PulseAugur
EN
LIVE 17:17:35

New NAR-MBR Decoding Boosts Speech Recognition Speed and Accuracy

By PulseAugur Editorial · [2 sources] ·

Researchers have developed a new non-autoregressive decoding framework for speech recognition, termed NAR-MBR decoding. This method aims to improve the speed of speech recognition by generating output tokens in parallel, overcoming the performance degradation typically associated with non-autoregressive models. By maximizing expected utility derived from samples rather than direct probability, NAR-MBR decoding achieves faster processing and outperforms previous non-autoregressive approaches on several benchmark datasets. AI

IMPACT This research offers a faster and potentially more accurate method for speech recognition, which could benefit real-time applications.

RANK_REASON The cluster contains an academic paper detailing a new research methodology for speech recognition.

Read on arXiv cs.CL →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New NAR-MBR Decoding Boosts Speech Recognition Speed and Accuracy

COVERAGE [2]

  1. arXiv cs.CL TIER_1 English(EN) · Hiroyuki Deguchi, Takatomo Kano, Katsuki Chousa, Marc Delcroix ·

    Non-Autoregressive Minimum Bayes' Risk Decoding for Fast Speech Recognition

    arXiv:2606.17537v1 Announce Type: cross Abstract: Non-autoregressive (NAR) decoding generates output tokens in parallel, making speech recognition faster than autoregressive decoding, which generates them sequentially from left to right. However, the recognition performance is de…

  2. arXiv cs.CL TIER_1 English(EN) · Marc Delcroix ·

    Non-Autoregressive Minimum Bayes' Risk Decoding for Fast Speech Recognition

    Non-autoregressive (NAR) decoding generates output tokens in parallel, making speech recognition faster than autoregressive decoding, which generates them sequentially from left to right. However, the recognition performance is degraded because NAR decoding cannot resolve uncerta…

RELATED ENTITIES

RELATED TOPICS