# I want to use my own LLM to get speech-to-text transcription

**URL:** <https://openvidu.discourse.group/t/i-want-to-use-my-own-llm-to-get-speech-to-text-transcription/5393>\
**Category:** How to implement?\
**Created:** [July 7, 2025, 3:22pm UTC](https://openvidu.discourse.group/t/i-want-to-use-my-own-llm-to-get-speech-to-text-transcription/5393 "2025-07-07T15:22:10Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![x3m4ikc](https://yyz2.discourse-cdn.com/free1/user_avatar/openvidu.discourse.group/x3m4ikc/32/2595_2.png) [@x3m4ikc](https://openvidu.discourse.group/u/x3m4ikc)\
**Post date:** [July 7, 2025, 3:22pm UTC](https://openvidu.discourse.group/t/i-want-to-use-my-own-llm-to-get-speech-to-text-transcription/5393/1 "2025-07-07T15:22:10Z")

</div>

Hello,

I’m using OpenVidu v3.3.0 single node. I want to use my own LLM to get speech-to-text transcription. I’ve already made the API server.

Can you help me how I can modify the project to use my own Speech Processing agent?

---

<div class="post-metadata">

**Author:** ![pabloFuente](https://yyz2.discourse-cdn.com/free1/user_avatar/openvidu.discourse.group/pablofuente/32/2536_2.png) [@pabloFuente](https://openvidu.discourse.group/u/pabloFuente)\
**Post date:** [July 7, 2025, 4:37pm UTC](https://openvidu.discourse.group/t/i-want-to-use-my-own-llm-to-get-speech-to-text-transcription/5393/2 "2025-07-07T16:37:46Z")

</div>

Hello,

It is very easy to integrate your own custom agent in an OpenVidu deployment. The steps are:

1. Implement your agent using LiveKit Agents framework.
2. Dockerized your agent.
3. Add your agent to your OpenVidu deployment simply by creating a YAML file and restarting your deployment.

These steps are explained in detail here: [Custom agents - openVidu](https://openvidu.io/latest/docs/ai/custom-agents)

---

<div class="post-metadata">

**Author:** ![x3m4ikc](https://yyz2.discourse-cdn.com/free1/user_avatar/openvidu.discourse.group/x3m4ikc/32/2595_2.png) [@x3m4ikc](https://openvidu.discourse.group/u/x3m4ikc)\
**Post date:** [July 8, 2025, 7:39am UTC](https://openvidu.discourse.group/t/i-want-to-use-my-own-llm-to-get-speech-to-text-transcription/5393/3 "2025-07-08T07:39:52Z")

</div>

Thank you a lot! I didn’t notice that in documentation