openapi: 3.0.3
info:
  title: Kits AI Stem Splitter Voice Conversion API
  version: v1
  description: 'The Kits AI API (operated by Arpeggi Labs) provides studio-quality AI music and audio tools as asynchronous inference jobs: voice conversion, vocal separation, stem splitting, and voice-model blending, plus a catalog of royalty-free artist voice models. All create endpoints enqueue a job and return its status; poll the corresponding fetch endpoint for the signed output file URLs. This specification was generated by API Evangelist from the published Kits AI documentation (https://docs.kits.ai/); the provider lists an openapi.json in its llms.txt but the file was not retrievable at generation time.'
  contact:
    name: Kits AI Support
    url: https://help.kits.ai/
  x-generated-by: api-evangelist-enrichment
  x-source: https://docs.kits.ai/api-reference
servers:
- url: https://arpeggi.io/api/kits/v1
  description: Production
security:
- bearerAuth: []
tags:
- name: Voice Conversion
  description: Convert an input performance to a target voice model.
paths:
  /voice-conversions:
    post:
      tags:
      - Voice Conversion
      operationId: createVoiceConversion
      summary: Create a new voice conversion job
      description: Creates a voice conversion inference job and adds it to the inference queue. All requests must be made as multipart form requests. The response contains the job status.
      requestBody:
        required: true
        content:
          multipart/form-data:
            schema:
              type: object
              required:
              - voiceModelId
              - soundFile
              properties:
                voiceModelId:
                  type: integer
                  description: ID of the voice model.
                soundFile:
                  type: string
                  format: binary
                  description: Input audio. Supports wav, mp3, or flac. Max 100MB.
                conversionStrength:
                  type: number
                  minimum: 0
                  maximum: 1
                  description: Amount of accent from the voice model (0 to 1).
                modelVolumeMix:
                  type: number
                  minimum: 0
                  maximum: 1
                  description: Mix between input volume and model volume (0 to 1).
                pitchShift:
                  type: number
                  minimum: -24
                  maximum: 24
                  description: Semitones to shift the input audio (-24 to 24).
                pre:
                  $ref: '#/components/schemas/PreProcessingEffects'
                post:
                  $ref: '#/components/schemas/PostProcessingEffects'
      responses:
        '201':
          description: Created
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/InferenceJob'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '422':
          $ref: '#/components/responses/UnprocessableEntity'
        '429':
          $ref: '#/components/responses/TooManyRequests'
    get:
      tags:
      - Voice Conversion
      operationId: listVoiceConversions
      summary: Fetch voice conversions
      description: Returns a paginated list of voice conversion inference jobs.
      parameters:
      - $ref: '#/components/parameters/Order'
      - $ref: '#/components/parameters/Page'
      - $ref: '#/components/parameters/PerPage'
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/PaginatedInferenceJobs'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '429':
          $ref: '#/components/responses/TooManyRequests'
  /voice-conversions/{id}:
    get:
      tags:
      - Voice Conversion
      operationId: getVoiceConversion
      summary: Fetch a voice conversion by ID
      description: Returns a single voice conversion inference job by id.
      parameters:
      - $ref: '#/components/parameters/JobId'
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/InferenceJob'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '429':
          $ref: '#/components/responses/TooManyRequests'
components:
  parameters:
    JobId:
      name: id
      in: path
      required: true
      schema:
        type: integer
    PerPage:
      name: perPage
      in: query
      description: Page size (default 10).
      schema:
        type: integer
        default: 10
    Page:
      name: page
      in: query
      description: Page offset (default 1).
      schema:
        type: integer
        default: 1
    Order:
      name: order
      in: query
      description: Sort order, asc (default) or desc.
      schema:
        type: string
        enum:
        - asc
        - desc
        default: asc
  schemas:
    PaginatedInferenceJobs:
      type: object
      properties:
        data:
          type: array
          items:
            $ref: '#/components/schemas/InferenceJob'
        meta:
          $ref: '#/components/schemas/PaginationMeta'
    InferenceJob:
      type: object
      description: An inference job created through the API.
      properties:
        id:
          type: integer
        createdAt:
          type: string
        type:
          type: string
          enum:
          - infer
          - tts
        status:
          type: string
          enum:
          - running
          - success
          - error
          - cancelled
        jobStartTime:
          type: string
          nullable: true
        jobEndTime:
          type: string
          nullable: true
        outputFileUrl:
          type: string
          nullable: true
          description: Signed URL for the output file (expires in 4 hours).
        lossyOutputFileUrl:
          type: string
          nullable: true
        recombinedAudioFileUrl:
          type: string
          nullable: true
        voiceModelId:
          type: integer
          nullable: true
        model:
          allOf:
          - $ref: '#/components/schemas/VoiceModel'
          nullable: true
    PostProcessingEffects:
      type: object
      description: Effects run on the output file after conversion.
      properties:
        Chorus:
          type: object
          properties:
            rate_hz:
              type: number
              minimum: 0
              maximum: 100
            depth:
              type: number
            centre_delay_ms:
              type: number
            feedback:
              type: number
            mix:
              type: number
        Reverb:
          type: object
          properties:
            room_size:
              type: number
              minimum: 0
              maximum: 1
            damping:
              type: number
              minimum: 0
              maximum: 1
            wet_level:
              type: number
              minimum: 0
              maximum: 1
            dry_level:
              type: number
              minimum: 0
              maximum: 1
            width:
              type: number
            freeze_mode:
              type: number
        Compressor:
          type: object
          properties:
            threshold_db:
              type: number
            ratio:
              type: number
            attack_ms:
              type: number
            release_ms:
              type: number
        Delay:
          type: object
          properties:
            delay_seconds:
              type: number
            feedback:
              type: number
            mix:
              type: number
              minimum: 0
              maximum: 1
    PreProcessingEffects:
      type: object
      description: Effects run on the input file before conversion.
      properties:
        NoiseGate:
          type: object
          properties:
            threshold_db:
              type: number
            ratio:
              type: number
            attack_ms:
              type: number
            release_ms:
              type: number
        HighPassFilter:
          type: object
          properties:
            cutoff_frequency_hz:
              type: number
              minimum: 0
              maximum: 99999
        LowPassFilter:
          type: object
          properties:
            cutoff_frequency_hz:
              type: number
              minimum: 0
              maximum: 99999
        Compressor:
          type: object
          properties:
            threshold_db:
              type: number
            ratio:
              type: number
            attack_ms:
              type: number
            release_ms:
              type: number
    VoiceModel:
      type: object
      description: A voice model that may be user created or internal to Kits.ai.
      properties:
        id:
          type: integer
        title:
          type: string
        isUsable:
          type: boolean
        tags:
          type: array
          nullable: true
          items:
            type: string
        imageUrl:
          type: string
          nullable: true
        demoUrl:
          type: string
          nullable: true
        twitterLink:
          type: string
          nullable: true
        instagramLink:
          type: string
          nullable: true
        tiktokLink:
          type: string
          nullable: true
        spotifyLink:
          type: string
          nullable: true
        youtubeLink:
          type: string
          nullable: true
    PaginationMeta:
      type: object
      properties:
        total:
          type: integer
        perPage:
          type: integer
        currentPage:
          type: integer
        lastPage:
          type: integer
        firstPage:
          type: integer
        firstPageUrl:
          type: string
        lastPageUrl:
          type: string
        nextPageUrl:
          type: string
          nullable: true
        previousPageUrl:
          type: string
          nullable: true
  responses:
    TooManyRequests:
      description: Rate limit exceeded.
      content:
        text/plain:
          schema:
            type: string
    Forbidden:
      description: Attempted to fetch a resource that does not belong to the requesting user.
      content:
        text/plain:
          schema:
            type: string
    Unauthorized:
      description: Authentication was not provided or was invalid.
      content:
        text/plain:
          schema:
            type: string
    UnprocessableEntity:
      description: Request body or query params are invalid.
      content:
        text/plain:
          schema:
            type: string
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      description: 'API key issued from the Kits AI API access page (https://app.kits.ai/api-access), sent as `Authorization: Bearer <api-key>`.'