search

Mintplex-Labs / anything-llm

The all-in-one Desktop & Docker AI application with built-in RAG, AI agents, and more.
https://anythingllm.com
MIT License
27.45k stars 2.76k forks source link

[BUG]: Can‘t use speech-to-text #2651

Closed Kurasa632 closed 2 days ago

Kurasa632 commented 2 days ago

How are you running AnythingLLM?

Docker (local)

What happened?

When I use STT for input.The error is shown in the figure below. 企业微信截图_17320057285826

Are there known steps to reproduce?

No response

timothycarambat commented 2 days ago

You are either offline or cannot download the model from HuggingFace (which is where it comes from). When you use STT for the first time it streams and stores the ONNX of whisper-tiny so you can use it. It is not pre-loaded into the application since not everyone uses it.

If you have internet connectivity issues or are prevented from reaching hf.co/** endpoints (by ISP or VPN) will result in this issue. We do not mirror this model to a fallback URL, so its HF or nothing

Kurasa632 commented 2 days ago

You are either offline or cannot download the model from HuggingFace (which is where it comes from). When you use STT for the first time it streams and stores the ONNX of whisper-tiny so you can use it. It is not pre-loaded into the application since not everyone uses it.

If you have internet connectivity issues or are prevented from reaching hf.co/** endpoints (by ISP or VPN) will result in this issue. We do not mirror this model to a fallback URL, so its HF or nothing

If I manually download whisper-tiny, where should I put it?