generated: '2026-07-18'
method: derived
source: openapi/arpeggi-labs-kits-openapi.yml
description: >-
  Entity-relationship graph for the Kits AI API, derived from the OpenAPI
  schemas and id-reference fields. Jobs (voice conversion / vocal separation /
  stem split / voice blend) reference Voice Models by id; the blender produces a
  new output Voice Model.
entities:
  - name: VoiceModel
    id: id (integer)
    description: A voice model, user-created or internal to Kits.ai. May be an instrument model.
    key_fields: [id, title, isUsable, tags, imageUrl, demoUrl]
  - name: InferenceJob
    id: id (integer)
    description: A voice conversion (type infer/tts) job that renders input audio through a voice model.
    key_fields: [id, type, status, voiceModelId, outputFileUrl]
  - name: SeparationJob
    id: id (integer)
    description: A vocal separation or stem splitter job (type separate) that emits vocal and instrument stems.
    key_fields: [id, type, status, vocalAudioFileUrl, stemFileUrls]
  - name: VoiceBlenderJob
    id: id (integer)
    description: A job that blends 2-4 voice models into a new output voice model.
    key_fields: [id, model1Id, model2Id, alpha, outputModelId, status]
relationships:
  - from: InferenceJob
    to: VoiceModel
    type: belongs_to
    via: voiceModelId
  - from: VoiceBlenderJob
    to: VoiceModel
    type: has_many
    via: [model1Id, model2Id, model3Id, model4Id]
    note: Two to four input models are blended.
  - from: VoiceBlenderJob
    to: VoiceModel
    type: has_one
    via: outputModelId
    note: The blend produces one new output voice model, usable once status is "success".
enumerations:
  job_status: [running, success, error, cancelled]
  inference_type: [infer, tts]
  separation_type: [separate]