Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-19 14:13:35 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #3598] [BUG]: Cannot enable speak-to-text (STT) function. #2325

New Issue
Closed
opened 2026-02-22 18:29:10 -05:00 by yindo · 3 comments
yindo commented 2026-02-22 18:29:10 -05:00
Owner
Copy Link
Copy Source

Originally created by @zhenyinzhang on GitHub (Apr 5, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3598

How are you running AnythingLLM?

AnythingLLM desktop app

What happened?

Click on the STT icon, encountered with error message: Could not locate file: "http://localhost:3001/static/stt/models/Xenova/whisper-tiny/tokenizer.json"

Platform: Windows 11
Version of AnythingLLM desktop app: 1.7.5

Problem solved manually on my own. I'm posting my steps for your information.

Are there known steps to reproduce?

1, AnythingLLM does not support proxy?
It was mentioned in issue #2651 that connectivity issues towards hf.co/** can possibly result in this error message.
In my country, such connectivity issue exists, but it's resolvable by a VPN proxy. However, I cannot find any proxy settings in AnythingLLM. My proxy software does not record any log message showing that anything_llm requests a download through proxy, when system proxy is enabled. So, it seems that anything_llm does not support proxy (yet) and always tries to visit internet directly.
Therefore, manually downloading models from huggingFace seems to be the only choice (as far as I know).

2, Trying to fix whisper-tiny for STT manually.
issue #2983 provides a clue that on linux there is a path $HOME/.config/anythingllm-desktop/storage possibly correspond to http://localhost:3001/static/stt.
In my case, AnythingLLM was installed on Windows 11 (for all users). I find:
$HOME\AppData\Roaming\anythingllm-desktop\storage. ($HOME==C:\Users\<username>)
According to the fact that tokenizer.json is usually at the root folder of a HF repo, I infer that the hugging face repo "whisper-tiny" should be downloaded (by git LFS) and placed at $HOME\AppData\Roaming\anythingllm-desktop\storage\models\Xenova\whisper-tiny.
After running the following code (successfully):

cd $HOME\AppData\Roaming\anythingllm-desktop\storage\models\Xenova
rm -r whisper-tiny
git lfs install
git clone https://huggingface.co/Xenova/whisper-tiny

And problem solved!

P.S. Make sure you pull from the repo of Xenova, not OpenAI (https://huggingface.co/openai/whisper-tiny). If you do so, the "could not locate file" error won't show up again, but a new error will come: "Unsupported model type: whisper".

  1. More feedback advices for AnythingLLM Developers:
  • Please consider adding proxy support (e.g. add a switch whether enable system proxy or not). (Or make the entrance more obvious, if AnythingLLM has already supported it?)
  • Please consider updating the documentation, and provide an official approach to fix such circumstance, if more users report similar issues.
  • Please add a feature that let user choose whether to send the STT result immediately or not. (Maybe I should make a new issue for it?)
Originally created by @zhenyinzhang on GitHub (Apr 5, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3598 ### How are you running AnythingLLM? AnythingLLM desktop app ### What happened? Click on the STT icon, encountered with error message: `Could not locate file: "http://localhost:3001/static/stt/models/Xenova/whisper-tiny/tokenizer.json"` Platform: Windows 11 Version of AnythingLLM desktop app: 1.7.5 Problem solved manually on my own. I'm posting my steps for your information. ### Are there known steps to reproduce? 1, AnythingLLM does not support proxy? It was mentioned in [issue #2651](https://github.com/Mintplex-Labs/anything-llm/issues/2651) that connectivity issues towards hf.co/** can possibly result in this error message. In my country, such connectivity issue exists, but it's resolvable by a VPN proxy. However, I cannot find any proxy settings in AnythingLLM. My proxy software does not record any log message showing that anything_llm requests a download through proxy, when system proxy is enabled. So, it seems that anything_llm does not support proxy (yet) and always tries to visit internet directly. Therefore, manually downloading models from huggingFace seems to be the only choice (as far as I know). 2, Trying to fix whisper-tiny for STT manually. [issue #2983](https://github.com/Mintplex-Labs/anything-llm/issues/2983) provides a clue that on linux there is a path `$HOME/.config/anythingllm-desktop/storage` possibly correspond to `http://localhost:3001/static/stt`. In my case, AnythingLLM was installed on Windows 11 (for all users). I find: `$HOME\AppData\Roaming\anythingllm-desktop\storage`. (`$HOME==C:\Users\<username>`) According to the fact that `tokenizer.json` is usually at the root folder of a HF repo, I infer that the hugging face repo "whisper-tiny" should be downloaded (by git LFS) and placed at `$HOME\AppData\Roaming\anythingllm-desktop\storage\models\Xenova\whisper-tiny`. After running the following code (successfully): ```powershell cd $HOME\AppData\Roaming\anythingllm-desktop\storage\models\Xenova rm -r whisper-tiny git lfs install git clone https://huggingface.co/Xenova/whisper-tiny ``` And problem solved! P.S. Make sure you pull from the repo of Xenova, not OpenAI (`https://huggingface.co/openai/whisper-tiny`). If you do so, the "could not locate file" error won't show up again, but a new error will come: "Unsupported model type: whisper". 3. More feedback advices for AnythingLLM Developers: - Please consider adding proxy support (e.g. add a switch whether enable system proxy or not). (Or make the entrance more obvious, if AnythingLLM has already supported it?) - Please consider updating the documentation, and provide an official approach to fix such circumstance, if more users report similar issues. - Please add a feature that let user choose whether to send the STT result immediately or not. (Maybe I should make a new issue for it?)
yindo added the possible bug label 2026-02-22 18:29:10 -05:00
yindo closed this issue 2026-02-22 18:29:10 -05:00
yindo commented 2026-02-22 18:29:11 -05:00
Author
Owner
Copy Link
Copy Source

@Louisredstone commented on GitHub (Apr 5, 2025):

I'm going to close this issue since problem has been solved.

@Louisredstone commented on GitHub (Apr 5, 2025): I'm going to close this issue since problem has been solved.
yindo changed title from [BUG]: Cannot enable speak-to-text (STT) function. to [GH-ISSUE #3598] [BUG]: Cannot enable speak-to-text (STT) function. 2026-06-05 14:45:54 -04:00
yindo commented 2026-06-05 14:45:55 -04:00
Author
Owner
Copy Link
Copy Source

@zhenyinzhang commented on GitHub (Apr 5, 2025):

I'm going to close this issue since problem has been solved.

<!-- gh-comment-id:2780648918 --> @zhenyinzhang commented on GitHub (Apr 5, 2025): I'm going to close this issue since problem has been solved.
yindo commented 2026-06-05 14:45:55 -04:00
Author
Owner
Copy Link
Copy Source

@ns167 commented on GitHub (Apr 13, 2026):

how you solved this problem?

<!-- gh-comment-id:4235404536 --> @ns167 commented on GitHub (Apr 13, 2026): how you solved this problem?
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label possible bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2325
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?