I just wonder whether I can use livekit for only reading a text

Thanks for reaching out, Baymurat. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

can you help with the flow?

Hey there! Let me walk you through how to use LiveKit for text-only reading – it's actually pretty straightforward! Basic Setup Flow Want to create a text-only agent? No worries! You'll just configure the AgentSession without any audio bells and whistles. Here's how you can set it up:

async def entrypoint(ctx: JobContext):
    session = AgentSession(
        llm=openai.LLM(model="gpt-4o-mini"),
        # note that no TTS or STT are needed here
    )
    await session.start(
        agent=MyAgent(),
        room=ctx.room,
        room_input_options=RoomInputOptions(text_enabled=True, audio_enabled=False),
        room_output_options=RoomOutputOptions(transcription_enabled=True, audio_enabled=False),
    )

Text Input and Output Flow • Text Input: You'll receive text input using TextStream on the

lk.chat

topic. • Text Output: Your agent will send responses through TextStream on the

lk.transcription

topic. Think of it like your agent is always listening for messages on the

lk.chat

text stream and ready to chat back! Configuration Options • Want to go full text-only? Just set

audio_enabled=False

in

RoomOutputOptions

. Your agent will happily send text responses to the

lk.transcription

text stream topic. • If you want to disable text input, you can set

text_enabled=False

in

RoomInputOptions

. Alternative Input Methods Got other ways to send text? No problem! You've got options: • Send text input via:

generate_reply(user_input="user's input text")

• Catch the agent's response with

session.on("conversation_item_added", ev)

Event Handling Want to stay in the loop? You can get notified whenever a text message comes in or goes out by listening to the

conversation_item_added

event. This approach is super efficient – perfect for creating smooth, text-based AI conversations without any audio complexity. Pretty cool, right? Sources: examples/other/chat-stream-receiver.py | Text and transcriptions | LiveKit Docs

if i diable TTS, how it reads the text?

Hey there! Let me break down how LiveKit handles text processing when Text-to-Speech is turned off. It's actually pretty cool and straightforward. • Text Processing Without TTS: Think of it like a direct pipeline - the text from your language model goes straight through the transcription system without needing to be converted to audio. • Transcription Node Processing: This is where the magic happens! The transcription node takes your text and can do some neat customization. By default, it'll pass text through as-is, but you can totally tweak how it handles things if you want. • Text Stream Output: Here's the fun part - your processed text gets sent to the frontend in real-time. It uses LiveKit's text streams, and gets published to the

lk.transcription

topic, complete with sender details. • Configuration Control: Want to turn off text output completely? No problem! Just set

transcription_enabled=False

in

RoomOutputOptions

. Basically, LiveKit treats your text like a live, direct output stream - super flexible and powerful, without the need for audio synthesis. Pretty neat, right? Sources: livekit-agents/livekit/agents/voice/agent_activity.py | Text and transcriptions | LiveKit Docs