[GH-ISSUE #3174] How should PiperTTS in Text-to-speech Preference be used? #2039

Closed
opened 2026-02-22 18:27:52 -05:00 by yindo · 1 comment
Owner

Originally created by @1oskao21 on GitHub (Feb 11, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3174

How should PiperTTS in Text-to-speech Preference be used? When a model is identified but an error is reported, how should it be configured? How to properly deploy piperTTS model

Image
Image
Image
Image

Originally created by @1oskao21 on GitHub (Feb 11, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3174 How should PiperTTS in Text-to-speech Preference be used? When a model is identified but an error is reported, how should it be configured? How to properly deploy piperTTS model ![Image](https://github.com/user-attachments/assets/b40db574-17fd-4a9a-ad01-791d29411496) ![Image](https://github.com/user-attachments/assets/3082d8bb-ce1f-4b1a-94f5-3800e6bec5fd) ![Image](https://github.com/user-attachments/assets/6af450b0-7494-42e8-bc87-25680433146a) ![Image](https://github.com/user-attachments/assets/029d98ee-e417-4e5f-bfcd-f7fae4105b27)
yindo closed this issue 2026-02-22 18:27:52 -05:00
Author
Owner

@timothycarambat commented on GitHub (Feb 11, 2025):

Are based in China mainland? Huggingface almost across the board blocks CN based IPs. While we mirror the WASM to run the models we do not mirror the voice files themselves since it would be a huge load on our CDN.

Inside of the package that we customized to make this work we have this:

const HF_BASE = "https://huggingface.co/diffusionstudio/piper-voices/resolve/main";
const ONNX_BASE = "https://cdnjs.cloudflare.com/ajax/libs/onnxruntime-web/1.18.0/";
const WASM_BASE = "https://cdn.jsdelivr.net/npm/@diffusionstudio/piper-wasm@1.0.0/build/piper_phonemize";

Which is almost certainly what is blocking you here. The files get stored in browserFS not in our normal models directory - which you have shown above. This is done via the package built from this repo

@timothycarambat commented on GitHub (Feb 11, 2025): Are based in China mainland? Huggingface almost across the board blocks CN based IPs. While we mirror the WASM to _run_ the models we do not mirror the voice files themselves since it would be a huge load on our CDN. Inside of the package that we customized to make this work we have this: ``` const HF_BASE = "https://huggingface.co/diffusionstudio/piper-voices/resolve/main"; const ONNX_BASE = "https://cdnjs.cloudflare.com/ajax/libs/onnxruntime-web/1.18.0/"; const WASM_BASE = "https://cdn.jsdelivr.net/npm/@diffusionstudio/piper-wasm@1.0.0/build/piper_phonemize"; ``` Which is almost certainly what is blocking you here. The files get stored in _browserFS_ not in our normal _models_ directory - which you have shown above. This is done via the package [built from this repo](https://github.com/Mintplex-Labs/piper-tts-web)
yindo changed title from How should PiperTTS in Text-to-speech Preference be used? to [GH-ISSUE #3174] How should PiperTTS in Text-to-speech Preference be used? 2026-06-05 14:44:11 -04:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2039