Resolved -
January 8th, 19:30 UTC - January 9th, 03:00 UTC
Impact: Users with a large number of voices experienced degraded performance with voice endpoints and increased Text-to-Speech (TTS) latency.
Root Cause: A scaling limitation in the voice loading system which prevented voices from being loaded into the cache.
Resolution: We have implemented performance improvements to resolve the immediate issue. Our engineering team is actively developing a architectural solution to prevent future occurrences.
Jan 9, 14:26 UTC