Early-career speech-AI researcher: full-duplex dialogue turn-taking, vocal emotion, and ASR evaluation 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: this snapshot provides no aggregate output measure beyond the three 2026 preprints, so publication volume and venue record are insufficient to compute this axis. It would measure total works, venue distribution, and output trajectory over time.
Impact—
Not yet scored: citation uptake is insufficient — the snapshot lists h-index 0 and i10-index 0, which is expected for 2026 arXiv preprints too recent to accrue citations. This axis would measure citations, h-index trajectory, and influence on later research.
Focus100/100
The focus score of 100 reflects a snapshot in which every recorded topic sits inside speech and audio: Speech Recognition and Synthesis (3), Speech and dialogue systems (2), Emotion and Mood Recognition (1), Speech and Audio Processing (1), and Music and Audio Processing (1), with the field distribution listing spoken dialogue systems / speech NLP. All three 2026 catalog papers are speech-centric — dialogue turn-taking, vocal emotion, and ASR token recovery — so topical concentration is maximal, though the corpus is small.
Collaboration48/100
The collaboration score of 48 is computed from a recent window showing 10 distinct co-authors across the three recorded works. Recurring names include Brandon Tai, Candice Fan, Luc Debaupte, Yi Zhong, Bill Wang, Yunqi Lu, Nikhil Johri, Ruben Aguilar, and Lisa Kaelin-Martin. All three 2026 papers are multi-authored, with him in a middle position on two and first author on VoiceCodeBench, and none as corresponding author — a contributor pattern typical of an early-career researcher embedded in a lab group.
Activity40/100
The activity score of 40 reflects a short recorded history: the counts-by-year series contains a single entry, 3 works in 2026, and the three-year window likewise reports 3 recent works. No earlier publication year appears in the snapshot, so the score captures a concentrated burst of arXiv preprints in one year rather than a sustained multi-year cadence; whether this pace continues is the main open question for the axis.
Reproducibility20/100
The reproducibility score of 20 comes from a mean of 20 across 1 evaluated paper. The reviewed work, 'Continue, Adapt, or Yield,' carries a low reproducibility-readiness rating, and Evolve's verdict flags partly circular evaluation labels in the Duplex Cue analysis. With only one paper reviewed so far, this axis rests on thin evidence and should be re-read as additional catalog papers receive verdicts.
Research summary
Tyler Baumgartner is an early-stage speech-AI researcher whose recorded work concentrates on how machines listen, respond, and evaluate: full-duplex spoken dialogue, vocal emotion recognition, and speech recognition of structured content. The OpenAlex snapshot credits three works, all 2026 arXiv preprints, spanning Speech Recognition and Synthesis (3), Speech and dialogue systems (2), Emotion and Mood Recognition (1), Speech and Audio Processing (1), and Music and Audio Processing (1) — a concentration that yields the maximum focus score of 100. The most developed line is turn-taking in full-duplex agents: 'Continue, Adapt, or Yield' introduces Duplex Cue, a 3x3 intent-by-response matrix, and reports that humans adapt on 68.2% of collaborative cues versus 34.8% for PersonaPlex, though Evolve cautions the labels are partly circular and rates reproducibility readiness low. He first-authored VoiceCodeBench on exact structured-token recovery in automatic speech recognition and co-authored VocalAffectBench on vocal emotion recognition in AI audio models; both await full Evolve review. Citation impact is not yet measurable — the snapshot lists h-index 0 and i10-index 0 — and the activity score of 40 reflects a single recorded year: 3 works in the last three years, 10 distinct co-authors, and a collaboration score of 48. The profile is name-resolved from public bibliographic data, and reproducibility (20) rests on a single reviewed paper.
Collaboration pattern
Baumgartner's recorded work is entirely collaborative: 10 distinct co-authors appear across his three 2026 works, with recurring collaborators including Brandon Tai, Candice Fan, Luc Debaupte, Yi Zhong, Bill Wang, Yunqi Lu, Nikhil Johri, Ruben Aguilar, and Lisa Kaelin-Martin. All three papers are multi-authored, with him in a middle position on two and first author on VoiceCodeBench, and none as corresponding author — a pattern consistent with an early-career researcher working within a lab group.
Research keywords
Highlights
Duplex Cue: quantifying in-turn adaptation to overlapping speech
VoiceCodeBench: first-authored ASR benchmark for structured-token recovery
Baumgartner is first author of this 2026 arXiv preprint, which addresses evaluating exact structured-token recovery in automatic speech recognition. It is listed in the catalog without an Evolve verdict, keywords, or a reproducibility rating yet.
VocalAffectBench: evaluating vocal emotion recognition in AI audio models
A 2026 arXiv preprint with Baumgartner as a middle-position co-author, proposing a benchmark for how AI audio models recognize vocal emotion — the evaluation side of his emotion-recognition topic footprint. It awaits a full Evolve review and currently carries no verdict, keywords, or reproducibility rating.
Timeline
- 2026Co-authored 'Continue, Adapt, or Yield: In-Turn Adaptation to Overlapping Speech in Full-Duplex Agents' on arXiv; this became the only catalog paper with a completed Evolve review.
- 2026First-authored 'VoiceCodeBench: Evaluating Exact Structured-Token Recovery in Automatic Speech Recognition' on arXiv.
- 2026Co-authored 'VocalAffectBench: Evaluating Vocal Emotion Recognition in AI Audio Models' on arXiv.
- 2026OpenAlex counts-by-year records all 3 of his works in 2026 (recent-3-year total: 3), making it his first recorded publication year in this snapshot.
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 are derived from OpenAlex only; institution, country, and ORCID are unresolved in this snapshot.
- Citation impact is absent from the record: h-index 0 and i10-index 0, consistent with three 2026 arXiv preprints too recent to have accrued citations.
- Only one of the three catalog papers ('Continue, Adapt, or Yield') has a completed Evolve review; the other two lack verdicts, keywords, and reproducibility ratings.
- The profile is name-resolved from public bibliographic data; with a relatively common name, identity confidence depends on the OpenAlex author cluster.