videopython

Minimal, LLM-friendly Python library for programmatic video editing, processing, and AI workflows.

Get Started API Reference

from videopython import Video
from videopython.base import FadeTransition

intro = Video.from_path("intro.mp4").resize(1080, 1920)
clip = Video.from_path("raw.mp4").cut(10, 25).resize(1080, 1920).resample_fps(30)
final = intro.transition_to(clip, FadeTransition(effect_time_seconds=0.5))
final = final.add_audio_from_file("music.mp3")
final.save("output.mp4")

Core Editing

Cut, resize, crop, change speed, reverse, freeze frames, picture-in-picture. Combine clips with fade, blur, and instant transitions. Multicam editing for podcast-style recordings.

Effects & Audio

Blur, zoom, color grading, vignette, Ken Burns, text and image overlay. Load, normalize, time-stretch, and mix audio tracks.

LLM-Driven Editing

JSON editing plans with full JSON Schema generation, dry-run validation, and an operation registry with rich constraints. Learn more →

AI Generation

Generate images, video, speech, and music from text prompts. SDXL, CogVideoX, Chatterbox Multilingual TTS, MusicGen — all local, no API keys. Learn more →

AI Video Analysis

Transcribe with speaker diarization, classify ambient audio, detect scene boundaries, and describe scenes with a vision-language model. One VideoAnalyzer runs the full pipeline and returns a serializable VideoAnalysis. Learn more →

AI Dubbing

Translate speech, clone the original voice, and re-time the dub onto the source — all in one pipeline. Whisper + MarianMT/Qwen3 + Chatterbox + Demucs. Resume cache for long runs, source-prosody-conditioned expressiveness, and a transcript-quality gate that rejects garbage input before paying for translation and TTS. Learn more →

Installation

pip install videopython          # core editing
pip install "videopython[ai]"    # + local AI features (GPU recommended)

Python >=3.10, <3.14. AI features run locally -- no cloud API keys required.

See the Installation Guide for FFmpeg setup and details.