University of Michigan researcher focused on Korean lipreading and post-training quantization of text-to-speech
2
papers here
0
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
4
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-27) 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 output score is available yet for this identity. This axis would measure publication volume, venue mix, and productivity over time; the OpenAlex record offers only scattered per-year counts (2011–2020), too sparse to score.
Impact—
No impact score is available yet. This axis would measure citation accumulation and scholarly influence; the resolved OpenAlex record lists an h-index of 4 and an i10-index of 2, a citation footprint too thin for a reliable score.
Focus—
No focus score is available yet. This axis would measure topical concentration across a career; OpenAlex topics cluster in imaging and compressive sensing while the reviewed papers are speech-focused, so a single concentration score is not yet reliable.
Collaboration32/100
Scores 32, reflecting a small network: the OpenAlex record lists only 4 distinct coauthors, and the catalog's frequent collaborators — Junyoung Park, Yutae Kim, Hakjun Kim, and Taehoon Roh — point to a compact team behind the recent speech-processing preprints. Both reviewed papers carry first-author position without corresponding-author status, consistent with early-career work inside a tight collaboration group rather than an extended international network.
Activity0/100
Scores 0 because the OpenAlex record reports 0 works in the last three years, with per-year counts ending in 2020 (1 work) after a 2013 peak of 4 and earlier counts in 2011, 2012, 2015, and 2018. The two 2026 arXiv preprints in the Evolve catalog show current output this bibliometric identity does not capture, so the zero should be read as a record-lag or identity-resolution signal, not confirmed inactivity.
Reproducibility—
No reproducibility score is available yet. This axis would measure artifact release, code availability, and replication evidence across outputs; the catalog flags the two reviewed papers as medium and high repro-readiness, but no aggregate score exists.
Research summary
Se Un Park is a researcher affiliated with the University of Michigan whose Evolve catalog centers on two 2026 first-author arXiv preprints in speech processing. The first reports personalized Korean lipreading as visual speech recognition on OLKAVS, combining transfer, a speaker census, and LoRA-based adaptation to reach 9.95–12.19% character error rate against a 26.64% baseline, with rigorous paired evaluation but limited prior-method baselines. The second is a cross-architecture study of post-training quantization for text-to-speech, validated with real-kernel measurements and blind tests, and rated high for reproducibility readiness. This recent machine-learning line contrasts sharply with the OpenAlex bibliometric record resolved to the name, which lists an h-index of 4, an i10-index of 2, per-year counts spanning 2011–2020 (peaking at 4 works in 2013), zero works in the last three years, and topics in sparse and compressive sensing, fluorescence microscopy, and photoacoustic imaging. The mismatch suggests either a lagging identity record or possible same-name conflation; this profile is therefore name-resolved from public bibliographic data rather than a verified unique identity.
Collaboration pattern
Collaboration is concentrated in a small team: the OpenAlex identity lists 4 distinct coauthors, and the catalog's frequent collaborators — Junyoung Park, Yutae Kim, Hakjun Kim, and Taehoon Roh — suggest a stable group behind the 2026 speech-processing preprints. Both reviewed papers are first-author, non-corresponding works, indicating an early-career contributor operating within a tight group rather than leading multi-institution consortia; no mentoring or industry partnership can be inferred from the available data.
Research keywords
Highlights
Personalized Korean lipreading: CER from 26.64% to 9.95–12.19% on OLKAVS
Cross-architecture post-training quantization of text-to-speech
Timeline
- 2011Earliest year in the resolved OpenAlex record, with 2 publications counted.
- 2013Peak output year in the OpenAlex record, with 4 publications counted.
- 2020Most recent year with a publication count in the OpenAlex record (1 work); no further works counted since.
- 2026Two first-author arXiv preprints enter the Evolve catalog: personalized Korean lipreading on OLKAVS and a cross-architecture TTS quantization study.
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 scores and raw metrics come from a single OpenAlex identity record; no ORCID is linked, so the profile is name-resolved from public bibliographic data.
- The OpenAlex topical profile (sparse/compressive sensing, fluorescence microscopy, photoacoustic imaging; counts ending 2020) diverges sharply from the reviewed 2026 speech-processing preprints, suggesting record lag or possible same-name conflation.
- Output, impact, focus, and reproducibility dimensions are unscored because the underlying citation and artifact data are too thin.
- Only two Evolve-reviewed papers are available, both first-author and non-corresponding, so leadership and mentoring patterns cannot be assessed.