To test

session.say

effectively and ensure it works as expected during tests, follow these steps: • Mock Components: Set up your test environment with mocked STT, LLM, and TTS components to isolate the behavior of

session.say

. • Invoke `session.say`: Call

session.say()

with your test message in your test method. • Track Playback Events: Use event listeners to capture

PlaybackFinishedEvent

and verify that the message was played. • Verify Chat Context: Ensure the message is added to the chat context if

add_to_chat_ctx=True

. If

session.run

seems to ignore

session.say

, consider the following: • Asynchronous Execution: Ensure you

await

session.say()

to allow it to complete before

session.run

proceeds. • TTS Configuration: Verify that TTS is correctly configured in the session, as

session.say

won’t play messages without it. • Event Handling: Set up event listeners to capture speech events indicating playback completion. • SpeechHandle Management: Use

SpeechHandle

to wait for speech completion, ensuring

session.run

doesn’t skip it. Here’s an example test code snippet:

async def test_session_say():
    speed = 5.0
    actions = FakeActions()
    actions.add_llm("Hello, test!")
    actions.add_tts(2.0)

    session = create_session(actions, speed_factor=speed)
    agent = MyAgent()

    playback_finished_events = []
    session.output.audio.on("playback_finished", playback_finished_events.append)

    await session.say("Test message")

    assert len(playback_finished_events) == 1
    assert not playback_finished_events[0].interrupted

By following these steps, you can ensure

session.say

is tested properly and works seamlessly with

session.run

.