Streaming Voice-Agent Interaction Efficiency on Spoken-MQA
1.96NFELTS-VoiceAgent
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LTS-VoiceAgent2026.01 | 1.96 | 0.16 | 8.16 | |
| LTS-VADThinking Mode=No-think2026.01 | 3.42 | 1.82 | 53.21 | |
| LTS-VADThinking Mode=Think2026.01 | 3.42 | 2.39 | 69.88 | |
| PredGenThinking Mode=No-think2026.01 | 59.39 | 58.14 | 97.9 | |
| PredGenThinking Mode=Think2026.01 | 59.39 | 58.39 | 98.31 |