How do I switch the elevenlabs models to streaming and Turbo v2.5 to reduce latency?

Original Discord Post by alexrw | 2024-09-18 18:53:30

I’m seeing no option to change the elevenlabs models to improve speed/latency, in either the unity scripts or the character dashboard

Reply by k3kalinix | 2024-09-18 21:03:20

Hello <@860384795295875082>,

Unfortunately, currently it is not possible, but I can add it to the feature request list.

Reply by alexrw | 2024-09-19 00:16:19

That would be amazing!

Having that option would not just save money, but cut latency in half in many cases.

I did also see the difference between two convai videos convai put out, about a year apart, for getting convai going in unity.

Was there a reason why the earlier version seems faster, in terms of latency, than the newest versions?

Old version: https://youtu.be/Vhr7IvfITgU?si=-WLD8hs3rPOVwGty

New version:

Embedded Content:
(OLD) Create AI Characters in Unity | Convai Plugin Setup Tutorial …
In this tutorial, we will show you how to import the Convai plugin into your Unity project and have a conversation with an NPC. We will guide you through the steps of setting up the plugin, including entering your API key, importing the package, and opening the test scene. We will also show you how to select your own character for the conversati…
Link: https://www.youtube.com/watch?v=Vhr7IvfITgU


Quick Guide On Adding AI Characters to Your Unity Project | Convai …
In this updated tutorial, we will show you how you can very easily import the Convai SDK into your Unity project to create intelligent NPCs that can communicate with you naturally in real-time, perceive their virtual surroundings, and generate complex actions on your command. Throughout this tutorial, we guide you on setting up the Convai plugin…
Link: https://www.youtube.com/watch?v=anb9ityi0MQ

Reply by tyke_18 | 2024-09-19 09:20:03

I see this thread is about Eleven Labs, may I add a question here please instead of starting a new thread : if I have an Eleven Labs Pro plan and add my own voices to it to use in my Convai account, if I select the cheaper Eleven voice models (v2 turbo, v2.5 turbo) for my voices in the Eleven Labs website config, will these voices then use these models when they are being used through Convai?

Reply by alexrw | 2024-09-20 23:43:04

It really feels like EVERY MODEL, by default should be a turbo model.

Is there no way to adjust the header or call to Elevenlabs inside Unity to override it with a simple “eleven_turbo_v2_5” somewhere?

This is the single easiest way to reduce latency using elevenlabs

This conversation happened on the Convai Discord Server, so this post will be closed.