Crowdsourcing MUSHRA Tests in the Age of Generative Speech Technologies: A Comparative Analysis of Subjective and Objective Testing Methods

TL;DR

This paper adapts the MUSHRA testing framework for crowdsourcing to evaluate generative speech models, comparing platform results with expert data and assessing objective metrics, to enable scalable and reliable speech quality evaluation.

Contribution

It introduces a crowdsourced MUSHRA testing methodology for generative speech codecs and evaluates platform-specific biases and metric effectiveness.

Findings

Crowdsourced tests show platform-specific biases.

Traditional metrics undervalue generative models.

Codec-aware metrics improve evaluation accuracy.

Abstract

The MUSHRA framework is widely used for detecting subtle audio quality differences but traditionally relies on expert listeners in controlled environments, making it costly and impractical for model development. As a result, objective metrics are often used during development, with expert evaluations conducted later. While effective for traditional DSP codecs, these metrics often fail to reliably evaluate generative models. This paper proposes adaptations for conducting MUSHRA tests with non-expert, crowdsourced listeners, focusing on generative speech codecs. We validate our approach by comparing results from MTurk and Prolific crowdsourcing platforms with expert listener data, assessing test-retest reliability and alignment. Additionally, we evaluate six objective metrics, showing that traditional metrics undervalue generative models. Our findings reveal platform-specific biases and emphasize codec-aware metrics, offering guidance for scalable perceptual testing of speech codecs.