a hosted app, Clips AI targets developers who want programmatic video-repurposing capability built into their own tools or pipelines, not end users looking for a point-and-click clip generator. No pricing applies – it’s free and open source.
