audioglow 0.1.3

GPU-accelerated audio-reactive music video generator

audioglow

GPU-accelerated audio-reactive music video generator. Turns your tracks into visual content with OpenGL shaders — album covers react to beats, bass, and frequencies in real time.

Installation

pip install audioglow

Requirements: ffmpeg must be installed and available in PATH.

Quick Start

Python API

from audioglow import render_video

# Minimal — just tracks, output, and video type:
result = render_video(
    tracks=["tracks/my_song/"],
    output="output.mp4",
    video_type="cover_audio_reactive",
)

# Custom settings:
result = render_video(
    tracks=["tracks/song1/", "tracks/song2/"],
    output="mix.mp4",
    video_type="cover_audio_reactive",
    fps=60,
    resolution=(1920, 1080),
    bloom_strength=2.0,
    bitrate="16M",
    crossfade_ms=3000,
)

# result.video_path  — Path to the rendered .mp4
# result.timestamps  — track start times:
#   [{"title": "song1", "start_ms": 0}, {"title": "song2", "start_ms": 187000}]

Each track directory must contain:

• One .wav audio file
• One cover image (.png, .jpg, or .jpeg)

CLI

# Single config
audioglow --config config.json --tracks-dir ./tracks --output ./out

# Batch — all configs in configs/
audioglow --all --output ./out

Video Types

Type	Description
cover_audio_reactive	Album cover with reactive zoom, shake, bloom on blurred background
cover_overlay_green	Cover + green-screen video converted to palette-colored particles
cover_overlay_spectrum	Cover + audio frequency spectrum wave visualization
cover_combo	All effects combined: particles + spectrum + cover

Overlay types (cover_overlay_green, cover_combo) require overlay_path pointing to a green-screen video file.

Parameters

Parameter	Default	Description
resolution	(1920, 1080)	Video resolution
fps	30	Frames per second
cover_size	800	Cover art size (px)
zoom_strength	0.3	Bass-reactive zoom intensity
shake_strength	4.0 / 0.5*	Beat-reactive camera shake
bloom_strength	1.5 / 3.0*	Glow intensity on beats
blur_sigma	40 / 35*	Background blur radius
bitrate	"12M"	Video bitrate
crossfade_ms	4000	Crossfade between tracks (ms)
overlay_path	None	Path to overlay video (required for overlay types)
config_file	None	JSON config file (overrides type defaults, kwargs override both)
tracks_count	all	Max tracks to use

*cover_combo type uses different defaults optimized for layered effects.

Config Priority

Type Defaults  →  JSON config file  →  kwargs
   (lowest)                           (highest)

JSON Config Example

{
  "video_type": "cover_audio_reactive",
  "resolution": [1920, 1080],
  "fps": 30,
  "zoom_strength": 0.3,
  "shake_strength": 4.0,
  "bloom_strength": 1.5,
  "blur_sigma": 40,
  "bitrate": "12M",
  "tracks_count": 2,
  "crossfade_ms": 4000
}

How It Works

1. Audio analysis — librosa extracts RMS energy, onset strength, and 64-band mel spectrogram
2. Palette extraction — dominant colors pulled from cover art
3. GPU rendering — ModernGL (OpenGL 3.3) fragment shaders render each frame
4. Encoding — ffmpeg encodes frames with hardware-accelerated H.264 (VideoToolbox / NVENC / VAAPI / QSV, fallback to libx264)

License

MIT