Converts agent responses to natural speech output. Use for: speak response, text to speech, voice output, read aloud.
