New framework uses calculus to optimize ASR vocabulary size

By PulseAugur Editorial · [1 sources] · 2026-05-14 06:19

Researchers have developed a calculus-based framework to determine the optimal vocabulary size for end-to-end Automatic Speech Recognition (ASR) systems. Unlike traditional hybrid ASR, end-to-end systems derive their vocabulary from training data, making vocabulary size a critical hyper-parameter. This new approach uses curve fitting and calculus principles to formally estimate the best vocabulary size, improving ASR performance on standard datasets like Librispeech. AI

IMPACT Formalizes an approach to optimize vocabulary size for end-to-end ASR, potentially improving model performance and training efficiency.

RANK_REASON Academic paper detailing a new methodology for optimizing a hyper-parameter in ASR systems. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

paper
other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

New framework uses calculus to optimize ASR vocabulary size

COVERAGE [1]

arXiv cs.CL TIER_1 English(EN) · Sunil Kumar Kopparapu · 2026-05-14 06:19

A Calculus-Based Framework for Determining Vocabulary Size in End-to-End ASR

In hybrid automatic speech recognition (ASR) systems, the vocabulary size is unambiguous, typically determined by the number of phones, bi-phones, or tri-phones present in the language. In contrast, end-to-end ASR systems derive their vocabulary, often referred to as tokens from …

COVERAGE [1]

A Calculus-Based Framework for Determining Vocabulary Size in End-to-End ASR

RELATED ENTITIES

RELATED TOPICS