music/release

Music release workflow — prompt → instrumental track → DistroKid-ready bundle.

Category: music
Source: workflows/music/release.py

Input Schema

Field	Type	Default	Description
`ai_disclosure`	`boolean`	`true`	Workflow invariant — must be True. AI-generated content tagging.
`album_name`	`object`	—	Album name (defaults to title for singles)
`artist_name`	`string`	required	Artist persona name (one of your registered DistroKid artists)
`audio_model`	`string`	`"fal-ai/stable-audio"`	Music generation model (commercial-cleared)
`bit_depth`	`integer`	`16`
`copyright`	`object`	—	© line; auto-built if omitted
`cover_prompt`	`string`	`""`	Cover art prompt. Defaults to a derivative of the music prompt if empty.
`cover_size`	`integer`	`3000`	Square pixel size
`distribution`	`object`	—	See _Distribution. auto_submit must be False in Phase 1.
`duration_secs`	`integer`	`180`	Track length in seconds (≥30 for Spotify play-counting)
`explicit`	`boolean`	`false`
`featured_artist`	`string`	`""`	Featured artist credit (optional)
`genre`	`string`	required	Spotify genre taxonomy (e.g. Ambient, Lo-Fi, Classical)
`image_model`	`object`	—	Image model override
`isrc`	`object`	—	ISRC (12 chars). Null lets DistroKid auto-issue.
`label`	`object`	—	Label name (defaults to artist_name)
`language`	`string`	`"instrumental"`	ISO code or ‘instrumental’
`lyrics`	`object`	—	Required when vocals=True
`mood_tags`	`string[]`	—
`output_dir`	`object`	—	Parent directory for release bundles. Defaults to a temp dir.
`pitch_to_editorial`	`boolean`	`true`
`playlist_targets`	`string[]`	—
`prompt`	`string`	required	Music generation prompt — describes mood, style, instrumentation
`publishing_rights`	`object`	—	℗ line; auto-built if omitted
`regenerate`	`object`	—	When set, this run is a regeneration. Workflows may read `direction` / `keep` / `extra_instructions` to modulate prompts; the engine persists `parent_run_id` and `parent_variant_index` as run lineage columns.
`release_date`	`string`	required	YYYY-MM-DD; ≥7 days out for S4A pitching
`sample_rate`	`integer`	`44100`
`seed`	`object`	—	Pinned seed shared by every segment generation — keeps the harmonic and timbral palette consistent across the whole track. When omitted, the workflow auto-generates one and records it in release.json so the run is reproducible.
`segment_duration_secs`	`integer`	`30`	Per-segment generation length (Stable Audio Open caps ~30-47s)
`segment_prompts`	`object`	—	Override the auto-generated per-segment prompts. When set, generates exactly len(segment_prompts) segments and skips the song-shape templates.
`songwriter_splits`	`object[]`	—	Publishing royalties — at least one entry; shares sum to 100. See _SongwriterSplit.
`subgenres`	`string[]`	—
`target_lufs`	`number`	`-14.0`	Spotify normalizes to -14 LUFS
`title`	`string`	required	Track title
`upc`	`object`	—	UPC (12 digits). Optional.
`variants`	`integer`	`1`	Number of independent variant executions (1–10). When > 1, the engine runs the workflow N times with different sampling, producing N outputs.
`vocals`	`boolean`	`false`	True routes to ACE-Step (Phase 3, deferred). Phase 1 supports instrumental only.

Output Schema

Field	Type	Default	Description
`audio_path`	`string`	required	Mastered 16-bit/44.1kHz WAV at -14 LUFS
`bundle_dir`	`string`	required	Directory containing the release triple
`cover_meta`	`object`	—
`cover_path`	`string`	required	Square sRGB JPEG, ≤10MB
`kind`	`object`	—	Variant card shape: video / carousel / image / text. Surfaced on the per-variant entry of the run-output API and used by gallery UIs to pick the right layout.
`loudness`	`object`	—
`metadata_csv_path`	`string`	required	DistroKid bulk-upload CSV row
`release_json_path`	`string`	required	Machine-readable sidecar with full metadata
`warnings`	`string[]`	—

Task Pipeline

prepare_release → generate_track → generate_cover → assemble_bundle

Task	Description
`prepare_release`	Validate Phase-1 invariants before any expensive work fires.
`generate_track`	Generate the instrumental track as N evolving segments.
`generate_cover`	Generate square cover art via the existing image stage.
`assemble_bundle`	Crossfade-concat segments, master, process cover, write metadata.

Run-spec example

Save the YAML below as my-run.yaml, edit the values, and run with the CLI or POST it to the API. Required fields are uncommented; optional knobs are documented above the input: block — copy any line under input: and uncomment to set.

workflow: music/release

# Optional fields — copy any line(s) under `input:` and uncomment to set:
# Workflow invariant — must be True. AI-generated content tagging.
#   ai_disclosure: true
#
# Album name (defaults to title for singles)
#   album_name: null
#
# Music generation model (commercial-cleared)
#   audio_model: fal-ai/stable-audio
#
#   bit_depth: 16
#
# © line; auto-built if omitted
#   copyright: null
#
# Cover art prompt. Defaults to a derivative of the music prompt if empty.
#   cover_prompt: ""
#
# Square pixel size
# [min=1400, max=3000]
#   cover_size: 3000
#
# See _Distribution. auto_submit must be False in Phase 1.
#   distribution: {}
#
# Track length in seconds (≥30 for Spotify play-counting)
# [min=30, max=300]
#   duration_secs: 180
#
#   explicit: false
#
# Featured artist credit (optional)
#   featured_artist: ""
#
# Image model override
#   image_model: null
#
# ISRC (12 chars). Null lets DistroKid auto-issue.
#   isrc: null
#
# Label name (defaults to artist_name)
#   label: null
#
# ISO code or 'instrumental'
#   language: instrumental
#
# Required when vocals=True
#   lyrics: null
#
#   mood_tags: []
#
# Parent directory for release bundles. Defaults to a temp dir.
#   output_dir: null
#
#   pitch_to_editorial: true
#
#   playlist_targets: []
#
# ℗ line; auto-built if omitted
#   publishing_rights: null
#
#   sample_rate: 44100
#
# Pinned seed shared by every segment generation — keeps the harmonic and timbral palette consistent across the whole track. When omitted, the workflow auto-generates one and records it in release.json so the run is reproducible.
#   seed: null
#
# Per-segment generation length (Stable Audio Open caps ~30-47s)
# [min=15, max=60]
#   segment_duration_secs: 30
#
# Override the auto-generated per-segment prompts. When set, generates exactly len(segment_prompts) segments and skips the song-shape templates.
#   segment_prompts: null
#
# Publishing royalties — at least one entry; shares sum to 100. See _SongwriterSplit.
#   songwriter_splits: []
#
#   subgenres: []
#
# Spotify normalizes to -14 LUFS
# [min=-23.0, max=-9.0]
#   target_lufs: -14.0
#
# UPC (12 digits). Optional.
#   upc: null
#
# True routes to ACE-Step (Phase 3, deferred). Phase 1 supports instrumental only.
#   vocals: false
#

input:
  # Artist persona name (one of your registered DistroKid artists)
  artist_name: ""

  # Spotify genre taxonomy (e.g. Ambient, Lo-Fi, Classical)
  genre: ""

  # Music generation prompt — describes mood, style, instrumentation
  prompt: ""

  # YYYY-MM-DD; ≥7 days out for S4A pitching
  release_date: ""

  # Track title
  title: ""

Run it locally:

fab-workflow --from-file my-run.yaml

Or submit over the wire — the same file is the request body:

curl -X POST 'https://gofabric.dev/v1/workflows/runs?name=music/release' \
  -H 'Authorization: Bearer fab_xxx' \
  -H 'content-type: application/yaml' \
  --data-binary @my-run.yaml

Every workflow also accepts the universal WorkflowInput fields — variants (1–10 fan-out) and regenerate (creative-direction hints with run lineage). See Run-specs (YAML / TOML / JSON) for the full top-level shape (metadata, priority, bundle, parent, etc.).

Worked examples

These checked-in run-specs exercise this workflow — good starting points to copy and tweak: