Speech processing researcher focused on low-resource language technology and cross-modal representation learning
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
13
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-09) 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: a complete, verified record of this researcher's total publication output — volume, venue types, and output diversity — is not available in the current snapshot. The output axis measures overall production volume and shape, and remains unscored pending fuller indexing.
Impact—
Not yet scored: although an h-index of 13 and an i10-index of 15 are recorded, the fuller citation evidence this axis needs — citation totals, citing works, and field-weighted influence — is insufficient in this snapshot, so no impact score is estimated here.
Focus0/100
Scores 0/100 for topical concentration: the field distribution is split exactly evenly (0.5/0.5) between speech processing / multimodal representation learning and speech processing / computational linguistics, and the OpenAlex topic record spans five clusters — Speech Recognition and Synthesis (75), Speech and Audio Processing (53), speech and dialogue systems (35), Emotion and Mood Recognition (31), and Natural Language Processing Techniques (15) — indicating breadth across adjacent speech and NLP areas rather than one narrow topic.
Collaboration22/100
Scores 22/100, reflecting a compact collaboration pattern: only 2 distinct co-authors appear in the recent three-year window, and the two Evolve-reviewed 2026 preprints place the researcher in middle and last author positions rather than corresponding author. The most frequent co-authors are Yannick Yomie Nzeuhang and Paulin Melatagia Yonta, consistent with a small, stable working circle.
Activity100/100
Scores 100/100: OpenAlex records 50 works in the recent three-year window, and the per-year series shows sustained acceleration — 8 works in 2020, 9 in 2021, 12 in 2022, 9 in 2023, a peak of 24 in 2024, then 8 in 2025 and 18 already in 2026 — against a 2009–2019 baseline of roughly 1–6 works per year, indicating a highly productive, currently active publishing pace.
Reproducibility55/100
Scores 55/100 (reproMean 55 across 2 reviewed papers). Both Evolve-reviewed papers carry medium repro-readiness: the HGNN cross-modal transfer study lacks a key acoustic-model baseline and its Yemba gain is not significant, while the frugal HGNN alignment study relies on single-run evaluation without frugal baselines — promising pipelines, not yet fully verifiable.
Research summary
Marie Tahon is a researcher at Université de Yaoundé I in Cameroon whose work centres on speech processing, with the bibliometric field distribution split exactly evenly (0.5/0.5) between multimodal representation learning and computational linguistics. Her OpenAlex-indexed record spans 2009–2026, with an h-index of 13 and an i10-index of 15, and publication counts accelerating markedly in recent years — 8 works in 2020, 12 in 2022, a peak of 24 in 2024, and 18 already recorded in 2026, for 50 works in the last three years. Her topic profile is led by Speech Recognition and Synthesis (75), Speech and Audio Processing (53), speech and dialogue systems (35), Emotion and Mood Recognition (31), and Natural Language Processing Techniques (15). The two Evolve-reviewed papers, both 2026 arXiv preprints, apply heterogeneous graph neural networks (HGNN) to cross-modal knowledge transfer and word-level speech–text alignment for Yemba, a low-resource Cameroonian language; Evolve judges the gains over baselines such as SAMU-XLSR as promising but limited by single-run evaluation, missing baselines, and a non-significant word-recognition gain. She publishes within a compact team, with Yannick Yomie Nzeuhang and Paulin Melatagia Yonta her most frequent co-authors and only 2 distinct co-authors in the recent three-year window.
Collaboration pattern
The collaboration pattern is compact and stable: only 2 distinct co-authors appear in the recent three-year window, and the Evolve-reviewed 2026 preprints place Marie Tahon in middle and last author positions, not corresponding author. The most frequent co-authors are Yannick Yomie Nzeuhang and Paulin Melatagia Yonta, suggesting a small, focused team around low-resource speech work rather than a large distributed network.
Research keywords
Highlights
HGNN cross-modal knowledge transfer for Yemba speech
Frugal HGNN link prediction for speech–text alignment
A focused programme on low-resource Cameroonian speech
Timeline
- 2009First works recorded in the OpenAlex series, with 1 publication in 2009 and 2 in 2010 — the start of a low-volume early period.
- 2016Annual output reaches 6 works, the highest yearly count of the first decade of the record.
- 2022Output rises to 12 works, the start of a sustained acceleration after 8 in 2020 and 9 in 2021.
- 2024Peak year in the per-year series, with 24 works recorded.
- 2026Two HGNN-based arXiv preprints on low-resource Yemba speech enter the Evolve catalog, with 18 works recorded so far this year.
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 (h-index, i10-index, per-year counts, topics, field distribution) come from OpenAlex records only; other databases may differ.
- This profile is name-resolved from public bibliographic data (OpenAlex, ORCID 0000-0002-6782-0332); identity confirmation is advised if the name is shared.
- Output and impact radar axes are unscored because the underlying evidence is insufficient in this snapshot.
- The Evolve-reviewed sample covers only two 2026 arXiv preprints on Yemba; 2025–2026 counts may reflect partial indexing and should be read with care.