Text to Speech

The audio_tts function converts text to audio using an audio model provider.

• Function name: extensions.com.cloud-apim.llm-extension.audio_tts

Parameters

Parameter	Type	Required	Description
provider	string	yes	The audio model provider id
encode_base64	boolean	no	Return the audio as base64 encoded string (default: false)
file_out	string	no	The file path to save the generated audio (default: a temporary file)
payload	object	yes	The payload object
payload.input	string	yes	The text to convert to speech
payload.model	string	no	The model name
payload.voice	string	no	The voice name
payload.instructions	string	no	Instructions for the generation
payload.responseFormat	string	no	The audio response format
payload.speed	number	no	The speech speed

Output

When encode_base64 is true:

{
  "content_type": "audio/mp3",
  "base64": "..."
}

When encode_base64 is false (default):

{
  "content_type": "audio/mp3",
  "file_out": "/tmp/audio-out-xxxxx.mp3"
}

Example

{
  "kind": "call",
  "function": "extensions.com.cloud-apim.llm-extension.audio_tts",
  "args": {
    "provider": "audio-model_xxxxx",
    "encode_base64": true,
    "payload": {
      "input": "Hello, world!",
      "model": "tts-1",
      "voice": "alloy",
      "speed": 1.0
    }
  },
  "result": "audio_result"
}