LLM Streaming¶
Contact Center supervisors can enable real-time streaming of LLM responses, reducing latency and enhancing the user experience, particularly for voice interactions.
Steps to enable LLM Streaming:
-
Go to Contact Center AI > CONFIGURATIONS > Advanced Settings > LLM Streaming.
-
Turn on the toggle to enable LLM streaming.
Note
LLM Streaming is disabled by default.
Benefits of LLM Streaming¶
- Reduced Latency: Real-time streaming of rephrased responses significantly decreases the time taken for the user to receive the modified content.
- Improved User Experience: Faster delivery of rephrased content creates a more natural and interactive conversation flow.
- Simplified Process: Streamlining the rephrasing process reduces complexity and potential points of failure.
LLM Streaming for Text-to-Speech (TTS) Providers¶
LLM streaming is enabled for the following TTS providers:
- PlayHT
- ElevenLabs
- Deepgram
Note
If a GenAI node's response contains special characters, such as punctuation marks, symbols, or brackets, the system might fail to play the prompt during a call, resulting in silence for the caller.