arazzo: 1.0.1
info:
  title: OpenAI Chat then Speak
  summary: Generate a chat reply, then synthesize it to spoken audio.
  description: >-
    Creates a chat completion from a user prompt and then sends the assistant
    reply to the text-to-speech endpoint to produce spoken audio. Every step
    spells out its request inline so the flow can be read and executed without
    opening the underlying OpenAPI description.
  version: 1.0.0
sourceDescriptions:
- name: openaiApi
  url: ../openapi/openai-openapi-master.yml
  type: openapi
workflows:
- workflowId: chat-then-speak
  summary: Create a chat reply and convert it to speech audio.
  description: >-
    Asks a chat model to answer a prompt, then feeds the reply text to the
    speech endpoint and returns the synthesized audio.
  inputs:
    type: object
    required:
    - apiKey
    - chatModel
    - prompt
    - speechModel
    - voice
    properties:
      apiKey:
        type: string
        description: OpenAI API key used as a Bearer token.
      chatModel:
        type: string
        description: The chat model id (e.g. gpt-4o-mini).
      prompt:
        type: string
        description: The user prompt to answer.
      speechModel:
        type: string
        description: The text-to-speech model id (e.g. gpt-4o-mini-tts or tts-1).
      voice:
        type: string
        description: The voice to use for synthesis (e.g. alloy).
  steps:
  - stepId: createChat
    description: Create a chat completion answering the user prompt.
    operationId: createChatCompletion
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.chatModel
        messages:
        - role: user
          content: $inputs.prompt
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      reply: $response.body#/choices/0/message/content
  - stepId: speak
    description: Synthesize the chat reply into spoken audio.
    operationId: createSpeech
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.speechModel
        input: $steps.createChat.outputs.reply
        voice: $inputs.voice
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      audio: $response.body
  outputs:
    reply: $steps.createChat.outputs.reply
    audio: $steps.speak.outputs.audio