Imitation Learning for Elder-Facing Speech Synthesis

2026-06-18

AI Summary

This study addresses the neglect of age-related sensory and cognitive decline in current text-to-speech systems, which often leads to reduced speech intelligibility for older adults, compounded by the high cost and user fatigue associated with collecting preference data from this population. To tackle these challenges, this work proposes a novel optimization framework that, for the first time, integrates imitation learning into elderly-oriented speech synthesis by combining expert demonstrations with a two-stage Online Policy Reward Learning (OPRL) approach. This method effectively mitigates reward hacking under limited data conditions and enhances the Group Relative Policy Optimization (GRPO) algorithm. Experimental results demonstrate that the proposed approach consistently outperforms both the GRPO baseline and supervised learning methods in both objective metrics and subjective evaluations, significantly improving speech intelligibility and listening comfort for older users.

Abstract

Recent advances in text-to-speech (TTS) synthesis have achieved highly natural and expressive speech generation. However, these systems are designed for general adults and overlook older adults' speech comprehension needs due to age-related sensory and cognitive decline. Prior work involves older adults by collecting preference feedback to tune model parameters. However, obtaining sufficient preference data is costly and difficult, as older adults quickly become fatigued during collection. In this paper, we propose a novel imitation learning (IL) framework to learn TTS models from expert demonstrations. We further improve Group Relative Policy Optimization (GRPO) with two-stage on-policy reward learning (OPRL) to mitigate reward hacking under limited supervision from expert demonstration. Experimental results show that GRPO w/ OPRL outperforms GRPO and supervised baselines in objective and subjective metrics. Audio samples are available at https://dongru1.github.io/demo/im-efss

Problem

• Elder-facing Speech Synthesis
• Imitation Learning
• Text-to-Speech
• Preference Data Collection
• Age-related Decline

Innovation

• Imitation Learning
• Elder-Facing Speech Synthesis
• On-Policy Reward Learning
• GRPO
• Text-to-Speech