Early-career speech AI researcher building full-duplex dialogue methods and vocal-emotion / ASR benchmarks
3
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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—
Not yet scored: the snapshot lacks aggregate output totals beyond three recent 2026 works. This axis would measure cumulative publications, per-year production, and venue spread across the full record, none of which is currently available for this researcher.
Impact—
Not yet scored: h-index and i10-index both register 0 in this OpenAlex snapshot and no citation counts are available, so research impact cannot be assessed. This axis would measure citations, h-index trajectory, and uptake of the benchmark papers.
Focus100/100
Focus scores 100 because every measured signal points to one area. The field distribution gives the sole cataloged field — spoken dialogue systems / speech NLP — a full weight of 1, and all topic counts stay inside speech and audio: Speech Recognition and Synthesis (3), speech and dialogue systems (2), plus single works in emotion and mood recognition, speech and audio processing, and music and audio processing.
Collaboration48/100
Collaboration scores 48 from a mid-range footprint: 10 distinct coauthors across 3 recent works, including recurring partners such as Tyler Baumgartner, Brandon Tai, Candice Fan, Yi Zhong, and Bill Wang. Authorship shows one first-author paper and two middle positions, none corresponding, consistent with a contributing member of a small speech-AI team rather than an independent group lead.
Activity40/100
Activity scores 40 on thin recent volume: recent3yWorks is 3, and the counts-by-year record shows all three works surfacing in 2026, with no earlier years in the snapshot. This pattern fits an early-career researcher whose output is only beginning to register in OpenAlex rather than an extended gap in research.
Reproducibility20/100
Reproducibility scores 20, based on a single assessed paper (reproPapers 1, reproMean 20). The reviewed full-duplex study is rated low reproducibility readiness, and its Evolve verdict flags partly circular adaptation labels — a caveat for anyone attempting independent replication of the 68.2% versus 34.8% comparison.
Research summary
Luc Debaupte is an early-stage researcher in speech and audio AI whose public bibliographic record — resolved via OpenAlex — concentrates on spoken dialogue systems and speech natural language processing. The Evolve catalog lists three arXiv preprints, all dated 2026: a study of in-turn adaptation to overlapping speech in full-duplex agents, a benchmark for vocal emotion recognition in AI audio models, and a benchmark for exact structured-token recovery in automatic speech recognition. Topic counts reinforce the picture: Speech Recognition and Synthesis (3 works), speech and dialogue systems (2), plus single works in emotion and mood recognition, speech and audio processing, and music and audio processing. Field distribution assigns all cataloged work to a single field, spoken dialogue systems / speech NLP, which drives a perfect focus score. Recent activity is modest — 3 works in the last three years, all surfacing in 2026 — with 10 distinct coauthors. Authorship shows one first-authored paper and two middle positions, none corresponding. Reproducibility data covers a single assessed paper, rated low readiness with a mean score of 20. Output and citation-impact metrics — h-index and i10-index both 0 in this snapshot — remain insufficient to score.
Collaboration pattern
Debaupte works in small teams: 10 distinct coauthors across 3 recent arXiv works, with recurring names including Tyler Baumgartner, Brandon Tai, Candice Fan, Yi Zhong, Bill Wang, Yunqi Lu, Nikhil Johri, Ruben Aguilar, and Lisa Kaelin-Martin. The catalog shows one first-authorship and two middle positions with no corresponding-author role, suggesting a hands-on contributor inside a speech-AI group rather than a PI running an independent lab. The collaboration score of 48 reflects this mid-range footprint.
Research keywords
Highlights
Duplex Cue: quantifying in-turn adaptation to overlapping speech in full-duplex agents
VocalAffectBench: a first-authored benchmark for vocal emotion recognition
First-authored 2026 arXiv preprint introducing a benchmark for evaluating vocal emotion recognition in AI audio models. No Evolve review, keywords, or reproducibility assessment is available yet, so its contribution rests on the benchmark design alone for now.
VoiceCodeBench: evaluating exact structured-token recovery in ASR
Middle-authored 2026 arXiv preprint targeting how precisely automatic speech recognition systems recover exact structured tokens. It has not yet been reviewed by Evolve, so no verdict, keywords, or reproducibility rating exists.
Timeline
- 2026Co-authors "Continue, Adapt, or Yield" on in-turn adaptation to overlapping speech in full-duplex agents (middle author position, arXiv).
- 2026First-authors VocalAffectBench, a benchmark for evaluating vocal emotion recognition in AI audio models (arXiv).
- 2026Co-authors VoiceCodeBench on evaluating exact structured-token recovery in automatic speech recognition (arXiv).
- 2026All three cataloged works land in a single year, matching the counts-by-year record of 3 works in 2026 and 3 works over the last three years.
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.
- Bibliometrics come from an OpenAlex snapshot; h-index and i10-index both register 0, reflecting an early-stage record and possibly incomplete author disambiguation.
- No institution, country, or ORCID is present in the source record; this profile is name-resolved from public bibliographic data.
- Only one of the three cataloged papers carries an Evolve review and reproducibility assessment; VocalAffectBench and VoiceCodeBench lack verdicts, keywords, and readiness ratings.
- Output and impact dimensions are unscored because aggregate publication and citation data are missing from the snapshot.