Speech and audio processing researcher; efficient int8 speech enhancement for on-device inference on microcontrollers
2
papers here
0
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
5
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 yet: the snapshot's publication-volume data is incomplete, so this axis is left unscored. It would measure total publication count and long-run productivity across the researcher's full career record.
Impact—
Impact is unscored: available citation signals (h-index 5, i10-index 2) are too early-stage to support a reliable rating. This axis would measure citation influence, uptake, and scholarly reach of the work.
Focus—
Focus is not yet scored: the topic sample is too small for a robust concentration measure, though topic labels point to speech and audio processing. It would measure how consistently output clusters around core research themes.
Collaboration22/100
The collaboration score is 22/100, reflecting a compact network: the OpenAlex record lists only 2 distinct coauthors, with Riccardo Miccini and Rasmus Kongsgaard Olsson as recurring collaborators. Both 2026 catalog papers are first-authored but not corresponding, consistent with a close, small-team working mode rather than large multi-institution consortia. Co-authorship breadth, measured by distinct collaborators and joint-output spread, is currently the narrowest dimension in this profile.
Activity83/100
Activity scores 83/100, driven by a clear acceleration: works per year rise from 1 in 2022 to 5 in 2023, 6 in 2024, and a record 7 in 2025, with 16 works in the last three years. The 2026 count (3) is already logged, and two first-author arXiv preprints entered Evolve's catalog this year, confirming sustained, current research momentum.
Reproducibility—
Reproducibility is unscored: the snapshot lacks systematic release, code, and replication evidence across the portfolio. It would measure code and data availability, documentation quality, and independent replication readiness; the two reviewed papers each carry only a medium repro-readiness rating.
Research summary
Clément Laroche is a speech and audio processing researcher whose recent work concentrates on efficient, on-device speech enhancement for microcontroller-class hardware. The profile is name-resolved from public bibliographic data (an OpenAlex record linked to ORCID 0000-0002-2021-5047); no institution or country is recorded. The OpenAlex record spans 2015 to 2026 and shows an accelerating curve: annual works rose from 1 in 2022 to 5 in 2023, 6 in 2024, and 7 in 2025, with 16 works in the last three years. Citation indicators remain early-stage (h-index 5, i10-index 2). The topic profile is squarely audio-centric: Speech and Audio Processing (20 topic-assigned works), Speech Recognition and Synthesis (10), Advanced Adaptive Filtering Techniques (7), Music and Audio Processing (6), and Blind Source Separation Techniques (6). Two first-author 2026 arXiv preprints anchor the current agenda. One is a compute-matched study of dynamic depth for on-device speech enhancement, finding that early-exit training yields an efficient static model family while per-frame routing adds little. The other redesigns LiSenNet into an int8-quantized NPU model reaching PESQ 3.01 at 4.83 ms/frame on a microcontroller. Evolve rates both as rigorous but preliminary (single-dataset; small gains without error bars; medium repro-readiness). The coauthor network is compact: 2 distinct coauthors, with Riccardo Miccini and Rasmus Kongsgaard Olsson recurring.
Collaboration pattern
The collaboration footprint is compact: the OpenAlex record lists only 2 distinct coauthors, and the radar collaboration score is low at 22/100. Riccardo Miccini and Rasmus Kongsgaard Olsson are the recurring collaborators. Both Evolve-catalog papers are first-authored but not corresponding, consistent with a hands-on lead-contributor role inside a small, stable team rather than large multi-institution consortia.
Research keywords
Highlights
Compute-matched test of per-frame early exit for on-device speech enhancement
LiSenNet redesigned as an int8 NPU model for microcontroller speech enhancement
A coherent thread in efficient on-device speech enhancement
The two 2026 preprints share one agenda: making neural speech enhancement cheap enough for microcontrollers via int8 quantization, streaming inference, and compute-matched evaluation. Evolve rates both papers as medium repro-readiness.
Timeline
- 2015Earliest publications appear in the OpenAlex record (2 works).
- 2023Annual output rises to 5 recorded works, up from 1 in 2022 — the start of a sustained acceleration.
- 2024Output grows again to 6 recorded works.
- 2025Peak annual output in the record: 7 works.
- 2026Two first-author arXiv preprints on efficient on-device speech enhancement enter the Evolve catalog; 3 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 are computed from an OpenAlex record only; citation and venue coverage may be incomplete.
- Institution, country, and career facts are not in the input data; the profile is name-resolved from public bibliographic data (OpenAlex, ORCID 0000-0002-2021-5047).
- Output, impact, focus, and reproducibility dimensions are unscored because the snapshot's underlying data is insufficient.
- Qualitative assessment draws on only two Evolve-reviewed papers, both 2026 arXiv preprints not yet peer-reviewed.