Universal, Explainable and Extensible Automatic Evaluation of Synthesized Speech

• Project/Area Number: 25K00143
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 61010:Perceptual information processing-related
• Research Institution: National Institute of Information and Communications Technology
• Principal Investigator: Cooper Erica 国立研究開発法人情報通信研究機構, ユニバーサルコミュニケーション研究所先進的音声翻訳研究開発推進センター, 主任研究員 (30843156)
• Co-Investigator(Kenkyū-buntansha): HUANG WENCHIN 名古屋大学, 情報学研究科, 助教 (91002385)
• Project Period (FY): 2025-04-01 – 2029-03-31
• Project Status: Granted (Fiscal Year 2025)
• Budget Amount: ¥18,720,000 (Direct Cost: ¥14,400,000、Indirect Cost: ¥4,320,000); Fiscal Year 2028: ¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000); Fiscal Year 2027: ¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000); Fiscal Year 2026: ¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000); Fiscal Year 2025: ¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
• Keywords: speech evaluation / speech synthesis

Outline of Research at the Start

Listening tests to evaluate speech synthesizers are time-consuming, costly, and a significant bottleneck to experimental evaluation, and better evaluation methodologies for synthesized speech are needed. To this end, we plan 1) collection of new datasets for automatic speech synthesis quality estimation; 2) development of quality predictors that consider explainability, context, and generalizability; 3) the continuation of international shared-task challenges on speech quality estimation.