Telefónica researcher advancing multilingual speech-LLMs, federated ASR, and European-language instruction tuning
3
papers here
3
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
1
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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—
Total-output metrics are insufficient here: the snapshot lacks a full publication count and venue breakdown, exposing only 15 works in the last three years. This axis would measure cumulative productivity and publication volume across the entire career.
Impact—
Impact cannot be scored yet: the OpenAlex record reports h-index 1 and i10-index 0, far too little citation signal to evaluate. This axis would measure citation counts, scholarly influence, and uptake of the work by other researchers.
Focus0/100
The focus score is 0/100, and the raw distribution supports it: fieldDist splits evenly at about 0.33 across multilingual ASR, speech-LLMs, and multilingual instruction tuning, while indexed topics span speech recognition and synthesis (9), speech and dialogue systems (6), topic modeling (5), speech and audio processing (5), and NLP techniques (4). Work is active but thematically dispersed rather than concentrated in one niche.
Collaboration41/100
Collaboration scores 41/100, based on 7 distinct coauthors in the recent window — Aleix Sant, Lorenzo Concina, Marco Matassoni, Alessio Brutti, Filippo Vella, Fernando López, and Carlos Escolano recur most often. First-author roles on both 2026 speech-LLM papers indicate leadership, while the middle-author EuroAlpaca paper shows cross-team participation; the modest coauthor count suggests small-team work rather than large consortia.
Activity81/100
Activity is the strongest axis at 81/100: the counts-by-year series shows 13 indexed works in 2026 versus 1 in each of 2025, 2024, and 2004, and 15 works in the last three years. The record is dominated by a concentrated 2026 burst — three arXiv preprints in the Evolve catalog are all from that year — indicating a steep recent acceleration.
Reproducibility55/100
Reproducibility averages 55 (reproMean) across 3 assessed papers, all rated medium readiness. Evolve's reviews flag recurring limits: single-run federated benchmarks without variance estimates, no significance tests in the challenge submission, and a single training seed with unreleased data in EuroAlpaca — consistent methodological transparency, but replication would need extra author input.
Research summary
Jordi Luque is a Spain-based researcher at Telefónica working at the intersection of speech technology and large language models. The profile is name-resolved from OpenAlex bibliographic records (no ORCID linked). The snapshot shows a low cumulative citation footprint — h-index 1 and i10-index 0 — but a pronounced recent surge: 13 works indexed in 2026 and 15 works over the last three years, with activity scoring 81/100. Topic distributions concentrate on speech recognition and synthesis (9), speech and dialogue systems (6), topic modeling (5), speech and audio processing (5), and NLP techniques (4), while the field split is even across multilingual ASR, speech-LLMs, and multilingual instruction tuning. Three 2026 arXiv papers appear in the Evolve catalog: a first-author federated-learning benchmark of four Speech-LLMs on MLS with 316 clients (FedProx helps only EuroLLM; single-run limits), a first-author Interspeech 2026 MLC-SLM challenge submission reaching 0.81 on 21-language speech MCQA with a frozen Voxtral-24B plus in-context learning, and a middle-author EuroAlpaca study on task-preserving localisation of instruction data for 50 languages. Reproducibility readiness averages 55 across three assessed papers; single-run results and unreleased data recur as limitations.
Collaboration pattern
Collaboration is small-team: 7 distinct coauthors appear in the recent window, and the same seven — Aleix Sant, Lorenzo Concina, Marco Matassoni, Alessio Brutti, Filippo Vella, Fernando López, and Carlos Escolano — recur as frequent collaborators. Jordi Luque takes first-author roles on both 2026 speech-LLM preprints while contributing as a middle author on EuroAlpaca, a pattern consistent with leading focused speech-model studies within a stable group.
Research keywords
Highlights
First-author federated benchmark of multilingual Speech-LLMs
Eloquence system: frozen Voxtral-24B with in-context learning
EuroAlpaca: task-preserving localisation of instruction data
Timeline
- 2026Released a first-author federated-learning benchmark of multilingual Speech-LLMs on arXiv, testing four models across 316 MLS clients.
- 2026Submitted the first-author Eloquence system to Task 2 of the Interspeech 2026 MLC-SLM challenge, scoring 0.81 on 21-language speech MCQA.
- 2026Co-authored EuroAlpaca on task-preserving localisation of instruction data for 50 languages.
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 metrics (h-index, counts-by-year, topics) come from an OpenAlex snapshot; no ORCID is linked, so identity is resolved from public bibliographic data.
- Output and impact dimensions are unscored: the snapshot lacks total publication counts and meaningful citation signal (h-index 1, i10-index 0).
- All catalog papers are 2026 arXiv preprints, so the 2026 activity spike reflects recent preprint posting rather than a long publication history.
- Reproducibility scores cover only 3 assessed papers, all rated medium readiness.