App Version
2.6.8
Platform
How you installed it
App Store
What happened?
When using the Server TTS engine, Conduit sends an explicit voice parameter on every /api/v1/audio/speech request. This overrides the per-model TTS voice configured in OpenWebUI — the server never gets a chance to apply the model's own voice setting.
What should have happened?
When "System Default" is selected in Conduit's TTS settings, no voice parameter should be sent, allowing OpenWebUI to use the voice configured for the active model or whatever is configured as default on server.
How to reproduce
Current behavior: Conduit fetches the server's global default voice and sends it explicitly, overriding model-specific voices.
Also worth noting: users with custom TTS backends (non-standard voice names) have no way to manually enter a voice ID in the app.
Does this happen every time?
Yes, every time
When did this start?
No response
Screenshots
No response
Logs
No response
App Version
2.6.8
Platform
How you installed it
App Store
What happened?
When using the Server TTS engine, Conduit sends an explicit voice parameter on every /api/v1/audio/speech request. This overrides the per-model TTS voice configured in OpenWebUI — the server never gets a chance to apply the model's own voice setting.
What should have happened?
When "System Default" is selected in Conduit's TTS settings, no voice parameter should be sent, allowing OpenWebUI to use the voice configured for the active model or whatever is configured as default on server.
How to reproduce
Current behavior: Conduit fetches the server's global default voice and sends it explicitly, overriding model-specific voices.
Also worth noting: users with custom TTS backends (non-standard voice names) have no way to manually enter a voice ID in the app.
Does this happen every time?
Yes, every time
When did this start?
No response
Screenshots
No response
Logs
No response