Thomas Ranzenberger
Georg Simon Ohm University of Applied Sciences Nuremberg
Speech processing researcher in Nuremberg — multimodal ASR, latent speech representations, and spoken-document corpora
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
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—
Not yet scored: the snapshot contains no publication-count, venue, or productivity metrics for this researcher beyond the two Evolve-reviewed arXiv papers. This axis would measure annual output volume and venue spread once a fuller bibliometric record is linked.
Impact—
Not yet scored: no citation counts, h-index, or influence metrics are present in the provided data. This axis would measure citation uptake and downstream reuse of the two 2026 arXiv papers once citation records become available.
Focus0/100
Records a focus score of 0, driven by the field distribution in the underlying data: the reviewed work splits evenly (0.5/0.5) between 'speech processing / automatic speech recognition' and 'speech processing and natural language processing'. With only two catalog papers, this even split across two adjacent field labels yields no dominant single field; it reads as an early speech-plus-NLP profile rather than scattered, unrelated interests.
Collaboration32/100
Scores 32 on a 0-100 scale, reflecting a compact collaboration footprint: only 4 distinct coauthors appear in the underlying record, and the frequent-collaborator list is a stable cluster of Steffen Freisinger, Tobias Bocklet, Korbinian Riedhammer, and Philipp Seeberger. On the two catalog papers he is first author once and middle author once, neither as corresponding author — a pattern consistent with close-knit single-group work rather than distributed multi-institution projects.
Activity—
Not yet scored: the snapshot provides no year-by-year publication cadence beyond the two 2026 papers, so recent activity cannot be judged. This axis would measure publication frequency and research continuity over time.
Reproducibility55/100
Mean reproducibility readiness is 55, computed over 2 papers, both rated medium. Evolve's reviews flag concrete gaps: the topic-localization paper reports consistent gains over text-only baselines but runs no significance tests, and the Spoken Wikipedia Presentation Corpus shows strong audio baselines while zero-shot slide context fails and the fine-tuning gain remains untested. Both artifacts are arXiv-available, leaving room to raise readiness with significance testing and fine-tuned baselines.
Research summary
Thomas Ranzenberger is a speech processing researcher at Georg Simon Ohm University of Applied Sciences Nuremberg whose Evolve-reviewed work sits at the intersection of automatic speech recognition and natural language processing: the catalog's field distribution splits evenly (0.5/0.5) between 'speech processing / automatic speech recognition' and 'speech processing and natural language processing'. He entered the Evolve catalog in 2026 with two arXiv papers. As first author of 'The Spoken Wikipedia Presentation Corpus' he contributed to a 553-hour multimodal ASR resource pairing Spoken Wikipedia audio with LLM-generated slides; Evolve's review finds strong audio baselines, zero-shot slide context failing, and an unproven fine-tuning gain. As a middle author on 'Reusing Latent Speech Representations for Query-Conditioned Topic Localization in Transcripts' he worked on reusing ASR encoder states for query-based span localization in transcripts, where gains over text-only approaches are consistent but domain-dependent and lack significance testing. His collaboration footprint is compact — 4 distinct coauthors (Steffen Freisinger, Tobias Bocklet, Korbinian Riedhammer, Philipp Seeberger), collaboration score 32 — and both papers carry medium reproducibility readiness with a mean of 55 across 2 papers. Output, impact, and activity metrics are not yet available in this snapshot.
Collaboration pattern
The collaboration footprint is small and stable: 4 distinct coauthors, a collaboration score of 32, and a recurring cluster of Steffen Freisinger, Tobias Bocklet, Korbinian Riedhammer, and Philipp Seeberger. Across the two 2026 catalog papers he appears once as first author and once in a middle position, never as corresponding author — consistent with contributing inside a single speech-processing group rather than wide external networking.
Research keywords
Highlights
First-authored 553-hour multimodal ASR corpus
Reusing ASR encoder states for query-conditioned transcript search
Both 2026 outputs rated medium reproducibility readiness
Both catalog papers carry a medium reproducibility rating, giving a mean score of 55 across 2 papers; Evolve flags missing significance tests and untested fine-tuning as the main gaps, both addressable since the artifacts are on arXiv.
Timeline
- 2026Released The Spoken Wikipedia Presentation Corpus on arXiv as first author — a 553-hour pairing of Spoken Wikipedia audio with LLM-generated slides for multimodal ASR.
- 2026Co-authored (middle position) the arXiv paper on reusing latent speech representations for query-conditioned topic localization in transcripts.
- 2026Both papers entered the Evolve review catalog with medium reproducibility ratings and a mean readiness score of 55.
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 are computed from the provided bibliometric snapshot only (2 Evolve-reviewed arXiv papers); no ORCID is linked, so this profile is name-resolved from public bibliographic data.
- Output, impact, and activity axes are unscored (dimMissing) because no publication counts, citation metrics, or year-by-year cadence data were provided.
- The reproducibility mean of 55 rests on only 2 papers, both arXiv preprints; peer-reviewed venue status is not recorded in the snapshot.
- Country is null and affiliation is limited to the institution name, so identity confidence may be lower in cases of name collision.