AI Product Manager for Speech Data Evaluation and Quality Assurance
As AI Product Manager, I led the quality acceptance and analysis of training data for Japanese and English speech synthesis models. I evaluated over 200 hours of AI speech training data to ensure compliance with model requirements and conducted comprehensive system assessments of TTS products. I developed an evaluation framework with multiple dimensions including pronunciation accuracy, rhythm, and dialect, and validated model performance through A/B testing and user research. • Quality acceptance of TTS model training audio data for AI applications • Developed a multi-dimensional evaluation system (accuracy, rhythm, dialect) • Over 200 hours of labeled and evaluated speech data • Led production of in-depth TTS model evaluation reports