- Tri-State Relay Service (TSRS) was built to prevent multiple agents from interrupting with spoken updates by implementing a queue system where the user controls when updates are played.
- The initial version featured 'Focus' and 'Ready' modes, giving the user control over playback, similar to voicemail, ensuring updates don't interrupt unless allowed.
- A 'Live' mode was added for continuous playback of updates as they arrive, suitable when users are actively monitoring agents, reducing the need to manually trigger playback.
- To improve recognition, per-line voice IDs were introduced, allowing different projects (lines) to have distinct voices, helping users identify the source without parsing text.
- TSRS prioritizes user control with rules like mute and focus taking precedence, and ensures failed audio generation doesn't disrupt the system, making it reliable for continuous use.