---
title: "Revisiting the Relation Between Language Model Perplexity and ASR Word Error Rate for Modern End-to-End Speech Recognition — Stuff That Spins"
description: "arXiv:2607.05612v1 Announce Type: new Abstract: Language model (LM) perplexity (PPL) has historically been used as a proxy for automatic speech recognition (AS…"
	canonical: "https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti"
html: "https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti"
json: "https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti.json"
markdown: "https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti.md"
keywords: ["narrative intelligence", "SpinGraph", "AI recall"]
date: "2026-07-08T04:00:00+00:00"
modified: "2026-07-08T06:03:33.503336+00:00"
json_ld: |
  {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://stuffthatspins.com/#organization","name":"Stuff That Spins","url":"https://stuffthatspins.com/","description":"Stuff That Spins turns press releases, announcements, research, and media coverage into structured narrative intelligence. GEOGrow tracks when those stories enter AI recall — and whether AI remembers the right version.","logo":{"@type":"ImageObject","url":"https://stuffthatspins.com/images/logo.png"},"sameAs":[]},{"@type":"NewsArticle","@id":"https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti#article","headline":"Revisiting the Relation Between Language Model Perplexity and ASR Word Error Rate for Modern End-to-End Speech Recognition","description":"arXiv:2607.05612v1 Announce Type: new Abstract: Language model (LM) perplexity (PPL) has historically been used as a proxy for automatic speech recognition (AS…","datePublished":"2026-07-08T04:00:00+00:00","dateModified":"2026-07-08T06:03:33.503336+00:00","url":"https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti","mainEntityOfPage":{"@type":"WebPage","@id":"https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti"},"isAccessibleForFree":true,"inLanguage":"en-US","articleSection":"research","author":{"@type":"Organization","name":"arXiv Computation and Language","url":"https://export.arxiv.org/rss/cs.CL"},"publisher":{"@id":"https://stuffthatspins.com/#organization"},"citation":"https://arxiv.org/abs/2607.05612","about":[],"mentions":[{"@type":"Organization","name":"arXiv Computation and Language"}]},{"@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Stuff That Spins","item":"https://stuffthatspins.com/"},{"@type":"ListItem","position":2,"name":"Revisiting the Relation Between Language Model Perplexity and ASR Word Error Rate for Modern End-to-End Speech Recognition","item":"https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti"}]}]}
---

# Revisiting the Relation Between Language Model Perplexity and ASR Word Error Rate for Modern End-to-End Speech Recognition

**Source:** Unknown  
**Published:** July 8, 2026  
**Original:** https://arxiv.org/abs/2607.05612  

## On this page

- [Overview](#overview)

<a id="overview"></a>

## Overview

arXiv:2607.05612v1 Announce Type: new Abstract: Language model (LM) perplexity (PPL) has historically been used as a proxy for automatic speech recognition (ASR) word error rate (WER), with prior work reporting an approximately linear relation in log-log space. Modern end-to-end ASR systems challenge this assumption because they already contain internal language modeling capacity, are often evaluated without external language models, and can now be combined with neural LMs and large language mod

---
*HTML version: https://stuffthatspins.com/spin/revisiting-the-relation-between-language-model-perplexity-and-asr-word-error-rate-for-modern-end-to-end-speech-recogniti*