Research map / Large language models / Language, safety and interpretability
SpokenWOZ: A Large-Scale Speech-Text Benchmark for Spoken Task-Oriented Dialogue Agents
Shuzheng Si, Wentao Ma, Haoyu Gao, Yuchuan Wu, Ting-En Lin, Yinpei Dai, Hangyu Li, Rui Yan, Fei Huang, Yongbin Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add aligned WavLM speech embeddings to SPACE | SPACE · SpokenWOZ DST | JGA: 22.73 -> 25.65 | Improved |
| Add WavLM speech embeddings without alignment | SPACE · SpokenWOZ DST | JGA: 22.73 -> 24.09 | Improved |
| Align words with audio pieces (WavLMalign vs WavLM) | SPACE+WavLM · SpokenWOZ DST | JGA: 24.09 -> 25.65 | Improved |
| Add WavLM fusion to SPACE+TripPy | SPACE+TripPy · SpokenWOZ DST | JGA: 16.24 -> 18.71 | Improved |
| Replace BERT encoder by SPACE in TripPy | BERT+TripPy · SpokenWOZ DST | JGA: 14.78 -> 16.24 | Improved |
| Generative (UBAR) instead of extractive (TripPy) DST | BERT+TripPy · SpokenWOZ DST | JGA: 14.78 -> 20.54 | Improved |
| Exclude cross-turn slots from evaluation | SPACE · SpokenWOZ DST | JGA: 22.73 -> 26.99 | Improved |
| Add aligned speech to SPACE in policy optimization | SPACE · SpokenWOZ policy optimization | Comb: 85.52 -> 88.01 | Improved |
| Add aligned speech to SPACE in end-to-end modeling | SPACE · SpokenWOZ end-to-end | Comb: 79.84 -> 82.32 | Improved |
| Align speech vs unaligned WavLM in end-to-end modeling | SPACE+WavLM · SpokenWOZ end-to-end | Comb: 80.71 -> 82.32 | Improved |
| Add WavLM without alignment to SPACE, policy BLEU | SPACE · SpokenWOZ policy optimization | BLEU: 18.72 -> 18.54 | No clear effect |
| Evaluate SPACE on spoken (SpokenWOZ) instead of written MultiWOZ | SPACE · DST | JGA: 57.5 -> 22.73 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- UBAR — UBAR: Towards Fully End-to-End Task-Oriented Dialog System with GPT-2 (AAAI 2021) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Aligning Spoken Dialogue Models from User Interactions (ICML 2025 · similarity 0.88)
- SD-Eval: A Benchmark Dataset for Spoken Dialogue Understanding Beyond Words (NeurIPS 2024 · similarity 0.88)
- MultiTalk: Scaling Full-Duplex Speech Models to Long, Multi-Party, Bilingual Conversation (NeurIPS 2026 · similarity 0.88)
- MMSU: A Massive Multi-task Spoken Language Understanding and Reasoning Benchmark (ICLR 2026 · similarity 0.87)
- Paralinguistics-Aware Speech-Empowered Large Language Models for Natural Conversation (NeurIPS 2024 · similarity 0.87)
- TA&AT: Enhancing Task-Oriented Dialog with Turn-Level Auxiliary Tasks and Action-Tree Based Scheduled Sampling (AAAI 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
