I wanted to ask if a response delay of around 4 to 5 seconds is considered normal. I understand there will always be some delay, but I’ve been trying for a while to find ways to reduce it, and I was wondering if there’s anything I can do to improve it.
For lip sync, I use another tool, so I don’t need Convai’s built-in lip sync or emotion features. I’ve tried removing those parts (though I’m not entirely sure if I’ve removed everything correctly), but I’m still experiencing a delay of 4 to 5 seconds — sometimes even more.
I understand that for very specific or complex questions, the response time might increase. However, for something as simple as saying “Hello,” I believe the response should be faster.
From what I’ve read in the forums, some voices and models are faster than others, but I haven’t been able to clearly identify which ones perform better. Also, I’m using Spanish voices for the responses, and I’m not sure if that could be affecting the speed.
On average, a response delay of around 3 seconds is considered normal, but this can vary depending on several factors:
What Affects Latency?
Voice Provider: ElevenLabs offers high-quality voices but often results in higher response times.
LLM Choice:
Models like GPT-4o and Claude typically respond faster than LLaMA.
Additional Features:
Elements like Core Description, Knowledge Bank, and Narrative Design add richness to responses but may also increase processing time.
Connection & Platform:
Latency may also be affected by internet connection quality and the platform you’re running on.
To Reduce Latency,
Switch to Azure or GCP voices – These usually offer much faster response times.
Use GPT-4o or Claude – These LLMs generally provide quicker replies.
We’re actively working to improve latency, and your feedback is valuable.