AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Learnable Fractional Superlets with a Spectro-Temporal Emotion Encoder for Speech Emotion Recognition

Alaa Nfissi, Wassim Bouachir, Nizar Bouguila, Brian L Mishara

ICLR 2026 · Architectures and efficiency › RNNs and normalization · approach: Non-autoregressive translation

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
make superlet orders, cycles and frequency grid learnable (LFST) instead of fixed superlet (ablation)STEE · NSPL-CRISEaccuracy (%): 74.9 -> 76.9Improved
make superlet orders, cycles and frequency grid learnable (LFST) instead of fixed superlet (ablation)STEE · IEMOCAPaccuracy (%): 86.0 -> 87.5Improved
make superlet orders, cycles and frequency grid learnable (LFST) instead of fixed superlet (ablation)STEE · EMO-DBaccuracy (%): 90.1 -> 91.4Improved
use LFST front-end instead of STFTSTEE · NSPL-CRISEaccuracy (%): 73.1 -> 76.9Improved
use LFST front-end instead of LEAFSTEE · NSPL-CRISEF1-score (%): 72.1 -> 76.6Improved
use LFST front-end instead of Morlet wavelet (c=3) (ablation)STEE · IEMOCAPF1-score (%): 84.8 -> 86.8Improved
use fixed Morlet wavelet instead of STFTSTEE · NSPL-CRISEaccuracy (%): 73.1 -> 74.6Improved
use LEAF instead of STFTSTEE · IEMOCAPaccuracy (%): 84.8 -> 84.9No clear effect
use fixed superlet instead of Morlet waveletSTEE · EMO-DBaccuracy (%): 90.1 -> 90.1No clear effect
use LFST instead of STFT front-endSTEE · complexity (1 s input)FLOPs (GF): 0.36 -> 201.5Hurt
use LFST instead of STFT front-endSTEE · complexity (1 s input)latency (ms): 2.2 -> 74.9Hurt
use LFST instead of LEAF front-endSTEE · complexity (1 s input)peak memory (MB): 1156.0 -> 4532.8Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy