Tian Tan: NLP and speech researcher — language-model negation understanding and open-source speech generation models
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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 publication-volume data is available yet. This axis would measure total research output — number of papers over time and authorship roles on them — so an output score cannot be computed from the current record.
Impact—
No citation-based data is available yet. This axis would measure scholarly influence — citations, citation velocity, and highly cited work — so no impact score is reported here.
Focus0/100
The focus dimension scores 0, and the raw field distribution explains why: cataloged output splits evenly — 0.5 speech processing / audio machine learning and 0.5 computational linguistics / NLP — across a negation-corpus paper and a speech-generation model report. Work currently spans two communities rather than concentrating in one.
Collaboration70/100
The collaboration dimension scores 70. The raw record shows 33 distinct coauthors across cataloged work, with 12 recurring collaborators including Ziyang Ma, Zhikang Niu, Tianrui Wang, Yang Liu, and Zeyu Xie. Both cataloged papers are multi-author arXiv efforts — first author on the negation corpus study and a middle-position co-author on the AuK speech model report — though this rests on only two cataloged papers.
Activity—
No longitudinal activity data is available yet. This axis would measure recent publishing cadence and year-over-year consistency, which the current two-paper, single-year snapshot cannot support.
Reproducibility78/100
Reproducibility scores 78, a mean computed over the 2 papers with reproducibility assessments (reproPapers = 2). The negation paper rates high readiness, backed by the released 1.8M-sample NegCue corpus, while the AuK speech report rates medium, with strong tables but no human MOS, variance, or full ablations. The average is respectable but the evidence base is thin.
Research summary
Tian Tan is a researcher whose cataloged work sits at the intersection of computational linguistics / NLP and speech processing / audio machine learning, with Evolve's field distribution assigning each 0.5. The record, name-resolved from public bibliographic data via ORCID 0000-0002-4723-0044, centers on two 2026 arXiv papers. As first author, Tian Tan led 'Not all Negation Cues are Equal', which builds NegCue, a 1.8M-sample corpus for continued pre-training of language models and LLMs on negation via self-supervised Next Sentence Polarity Prediction, and shows affixal negation cues yield the best negation understanding; Evolve judges the LM results convincing while noting small LLM gains and several tiny test sets. As a middle-position co-author, Tian Tan contributed to the AuK technical report, an open-source 1.5B-parameter foundational model for speech generation and editing built on rectified flow, flow matching, and consistency distillation, with claimed state-of-the-art TTS and editing results. Bibliometrics show 33 distinct coauthors, 12 recurring collaborators, a collaboration score of 70, and a reproducibility score of 78 averaged over the two assessed papers; the focus score is 0, reflecting the even two-field split. No institution, country, citation, or publication-volume data is currently available.
Collaboration pattern
The collaboration dimension scores 70, with 33 distinct coauthors across the cataloged record and 12 recurring collaborators — including Ziyang Ma, Zhikang Niu, Wenming Tu, Tianrui Wang, Ruiqi Yan, Nickk Huang, Yang Liu, Zeyu Xie, and Hui Wang. Both cataloged papers are large-team arXiv efforts, one first-authored and one middle-position, with no corresponding authorship observed. The evidence base is two papers, so the pattern is indicative rather than settled.
Research keywords
Highlights
Built NegCue: a 1.8M-sample negation corpus for continued pre-training
Affixal negation cues yield better negation understanding
Co-authored AuK, an open-source 1.5B-parameter speech generation and editing model
Timeline
- 2026First-authored 'Not all Negation Cues are Equal', introducing the 1.8M-sample NegCue corpus for negation-focused continued pre-training.
- 2026Co-authored the AuK technical report on an open-source foundational model for unified speech generation and editing.
- 2026Cataloged output spans both NLP and speech/audio machine learning, with 33 distinct coauthors observed and reproducibility assessed on 2 papers.
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.
- Dimension scores, the 33-coauthor count, and the field distribution come from public bibliometric data only; citation and publication-volume metrics were not available.
- Only 2 papers from the 2026 arXiv record are reviewed in the Evolve catalog; peer-reviewed venue information is not included.
- No institution or country is on record; identity is name-resolved from public bibliographic data via ORCID 0000-0002-4723-0044.
- Output, impact, and activity dimensions are unscored because their underlying metrics are missing from the current snapshot.