Hosted TTS first audio chunk p50 (warm)
Hosted speech synthesis missed its 220 ms budget, which was recorded as a failure.
Single reading
261 ms
Protocol
Test definition
Real Flash v2.5 WebSocket synthesis on a persistent connection, first audio chunk.
Limits
Limitations
- n=9, p95 410.38 ms against a 220 ms budget (recorded as FAILED); cold first call 228.51 ms; provider since replaced.
Log
Recorded values
| Date | Value | Evidence | Samples | Source |
|---|---|---|---|---|
| 1 August 2026 | 261 ms | MEASURED | 9 | evidence/live/provider_pass.json |
Source paths refer to the private JARVIS repository and runtime records. They are listed for audit and are not published.