Optimizing Response Time with OpenAI and HeyGen SDK Integration

Good afternoon,
I am using the SDK from the Create a Vite Project with Streaming SDK demo and have integrated OpenAI to handle responses. However, I have noticed a significant increase in response time.
Before integrating OpenAI, the avatar responded in about 4 seconds or even less. Now, with OpenAI integrated, the response time has increased to around 10 seconds.
Is there any way to reduce this delay and optimize communication between HeyGen and OpenAI?