9b9206f50d
This commit restructures the audiobook generation and serving layer to rely exclusively on S3-compatible blob storage, removing the dependency on local filesystem paths. - Bundle `ffmpeg` and `ffprobe` binaries via npm to ensure portability across environments. - Update API routes to stream audio directly from blob storage instead of local disk. - Remove legacy migration endpoints and filesystem-based indexing logic. - Add startup scripts to facilitate the transition from local to remote storage. BREAKING CHANGE: Audiobook functionality is now strictly dependent on S3 configuration. The previous filesystem-based storage method has been removed.
19 lines
645 B
Markdown
19 lines
645 B
Markdown
---
|
|
title: Deepinfra
|
|
---
|
|
|
|
Use Deepinfra as a hosted OpenAI-compatible TTS provider.
|
|
|
|
## OpenReader setup
|
|
|
|
1. In OpenReader settings, choose provider `Deepinfra`.
|
|
2. Leave `API_BASE` unset unless you want to override the default endpoint.
|
|
- Default value: `https://api.deepinfra.com/v1/openai`
|
|
3. Set `API_KEY` to your Deepinfra API key if needed.
|
|
4. Choose model and voice.
|
|
|
|
## Notes
|
|
|
|
- `Deepinfra` is a built-in provider in the dropdown, so `API_BASE` is usually not required.
|
|
- Deepinfra supports multiple TTS models, including Kokoro-family options.
|
|
- For variable details, see [Environment Variables](../reference/environment-variables).
|