vocabulary: 0.1
provider: LALAL.AI
provider_id: lalal-ai
description: Operational and capability vocabulary for the LALAL.AI audio separation and voice technology
  platform, covering stems, neural splitters, processing modes, and task lifecycle.
updated: '2026-05-25'

stems:
  - id: vocals
    label: Vocals
    description: All vocal performance separated from the instrumental bed.
  - id: voice
    label: Voice
    description: Spoken-word voice isolated from background music and ambient noise (voice-clean surface).
  - id: drum
    label: Drum
    description: Drum kit and percussion stem.
  - id: bass
    label: Bass
    description: Bass guitar and low-frequency stem.
  - id: piano
    label: Piano
    description: Acoustic and electric piano stem.
  - id: electric_guitar
    label: Electric Guitar
  - id: acoustic_guitar
    label: Acoustic Guitar
  - id: synthesizer
    label: Synthesizer
    description: Synth stem (Phoenix splitter only).
  - id: strings
    label: Strings
    description: Strings stem (Phoenix splitter only).
  - id: wind
    label: Wind
    description: Wind instruments stem (Phoenix splitter only).
  - id: music
    label: Music
    description: Music component isolated from a mixed source (demuser surface).

splitters:
  - id: andromeda
    label: Andromeda
    description: Sixth-generation neural engine; default for vocal/instrumental and voice/noise separation.
    released: 2025-11-21
  - id: perseus
    label: Perseus
    description: Transformer-based network supporting Clear Cut and Deep Extraction; covers vocals, drums, bass,
      acoustic guitar, electric guitar, and piano.
    released: 2024-09-25
  - id: orion
    label: Orion
  - id: phoenix
    label: Phoenix
    description: Only splitter supporting synthesizer, strings, and wind stems.
  - id: lyra
    label: Lyra
    description: Proprietary dereverb and Demuser-stem network.
    released: 2025-09-30
  - id: lynx
    label: Lynx
  - id: auto
    label: Auto
    description: Server selects the latest splitter compatible with the requested stems.

processing_modes:
  - id: clear_cut
    label: Clear Cut
    description: Perseus extraction mode optimised for natural-sounding stems.
  - id: deep_extraction
    label: Deep Extraction
    description: Perseus extraction mode optimised for aggressive isolation.

encoder_formats:
  - mp3
  - wav
  - flac
  - aac
  - ogg

task_states:
  - id: success
    description: Task complete; stem_track and back_track URLs available.
  - id: progress
    description: Task still running; progress 0-100 reported on /check/.
  - id: error
    description: Task failed; check the error.code and error.message fields.
  - id: cancelled
    description: Task explicitly cancelled by /cancel/ or /cancel/all/.

voice_pack_categories:
  - id: legal_artist
    label: Legal Artist Pack
    description: LALAL.AI-licensed artist Voice Packs (ALEX_KAYE, STASIA_FAYE, NICOLAAS_HAAS, NIK_ZEL,
      OLIA_CHEBO, YVAR_DE_GROOT, VETRANA).
  - id: custom_clone
    label: Custom Clone Pack
    description: User-trained Voice Pack produced by the LALAL.AI Voice Cloner.

queues:
  - id: fast
    label: Fast Queue
    description: Priority processing queue with monthly minute allocation on paid plans.
  - id: relaxed
    label: Relaxed Queue
    description: Best-effort queue, unlimited minutes on paid plans, free 10 minutes on Starter.

retention:
  source_files_hours: 24
  description: Uploaded source files are deleted 24 hours after upload.