diff options
| author | historia <historiavg@proton.me> | 2026-08-24 14:13:06 -0400 |
|---|---|---|
| committer | historia <historiavg@proton.me> | 2026-08-24 14:13:06 -0400 |
| commit | e7a3d65f68659d17f37b79e8bfefea19d7ac0648 (patch) | |
| tree | 2c6d65391e2160ffb800eb237d5e25f842771a3e /app/docs/backend-audiocpp.md | |
| parent | dff790664389d60d16729092a58d9c0dc490a953 (diff) | |
| download | tts-audiobook-generator-e7a3d65f68659d17f37b79e8bfefea19d7ac0648.tar.gz | |
feat: audio.cpp unloads model before converting
Diffstat (limited to 'app/docs/backend-audiocpp.md')
| -rw-r--r-- | app/docs/backend-audiocpp.md | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/app/docs/backend-audiocpp.md b/app/docs/backend-audiocpp.md index c193e92..deb8fbe 100644 --- a/app/docs/backend-audiocpp.md +++ b/app/docs/backend-audiocpp.md @@ -91,3 +91,5 @@ python audiobook.py --backend audiocpp --model qwen-design \ ``` The hub also works with an audio.cpp server that runs somewhere else (another checkout, another machine) as long as it answers on the configured port: when there is no local `server.json`, the convert menus query the running server directly (`GET /v1/models` and `GET /v1/audio/voices`) instead of reading one. On the CLI, pass `--model`/`--voice` matching that server's config. + +Before converting, `audiobook.py` asks the server to unload all currently loaded models (`POST /v1/tasks/unload_all_models`) so models left resident by earlier runs free their memory (e.g. VRAM on GPU backends) and only the selected entry loads. A server without that endpoint, or one busy unloading, only produces a warning. |