ChristophSchuhmann/advanced-soundscapes-stage-1

Advanced Soundscapes Stage 1 — Raw Components

This dataset contains Stage 1 output from the LAION Universal Audio Annotation Pipeline (UAAP) data generation plan.

Contents

• —0 shard(s) containing 0 soundscape recipes with raw audio components
• —Each soundscape row includes:
• —recipe.json — full recipe with timeline, events, loudness, speaker IDs, overlap/density settings
• —spkN.flac / spkN.json — raw speech components + full source metadata
• —musicN.flac / musicN.json — raw music components + captions
• —sfxN.flac / sfxN.json — raw SFX components + captions
• —vbN.flac / vbN.json — raw vocal burst components + metadata

Format

WebDataset .tar shards. Each row is a directory inside the tar with the clip ID as the key.

Source Datasets

Stage 1 Pipeline

Stage 1 is CPU-only. It:

1. 1.Samples a recipe (archetype, duration, density, overlap, speaker config)
2. 2.Draws raw audio pieces from source datasets (language-balanced, deduplicated)
3. 3.Decodes & resamples to 16 kHz mono
4. 4.Computes exact mix plan (ground-truth timings, loudness, speaker IDs)
5. 5.Gathers full source metadata
6. 6.Writes WebDataset rows

Stage 2 (GPU) will: caption missing fields, fuse via Gemma-4-12B, render final mixdown MP3.

Recipe Archetypes

Pipeline Script

The generation pipeline is included as stage1_pipeline.py. Run with:

python stage1_pipeline.py --num-soundscapes 1000 --num-workers 4 --output-dir ./output --upload