Salvato in:
Dettagli Bibliografici
Autori principali: Hsu, Chan-Jan, Tseng, Liang-Hsuan, Lin, Yi-Cheng, Kuo, Yen-Chun, Chou, Ju-Chieh, Chang, Kai-Wei, Lee, Hung-yi, Busso, Carlos
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2601.06329
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911722294476800
author Hsu, Chan-Jan
Tseng, Liang-Hsuan
Lin, Yi-Cheng
Kuo, Yen-Chun
Chou, Ju-Chieh
Chang, Kai-Wei
Lee, Hung-yi
Busso, Carlos
author_facet Hsu, Chan-Jan
Tseng, Liang-Hsuan
Lin, Yi-Cheng
Kuo, Yen-Chun
Chou, Ju-Chieh
Chang, Kai-Wei
Lee, Hung-yi
Busso, Carlos
contents Generative spoken language models pretrained on large-scale raw audio can continue a speech prompt with appropriate content while preserving attributes like speaker and emotion, serving as foundation models for spoken dialogue. In prior literature, these models are often evaluated using ``global token perplexity'', which directly applies the text perplexity formulation to speech tokens. However, this practice overlooks fundamental differences between speech and text modalities, possibly leading to an underestimation of the speech characteristics. In this work, we propose a variety of likelihood- and generative-based evaluation methods that serve in place of naive global token perplexity. We demonstrate that the proposed evaluations more faithfully reflect perceived generation quality, as evidenced by stronger correlations with human-rated mean opinion scores (MOS). When assessed under the new metrics, the relative performance landscape of spoken language models is reshaped, revealing a significantly reduced gap between the best-performing model and the human topline. Together, these results suggest that appropriate evaluation is critical for accurately assessing progress in spoken language modeling.
format Preprint
id arxiv_https___arxiv_org_abs_2601_06329
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle On the Fallacy of Global Token Perplexity in Spoken Language Model Evaluation
Hsu, Chan-Jan
Tseng, Liang-Hsuan
Lin, Yi-Cheng
Kuo, Yen-Chun
Chou, Ju-Chieh
Chang, Kai-Wei
Lee, Hung-yi
Busso, Carlos
Computation and Language
Artificial Intelligence
Generative spoken language models pretrained on large-scale raw audio can continue a speech prompt with appropriate content while preserving attributes like speaker and emotion, serving as foundation models for spoken dialogue. In prior literature, these models are often evaluated using ``global token perplexity'', which directly applies the text perplexity formulation to speech tokens. However, this practice overlooks fundamental differences between speech and text modalities, possibly leading to an underestimation of the speech characteristics. In this work, we propose a variety of likelihood- and generative-based evaluation methods that serve in place of naive global token perplexity. We demonstrate that the proposed evaluations more faithfully reflect perceived generation quality, as evidenced by stronger correlations with human-rated mean opinion scores (MOS). When assessed under the new metrics, the relative performance landscape of spoken language models is reshaped, revealing a significantly reduced gap between the best-performing model and the human topline. Together, these results suggest that appropriate evaluation is critical for accurately assessing progress in spoken language modeling.
title On the Fallacy of Global Token Perplexity in Spoken Language Model Evaluation
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2601.06329