Publications

Paper 1

Figure 1

Speaker Disentanglement of Speech Pre-trained Model Based on Interpretability

Xiaoxu Zhu

Technical Report, 2024

Paper 2

Figure 2

A polyphone BERT for Polyphone Disambiguation in Mandarin Chinese

Xiaoxu Zhu

Interspeech, 2022

Paper 3

Figure 3

Multimodal Sentiment Analysis via Efficient Multimodal Transformer and Modality-Aware Adaptive Training Strategy

Xiaoxu Zhu

ACM MuSe-Mimic Subchallenge, 2023

Patents & Awards

Patents

  • 多音字读音预测网络的训练方法、语音生成方法及装置
    Patent No: CN115273809A
  • 残差网络的训练和语音合成方法、装置、设备及介质
    Patent No: CN112562655A
  • 模型训练和语音合成方法、装置、设备及介质
    Patent No: CN116206591A
  • 一种模型训练和语音合成方法、装置、设备及介质
    Patent No: CN115294955A

Awards & Recognitions

  • 2024 Intel Mini Hackathon - Excellent Work Award
  • 2023 ACM MuSe-Mimic Subchallenge - Second Place
  • LPCNet Contributor - GRU B conditioning optimization ~10% speedup