[PR #121] [MERGED] Refactoring of Coqui TTS extension #203

New issue

Closed

opened 2026-10-15 18:05:32 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-15 18:05:32 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/SillyTavern/SillyTavern-Extras/pull/121
Author: @Tony-sama
Created: 8/14/2023
Status: ✅ Merged
Merged: 8/15/2023
Merged by: @Cohee1207

Base: neo ← Head: neo

📝 Commits (8)

1389d38 Started refactoring of Coqui-tts extension.
db63c58 Added missing route to populate ST coqui UI settings and download/check models of coqui api.
0c87844 Merge branch 'SillyTavern:neo' into neo
f87ac34 Merge branch 'SillyTavern:neo' into neo
a440177 reduced coqui api call to minimum, added model download
46fecb5 Merge branch 'neo' of https://github.com/Tony-sama/SillyTavern-extras into neo
166b77b Add download detection when initiated through tts request
c800eae Cleanup remaining old coqui implementation

📊 Changes

3 files changed (+228 additions, -473 deletions)

View changed files

➕ modules/text_to_speech/coqui/coqui_module.py (+215 -0)
📝 server.py (+13 -62)
➖ tts_coqui.py (+0 -411)

📄 Description

Simplified coqui extension by reducing to the following functionalities:

check if a model installed
- just look for files
- check if install is corrupt (fail download)
install a model
- using coqui api automatic download, can be used to reinstall a model
generate tts
- much simpler than before
- assume request payload provide proper model id and options

Notes:

Only tested on gpu, but should work on cpu too.
Can be optimized:
- by keeping model loaded (maybe already managed by coqui api)
- by keeping multiple model loaded for group chat (not sure the api allows it)

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/SillyTavern/SillyTavern-Extras/pull/121 **Author:** [@Tony-sama](https://github.com/Tony-sama) **Created:** 8/14/2023 **Status:** ✅ Merged **Merged:** 8/15/2023 **Merged by:** [@Cohee1207](https://github.com/Cohee1207) **Base:** `neo` ← **Head:** `neo` --- ### 📝 Commits (8) - [`1389d38`](https://github.com/SillyTavern/SillyTavern-Extras/commit/1389d3806d6f8b5090c7fbb67108fefccb8f5032) Started refactoring of Coqui-tts extension. - [`db63c58`](https://github.com/SillyTavern/SillyTavern-Extras/commit/db63c58c30ff634799d2e35873dcb7a0a4077863) Added missing route to populate ST coqui UI settings and download/check models of coqui api. - [`0c87844`](https://github.com/SillyTavern/SillyTavern-Extras/commit/0c87844b8a013b477a19e874024e000c98909c34) Merge branch 'SillyTavern:neo' into neo - [`f87ac34`](https://github.com/SillyTavern/SillyTavern-Extras/commit/f87ac349283b5d9ca5ab3dd17b7a6850541683d5) Merge branch 'SillyTavern:neo' into neo - [`a440177`](https://github.com/SillyTavern/SillyTavern-Extras/commit/a440177f257498502b216ea82df05b38aaad057a) reduced coqui api call to minimum, added model download - [`46fecb5`](https://github.com/SillyTavern/SillyTavern-Extras/commit/46fecb5b488740180bd2bb76f370f62019cfb3dc) Merge branch 'neo' of https://github.com/Tony-sama/SillyTavern-extras into neo - [`166b77b`](https://github.com/SillyTavern/SillyTavern-Extras/commit/166b77b7fd8046a0fe51bd8cea88f1db09b75427) Add download detection when initiated through tts request - [`c800eae`](https://github.com/SillyTavern/SillyTavern-Extras/commit/c800eae3e88f6b36cd9f76dad4e1a622c90eb183) Cleanup remaining old coqui implementation ### 📊 Changes **3 files changed** (+228 additions, -473 deletions) <details> <summary>View changed files</summary> ➕ `modules/text_to_speech/coqui/coqui_module.py` (+215 -0) 📝 `server.py` (+13 -62) ➖ `tts_coqui.py` (+0 -411) </details> ### 📄 Description Simplified coqui extension by reducing to the following functionalities: - check if a model installed - just look for files - check if install is corrupt (fail download) - install a model - using coqui api automatic download, can be used to reinstall a model - generate tts - much simpler than before - assume request payload provide proper model id and options Notes: - Only tested on gpu, but should work on cpu too. - Can be optimized: - by keeping model loaded (maybe already managed by coqui api) - by keeping multiple model loaded for group chat (not sure the api allows it) --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-15 18:05:32 +00:00