Can the Microsoft Speech SDK be configured to use plain HTTP WebSocket (ws) connections, allowing a proxy to upgrade to wss? #2680
Labels
update needed
For items that are in progress but have not been updated
Description:
We are using the Microsoft Speech SDK to connect to the Microsoft Speech Service via WebSocket calls. Our architecture relies on an Envoy outbound proxy configured for TLS termination.
Key setup details:
The instance is restricted to HTTP traffic only, with iptables redirecting this traffic to Envoy for TLS upgrades (HTTP → HTTPS, ws → wss).
CONNECT requests are not allowed through the proxy, as TLS termination is exclusively handled by Envoy.
This setup works for an internal service where WebSocket connections (ws) are successfully upgraded to wss by Envoy.
However, when using the Microsoft Speech SDK, the WebSocket upgrade (ws → wss) via Envoy does not work. It seems the SDK attempts to initiate wss connections directly, bypassing the proxy's TLS termination process.
Question:
Is it possible to configure the Microsoft Speech SDK to initiate plain HTTP WebSocket (ws) connections, enabling our proxy to upgrade these to wss?
Is there anything I am missing ?
The text was updated successfully, but these errors were encountered: