{
  "kind": "open-flow-workflow",
  "version": 1,
  "name": "OpenAI Text to Speech",
  "content": {
    "modelVersion": 14,
    "document": {
      "bindings": {},
      "graph": {
        "nodes": {
          "n_7567d8226b9ae371": {
            "name": "Speech text and voice",
            "kind": "start",
            "inputDefs": [
              {
                "jsonSchema": {
                  "type": "string",
                  "minLength": 1,
                  "maxLength": 4096
                },
                "nullable": false,
                "handle": "input_text"
              },
              {
                "jsonSchema": {
                  "type": "string"
                },
                "nullable": true,
                "handle": "voice"
              }
            ]
          },
          "n_984162e9021686b2": {
            "inputs": {
              "input_text": {
                "kind": "sources",
                "sources": [
                  {
                    "kind": "node",
                    "nodeId": "n_7567d8226b9ae371",
                    "output": "input_text"
                  }
                ]
              },
              "voice": {
                "kind": "sources",
                "sources": [
                  {
                    "kind": "node",
                    "nodeId": "n_7567d8226b9ae371",
                    "output": "voice"
                  }
                ]
              }
            },
            "name": "Validate speech input",
            "kind": "task",
            "task": {
              "inputs": [
                {
                  "jsonSchema": {
                    "type": "string"
                  },
                  "nullable": false,
                  "handle": "input_text"
                },
                {
                  "jsonSchema": {
                    "type": "string"
                  },
                  "nullable": true,
                  "handle": "voice"
                }
              ],
              "outputs": [
                {
                  "jsonSchema": {
                    "type": "string"
                  },
                  "nullable": false,
                  "handle": "input_text"
                },
                {
                  "jsonSchema": {
                    "type": "string"
                  },
                  "nullable": false,
                  "handle": "voice"
                }
              ],
              "name": "Validate speech input",
              "moduleId": "n_1dfcc0a72c29193a",
              "capabilities": [
                {
                  "kind": "connector",
                  "mode": "independent",
                  "actions": []
                }
              ]
            }
          },
          "n_a9b1c51c6e87c00a": {
            "inputs": {
              "model": {
                "kind": "value",
                "value": "tts-1"
              },
              "input": {
                "kind": "sources",
                "sources": [
                  {
                    "kind": "node",
                    "nodeId": "n_984162e9021686b2",
                    "output": "input_text"
                  }
                ]
              },
              "voice": {
                "kind": "sources",
                "sources": [
                  {
                    "kind": "node",
                    "nodeId": "n_984162e9021686b2",
                    "output": "voice"
                  }
                ]
              },
              "response_format": {
                "kind": "value",
                "value": "mp3"
              },
              "stream_format": {
                "kind": "value",
                "value": "audio"
              }
            },
            "name": "Generate MP3 speech",
            "kind": "task",
            "task": {
              "inputs": [
                {
                  "description": "The text-to-speech model to use.",
                  "jsonSchema": {
                    "type": "string",
                    "description": "The text-to-speech model to use."
                  },
                  "nullable": false,
                  "handle": "model"
                },
                {
                  "description": "The text to synthesize into speech.",
                  "jsonSchema": {
                    "type": "string",
                    "description": "The text to synthesize into speech."
                  },
                  "nullable": false,
                  "handle": "input"
                },
                {
                  "description": "The voice configuration used for synthesis.",
                  "jsonSchema": {
                    "anyOf": [
                      {
                        "type": "string",
                        "description": "A built-in voice name."
                      },
                      {
                        "type": "object",
                        "additionalProperties": {
                          "description": "Any JSON value accepted by the upstream API."
                        },
                        "description": "A structured voice object accepted by the API."
                      }
                    ],
                    "description": "The voice configuration used for synthesis."
                  },
                  "nullable": false,
                  "handle": "voice"
                },
                {
                  "description": "Optional voice instructions that guide the synthesis style.",
                  "jsonSchema": {
                    "type": "string",
                    "description": "Optional voice instructions that guide the synthesis style."
                  },
                  "nullable": true,
                  "handle": "instructions",
                  "value": null
                },
                {
                  "description": "The audio format to return.",
                  "jsonSchema": {
                    "type": "string",
                    "enum": [
                      "mp3",
                      "opus",
                      "aac",
                      "flac",
                      "wav",
                      "pcm"
                    ],
                    "description": "The audio format to return."
                  },
                  "nullable": true,
                  "handle": "response_format",
                  "value": null
                },
                {
                  "description": "The playback speed multiplier.",
                  "jsonSchema": {
                    "type": "number",
                    "minimum": 0.25,
                    "maximum": 4,
                    "description": "The playback speed multiplier."
                  },
                  "nullable": true,
                  "handle": "speed",
                  "value": null
                },
                {
                  "description": "The speech response delivery format. This connector only accepts audio or an omitted value.",
                  "jsonSchema": {
                    "type": "string",
                    "enum": [
                      "audio",
                      "sse"
                    ],
                    "description": "The speech response delivery format. This connector only accepts audio or an omitted value."
                  },
                  "nullable": true,
                  "handle": "stream_format",
                  "value": null
                }
              ],
              "outputs": [
                {
                  "description": "The synthesized audio encoded as base64.",
                  "jsonSchema": {
                    "type": "string",
                    "description": "The synthesized audio encoded as base64."
                  },
                  "nullable": false,
                  "handle": "content_base64"
                },
                {
                  "description": "The MIME type of the synthesized audio.",
                  "jsonSchema": {
                    "type": "string",
                    "description": "The MIME type of the synthesized audio."
                  },
                  "nullable": false,
                  "handle": "content_type"
                }
              ],
              "name": "Generate MP3 speech",
              "executor": {
                "kind": "connector",
                "action": "openai.create_speech"
              }
            }
          },
          "n_93e5f554aa8c8c31": {
            "inputs": {
              "content_base64": {
                "kind": "sources",
                "sources": [
                  {
                    "kind": "node",
                    "nodeId": "n_a9b1c51c6e87c00a",
                    "output": "content_base64"
                  }
                ]
              },
              "content_type": {
                "kind": "sources",
                "sources": [
                  {
                    "kind": "node",
                    "nodeId": "n_a9b1c51c6e87c00a",
                    "output": "content_type"
                  }
                ]
              }
            },
            "name": "Save audio artifact",
            "kind": "task",
            "task": {
              "inputs": [
                {
                  "jsonSchema": {
                    "type": "string",
                    "description": "The synthesized audio encoded as base64."
                  },
                  "nullable": false,
                  "handle": "content_base64"
                },
                {
                  "jsonSchema": {
                    "type": "string",
                    "description": "The MIME type of the synthesized audio."
                  },
                  "nullable": false,
                  "handle": "content_type"
                }
              ],
              "outputs": [
                {
                  "jsonSchema": {
                    "type": "object",
                    "properties": {
                      "kind": {
                        "const": "artifact"
                      },
                      "id": {
                        "type": "string"
                      },
                      "name": {
                        "type": "string"
                      },
                      "mediaType": {
                        "type": "string"
                      },
                      "size": {
                        "type": "integer",
                        "minimum": 1
                      },
                      "digest": {
                        "type": "string"
                      }
                    },
                    "required": [
                      "kind",
                      "id",
                      "name",
                      "size",
                      "digest"
                    ]
                  },
                  "nullable": false,
                  "handle": "audio"
                }
              ],
              "name": "Save audio artifact",
              "moduleId": "n_d4aa114b1fbfda6b",
              "capabilities": [
                {
                  "kind": "connector",
                  "mode": "independent",
                  "actions": []
                }
              ]
            }
          },
          "n_6b061d0175e9795f": {
            "name": "Download speech audio",
            "kind": "outputs",
            "outputDefs": [
              {
                "jsonSchema": {
                  "type": "object",
                  "properties": {
                    "kind": {
                      "const": "artifact"
                    },
                    "id": {
                      "type": "string"
                    },
                    "name": {
                      "type": "string"
                    },
                    "mediaType": {
                      "type": "string"
                    },
                    "size": {
                      "type": "integer",
                      "minimum": 1
                    },
                    "digest": {
                      "type": "string"
                    }
                  },
                  "required": [
                    "kind",
                    "id",
                    "name",
                    "size",
                    "digest"
                  ]
                },
                "nullable": false,
                "handle": "audio"
              }
            ],
            "outputMappings": {
              "audio": {
                "kind": "sources",
                "sources": [
                  {
                    "kind": "node",
                    "nodeId": "n_93e5f554aa8c8c31",
                    "output": "audio"
                  }
                ]
              }
            }
          }
        },
        "edges": [
          {
            "source": "n_7567d8226b9ae371",
            "target": "n_984162e9021686b2"
          },
          {
            "source": "n_984162e9021686b2",
            "target": "n_a9b1c51c6e87c00a"
          },
          {
            "source": "n_a9b1c51c6e87c00a",
            "target": "n_93e5f554aa8c8c31"
          },
          {
            "source": "n_93e5f554aa8c8c31",
            "target": "n_6b061d0175e9795f"
          }
        ]
      }
    },
    "modules": {
      "n_1dfcc0a72c29193a": {
        "name": "Validate speech input",
        "imports": [],
        "source": "export default function prepareSpeech({ input_text, voice }) {\n  if (typeof input_text !== 'string' || !input_text.trim()) throw new Error('Input text is required.');\n  if (Array.from(input_text).length > 4096) throw new Error('Input text must contain at most 4096 characters.');\n  const selected = voice == null || voice === '' ? 'alloy' : voice;\n  if (!['alloy', 'ash', 'coral', 'echo', 'fable', 'onyx', 'nova', 'sage', 'shimmer'].includes(selected)) throw new Error('Choose a supported tts-1 voice.');\n  return { input_text, voice: selected };\n}\n"
      },
      "n_d4aa114b1fbfda6b": {
        "name": "Save audio artifact",
        "imports": [],
        "source": "import { Buffer } from 'node:buffer';\n\nexport default async function saveSpeech({ content_base64, content_type }, context) {\n  if (typeof content_base64 !== 'string' || !content_base64 || content_base64.length % 4 !== 0 || !/^(?:[A-Za-z0-9+/]{4})*(?:[A-Za-z0-9+/]{2}==|[A-Za-z0-9+/]{3}=)?$/.test(content_base64)) throw new Error('The speech response does not contain valid base64 audio.');\n  const mediaType = typeof content_type === 'string' ? content_type.split(';')[0].trim().toLowerCase() : '';\n  if (!['audio/mpeg', 'audio/mp3', 'application/octet-stream'].includes(mediaType)) throw new Error('The speech response is not MP3 audio.');\n  const bytes = new Uint8Array(Buffer.from(content_base64, 'base64'));\n  if (!bytes.length || Buffer.from(bytes).toString('base64') !== content_base64) throw new Error('The speech response is empty or has invalid base64 padding.');\n  return { audio: await context.artifact.put(bytes, { name: 'speech.mp3', mediaType: 'audio/mpeg' }) };\n}\n"
      }
    }
  },
  "presentation": {
    "designer": {
      "version": 1,
      "flow": {
        "direction": "vertical"
      }
    }
  }
}