TTS

• Key points are in the visual diagram above.

Reading agent responses on screen isn't always possible or desirable. TTS enables phone agents, accessibility features, and conversational experiences that feel human.

TTS is an audiobook narrator — it reads written text aloud with appropriate pacing, emphasis, and tone.

Scenario

Agent generates answer text → TTS converts to audio stream → played through phone system IVR. Voice: 'professional female, en-US'.

Solution

In Voice & Multimodal Agents, apply TTS to this scenario: Agent generates answer text → TTS converts to audio stream → played through phone system IVR. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to TTS (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where TTS happens in the code.

1from openai import OpenAI  # import dependencies2
3client = OpenAI()  # create API client4
5def speak(text: str, output_path: str = "response.mp3") -> str:  # define a reusable function6    response = client.audio.speech.create(  # call the API7        model="tts-1",8        voice="nova",9        input=text,10    )11    response.stream_to_file(output_path)12    return output_path  # return the result

• Treating TTS as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for tts

Quick recap — the most important points from this module.