Thinline-Dynamic-Solutions/ThinLineRadio v7.0.0-beta9.7.3

v7.0.0 Beta 9.7.3 — Fix stale AssemblyAI speech model names (hotfix)

on GitHub

Hotfix — Beta 9.7.3

Bug Fixes

• AssemblyAI: Gracefully handle stale/unrecognized speech model names
  • If a user had an old model value (e.g. best, nano, slam-1-5) saved in their config from before the AssemblyAI API change, the server would send it as ["best"] which the API rejects with HTTP 400
  • Added a whitelist check: only universal-3-pro and universal-2 are sent; any other stored value is silently ignored and AssemblyAI's own default is used instead

⚠️ Action Required for Affected Users

Go to Admin → Options → Transcription → AssemblyAI Speech Model and set it to universal-2 or universal-3-pro, or leave it blank to use the API default.

See CHANGELOG.md for the full history.