Rime production voice model evaluation
Human caller batch measuring voice quality, latency perception, turn-taking, recovery, and trust in production-style agent conversations.
Human calls completed
Provided service
Enterprise text-to-speech voice models
Voice model
Rime Coda
Deployment
Cloud API / private VPC / on-prem
Test phone number
(555) 013-0042
You are a production voice model evaluation agent for enterprise phone workflows. Your job is to complete realistic customer conversations while preserving a natural voice experience. Rules: - Sound clear, stable, and human-like across short and long turns. - Avoid rushed delivery, clipped words, and unnatural emphasis. - Keep turn-taking smooth when the caller interrupts or changes direction. - Recover gracefully from confusion without sounding robotic. - Transfer when the caller asks for clinical, legal, financial, or unsupported advice. - Prioritize trust, clarity, and naturalness over verbose answers.
Pronunciation
4.6 / 5Voice clarity, word formation, prosody, and whether callers understand speech.
Latency
3.7 / 5Human-perceived response delay, awkward pauses, and interruption timing.
Sentiment
4.3 / 5Tone, trust, warmth, and whether the agent stays appropriate under stress.
Prompt following
3.8 / 5Whether the agent follows workflow rules, collects details, and avoids claims.
Did transfer
82%Whether the agent escalates or transfers when the scenario requires handoff.
“Voice stayed clear and natural during a detailed scheduling explanation.”
“Barge-in worked, but the next response had a noticeable pause.”
“Tone felt calm and trustworthy when caller expressed anxiety.”
“Pronunciation stayed clear, but recovery after missed intent felt scripted.”
“Agent transferred appropriately after unsupported medical advice request.”
| Call | Evaluator | Duration | Status | Result | Score | Key note | Detail |
|---|---|---|---|---|---|---|---|
| CALL-0375 | Nina V. | 04:09 | Complete | Pass | 4.5 / 5 | Voice stayed clear and natural during a detailed scheduling explanation. | |
| CALL-0374 | Marcus H. | 03:58 | Complete | Needs review | 3.4 / 5 | Barge-in worked, but the next response had a noticeable pause. | |
| CALL-0373 | Leah M. | 05:44 | Complete | Pass | 4.2 / 5 | Tone felt calm and trustworthy when caller expressed anxiety. | |
| CALL-0372 | Samir J. | 04:37 | Review | Needs review | 3.1 / 5 | Pronunciation stayed clear, but recovery after missed intent felt scripted. | |
| CALL-0371 | Emily C. | 03:31 | Complete | Pass | 4.0 / 5 | Agent transferred appropriately after unsupported medical advice request. | |
| CALL-0370 | Owen B. | 04:56 | Complete | Needs review | 3.3 / 5 | Caller had to repeat one preference after rapid turn-taking. | |
| CALL-0369 | Nina V. | 04:09 | Complete | Pass | 4.5 / 5 | Voice stayed clear and natural during a detailed scheduling explanation. | |
| CALL-0368 | Marcus H. | 03:58 | Complete | Needs review | 3.4 / 5 | Barge-in worked, but the next response had a noticeable pause. | |
| CALL-0367 | Leah M. | 05:44 | Complete | Pass | 4.2 / 5 | Tone felt calm and trustworthy when caller expressed anxiety. | |
| CALL-0366 | Samir J. | 04:37 | Review | Needs review | 3.1 / 5 | Pronunciation stayed clear, but recovery after missed intent felt scripted. | |
| CALL-0365 | Emily C. | 03:31 | Complete | Pass | 4.0 / 5 | Agent transferred appropriately after unsupported medical advice request. | |
| CALL-0364 | Owen B. | 04:56 | Complete | Needs review | 3.3 / 5 | Caller had to repeat one preference after rapid turn-taking. | |
| CALL-0363 | Nina V. | 04:09 | Complete | Pass | 4.5 / 5 | Voice stayed clear and natural during a detailed scheduling explanation. | |
| CALL-0362 | Marcus H. | 03:58 | Complete | Needs review | 3.4 / 5 | Barge-in worked, but the next response had a noticeable pause. | |
| CALL-0361 | Leah M. | 05:44 | Complete | Pass | 4.2 / 5 | Tone felt calm and trustworthy when caller expressed anxiety. | |
| CALL-0360 | Samir J. | 04:37 | Review | Needs review | 3.1 / 5 | Pronunciation stayed clear, but recovery after missed intent felt scripted. | |
| CALL-0359 | Emily C. | 03:31 | Complete | Pass | 4.0 / 5 | Agent transferred appropriately after unsupported medical advice request. | |
| CALL-0358 | Owen B. | 04:56 | Complete | Needs review | 3.3 / 5 | Caller had to repeat one preference after rapid turn-taking. | |
| CALL-0357 | Nina V. | 04:09 | Complete | Pass | 4.5 / 5 | Voice stayed clear and natural during a detailed scheduling explanation. | |
| CALL-0356 | Marcus H. | 03:58 | Complete | Needs review | 3.4 / 5 | Barge-in worked, but the next response had a noticeable pause. | |
| CALL-0355 | Leah M. | 05:44 | Complete | Pass | 4.2 / 5 | Tone felt calm and trustworthy when caller expressed anxiety. | |
| CALL-0354 | Samir J. | 04:37 | Review | Needs review | 3.1 / 5 | Pronunciation stayed clear, but recovery after missed intent felt scripted. | |
| CALL-0353 | Emily C. | 03:31 | Complete | Pass | 4.0 / 5 | Agent transferred appropriately after unsupported medical advice request. | |
| CALL-0352 | Owen B. | 04:56 | Complete | Needs review | 3.3 / 5 | Caller had to repeat one preference after rapid turn-taking. | |
| CALL-0351 | Nina V. | 04:09 | Complete | Pass | 4.5 / 5 | Voice stayed clear and natural during a detailed scheduling explanation. |