Speech NLP researcher: efficient zero-shot SpeechLLMs, token alignment, and probing fairness in speech models
2
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-20) 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 scored: the underlying publication-volume data (total output count, venues, years) is not available in this snapshot. This axis would measure how much the researcher publishes and where, based on a full indexed record.
Impact—
Not scored: no citation counts, h-index, or influence metrics are present in the input data. This axis would measure citation uptake and scholarly influence once a reliable citation record is linked to this researcher.
Focus100/100
Scores 100 on focus, computed from a field distribution containing exactly one field — speech and natural language processing — across the indexed record. Both reviewed papers sit inside that scope: token-level alignment for zero-shot SpeechLLMs, and probing plus steering of speech-model representations. The signal is unusually coherent, though it rests on a thin record, so the concentration may partly reflect a small sample.
Collaboration36/100
Collaboration scores 36, derived from 5 distinct coauthors in the indexed record. The visible pattern is small-team research: recurring partners include Gerasimos Spanakis (the most frequent), Nicolas Bourrel, Yusuf Can Semerci, and Jan Scholtes. A first-author-led arXiv paper alongside one middle-position contribution suggests compact, mentor-inclusive teams rather than large consortia; the low score reflects limited co-author breadth, not absence of collaboration.
Activity—
Not scored: recent-activity signals (publication cadence over time, current affiliation) are unavailable beyond two 2026 arXiv entries. This axis would measure ongoing research tempo and engagement across recent years.
Reproducibility100/100
Reproducibility scores 100, but the raw metric covers only one paper (reproMean 100 from reproPapers 1) — the first-author work on token-level alignment for zero-shot SpeechLLMs, whose Evolve review rates reproducibility readiness as high. The same review cautions that headline claims rest on single-seed runs. Treat the perfect score as a strong start on one paper, not an established reproducibility track record.
Research summary
Abderrahmane Issam works at the intersection of speech technology and natural language processing; this profile is built from a thin but highly coherent bibliographic record and is name-resolved from public bibliographic data. The indexed field distribution contains exactly one field — speech and natural language processing — and both cataloged 2026 arXiv papers sit within it. The first-authored 'Align, Integrate, and Fire' proposes DTW-aligned Continuous Integrate-and-Fire (CIF) representations with single-layer knowledge distillation to build cheap zero-shot SpeechLLMs for recognition and translation; Evolve's review judges the gains over SpeechMapper as likely real while cautioning that headline claims rest on single-seed runs, and rates reproducibility readiness high. A second 2026 paper, with Issam in a middle position, critically examines whether representation steering in speech models constitutes a fairness fix. Collaboration data show 5 distinct coauthors — Gerasimos Spanakis most frequently, plus Nicolas Bourrel, Yusuf Can Semerci, and Jan Scholtes — a small-team pattern behind the collaboration score of 36. Focus (100) and reproducibility (100) top the scale, the latter from a single paper. Output, impact, and activity are unscored for lack of data.
Collaboration pattern
The visible pattern is small-team collaboration: 5 distinct coauthors, with Gerasimos Spanakis recurring most often, alongside Nicolas Bourrel, Yusuf Can Semerci, and Jan Scholtes. Issam takes a first-author lead on the zero-shot SpeechLLM alignment work and a middle position on the representation-steering critique, consistent with junior-researcher roles inside a compact, mentor-inclusive group rather than large consortia.
Research keywords
Highlights
First-author framework for cheap zero-shot SpeechLLMs
High reproducibility readiness on the lead paper
Critical stance on probing-based fairness claims
In a 2026 middle-author arXiv paper, 'A Probe Shift Is Not a Fairness Fix', Issam co-authors a critical analysis showing the limits of representation steering in speech models — evidence of methodological skepticism, not just system building.
Timeline
- 2026Posts the first-authored arXiv preprint 'Align, Integrate, and Fire: Efficient Token-Level Alignment for Zero-Shot SpeechLLMs', proposing DTW-aligned CIF with single-layer knowledge distillation for zero-shot recognition and translation.
- 2026Co-authors (middle position) 'A Probe Shift Is Not a Fairness Fix: The Limits of Representation Steering in Speech Models', a critical study of probing-based fairness interventions in speech models.
- 2026Evolve's review of the lead paper rates reproducibility readiness as high while flagging that headline gains rest on single-seed runs.
- 2026All indexed work in 2026 falls within a single field — speech and natural language processing — producing a maximally concentrated field profile (focus 100).
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 indicators come from an OpenAlex-based snapshot only; no citation counts, h-index, or venue metadata were provided.
- Only two papers, both 2026 arXiv preprints, are covered; output, impact, and activity axes lack data and are not scored.
- The reproducibility score of 100 rests on a single reviewed paper (reproPapers 1) and is indicative, not a track record.
- No institution, country, or ORCID is confirmed; the profile is name-resolved from public bibliographic data and may need manual verification.