method · first seen 9 Oct 2026
SpecAugment
A data augmentation method for speech recognition that masks blocks of time and frequency in spectrograms.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Error Rate is Not Enough: Costs of Augmentation in Low-Resource ASRPapers · 9 Oct 2026 · Katsumi Ibaraki, David Chiang
- EmoNet: A Lightweight Attention-Based CNN for Speech Emotion RecognitionPapers · 25 Oct 2026 · Muhammad Moiz Khalid, Abdur Rehman Tariq, Muhammad Okasha Yamin
Latest sources
- EmoNet: A Lightweight Attention-Based CNN for Speech Emotion RecognitionPapers · 25 Oct 2026 · Muhammad Moiz Khalid, Abdur Rehman Tariq, Muhammad Okasha Yamin
- Error Rate is Not Enough: Costs of Augmentation in Low-Resource ASRPapers · 9 Oct 2026 · Katsumi Ibaraki, David Chiang