Evaluating Gender Bias in Speech Translation

TL;DR

This paper introduces WinoST, a new evaluation set for measuring gender bias in speech translation, revealing significant bias disparities across four language pairs and emphasizing the need for bias mitigation.

Contribution

The paper presents WinoST, the first speech translation challenge set for gender bias evaluation, along with an evaluation protocol and empirical bias measurements.

Findings

Gender accuracy in speech translation is over 23% lower than in machine translation.

WinoST enables standardized bias measurement in speech translation systems.

Gender bias varies significantly across different language pairs.

Abstract

The scientific community is increasingly aware of the necessity to embrace pluralism and consistently represent major and minor social groups. Currently, there are no standard evaluation techniques for different types of biases. Accordingly, there is an urgent need to provide evaluation sets and protocols to measure existing biases in our automatic systems. Evaluating the biases should be an essential step towards mitigating them in the systems. This paper introduces WinoST, a new freely available challenge set for evaluating gender bias in speech translation. WinoST is the speech version of WinoMT which is a MT challenge set and both follow an evaluation protocol to measure gender accuracy. Using a state-of-the-art end-to-end speech translation system, we report the gender bias evaluation on four language pairs and we show that gender accuracy in speech translation is more than 23% lower than in MT.