Based on the new GPT real-time model, is there a w...
# ask-aib
bulky-actor-93512
08/29/2025, 3:08 AMBased on the new GPT real-time model, is there a way I can configure an STT model and use the real-time with a text input and audio output? \
t
tall-airline-16756
08/29/2025, 3:08 AMThanks for reaching out, Sebastian. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
tall-airline-16756
08/29/2025, 3:09 AMLiveKit makes it super easy to set up an STT model that works seamlessly with real-time GPT for both text input and audio output. Here’s a quick rundown on how you can get this going:
• Built-in STT: You can use OpenAI’s Realtime API, which handles the STT for audio input right out of the box. Just pick your transcription model (like "whisper-1") and tweak settings like language and turn detection to fit your needs.
• Text Input & Audio Output: Simply send text input straight to GPT-4o through the Realtime API, and it’ll respond with audio using built-in TTS. Just set
modalities=['text', 'audio']b
bulky-actor-93512
08/29/2025, 3:09 AMi mean if i add a
stt=deepgram.STT
llm=openai.realtime.RealtimeModel()
Will it work using the deepgram stt?
t
tall-airline-16756
08/29/2025, 3:10 AMAbsolutely, you can integrate Deepgram STT with OpenAI's RealtimeModel for real-time processing. Here’s a simple way to do it:
• STT Configuration: Set Deepgram as your transcription provider like this:
Copy code
stt=deepgram.STT(
model="nova-3",
language="en-US",
api_key="YOUR_DEEPGRAM_API_KEY"
) Copy code
llm=openai.realtime.RealtimeModel(
input_audio_transcription=None,
turn_detection=None,
api_key="YOUR_OPENAI_API_KEY"
) Copy code
pip install "livekit-agents[deepgram]" "livekit-agents[openai]"5 Views