Early-stage speech researcher at Sun Yat-sen University building full-duplex spoken dialogue benchmarks and datasets
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
No reliable output totals are available yet: OpenAlex indexes 6 recent works, but a complete record of publications, venues, and authorship share is not established. This axis would measure total and per-year scholarly output.
Impact—
Citation impact cannot be scored: h-index and i10-index are recorded as 0, and the cataloged works are 2026 arXiv preprints too recent for citations. This axis would measure citation accumulation and scholarly influence.
Focus0/100
The computed focus score is 0, but the raw field distribution splits 0.5/0.5 between two nearly identical spoken-dialogue labels, and topic counts cluster tightly: Speech Recognition and Synthesis (4), Speech and dialogue systems (4), Speech and Audio Processing (3). The zero score appears to reflect label fragmentation rather than genuine topic scattering, with a narrow outlier in Social and Intergroup Psychology (1).
Collaboration60/100
Collaboration scores 60 with 19 distinct coauthors recorded. The cataloged papers are middle-author positions, and the recurring coauthor list — including Qingxiang Guo, Wenke Fan, Yingxin Shang, Weixu Wang, and Shuofeng Zhao — suggests a stable team around full-duplex dialogue work rather than solo output.
Activity57/100
Activity scores 57. OpenAlex counts 6 works in the recent three-year window, all falling in 2026, indicating a concentrated burst of arXiv output rather than a sustained multi-year record. Both cataloged papers appeared in 2026, consistent with an early-stage researcher ramping up.
Reproducibility38/100
Evolve's reproducibility assessment covers 2 papers with a mean score of 38. ECHO is rated medium readiness, while DuplexDrama is rated low, consistent with the verdict that its 800-hour synthesized dataset lacks human perceptual validation. The two-paper sample means this score should be read as preliminary.
Research summary
Hongwei Cai is an early-stage researcher at Sun Yat-sen University working on speech processing and spoken dialogue systems, with a profile name-resolved from public bibliographic data (OpenAlex). The cataloged output is concentrated in full-duplex spoken dialogue: two 2026 arXiv papers — ECHO, a matched-contrast benchmark for context-sensitive turn-taking, and DuplexDrama, a synthesized 800-hour TTS dialogue dataset covering personas, expressive speech, and sound events. OpenAlex records 6 works in the recent three-year window, all in 2026, with topic counts led by Speech Recognition and Synthesis (4), Speech and dialogue systems (4), and Speech and Audio Processing (3), plus single mentions in Social and Intergroup Psychology and discourse/communication strategies. Citation indicators (h-index 0, i10-index 0) reflect output too recent to have accumulated citations. Field distribution splits evenly (0.5/0.5) between two closely related spoken-dialogue labels, which likely depresses the computed focus score of 0 despite a coherent topical core. Collaboration is the strongest computed axis (60, with 19 distinct coauthors). Evolve's reproducibility assessment across the two reviewed papers averages 38 (ECHO medium, DuplexDrama low); verdicts credit ECHO's matched-insertion design for exposing YIELD bias while flagging label-dependent interruption acoustic processing and a missing gain-free control, and note DuplexDrama's pipeline is solid but lacks human perceptual validation.
Collaboration pattern
The recorded collaboration pattern is team-based: 19 distinct coauthors, with a recurring core including Qingxiang Guo, Wenke Fan, Yingxin Shang, Weixu Wang, Shuofeng Zhao, Dawei Yang, Zhiyang Zhou, and Lin Yang. On both cataloged papers Hongwei Cai sits in a middle author position rather than first or corresponding, consistent with contributing within a larger Sun Yat-sen University group working on full-duplex spoken dialogue.
Research keywords
Highlights
ECHO: matched-contrast benchmark exposes YIELD bias in full-duplex turn-taking
DuplexDrama: 800-hour synthesized full-duplex dialogue dataset
Timeline
- 2026DuplexDrama, a synthesized full-duplex dialogue dataset covering scenarios, full-duplex behaviors, expressive speech, and sound events, appeared on arXiv.
- 2026ECHO, a matched-contrast benchmark for context-sensitive turn-taking in full-duplex dialogue, appeared on arXiv.
- 2026OpenAlex recorded 6 works in the recent three-year window, all in 2026, with topics led by Speech Recognition and Synthesis (4) and Speech and dialogue systems (4).
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric indicators (coauthor counts, topics, field distribution, h-index) are computed from OpenAlex records only; the profile is name-resolved from public bibliographic data.
- Both reviewed papers are 2026 arXiv preprints; citation metrics (h-index 0, i10-index 0) reflect recency rather than confirmed absence of impact.
- The reproducibility mean of 38 rests on only 2 reviewed papers, so the score is preliminary.
- No venue, funding, award, or career-history data was available in the provided snapshot.