Yi Liu (刘艺), Ph.D.

Beijing, China
Email: liu-yi15@tsinghua.org.cn


I'm now a software engineer working on speech processing in ByteDance AI Lab. My research interests include speaker recognition and diarization, speech recognition, and other speech/audio processing technologies. I am also the reviewer of INTERSPEECH and ICASSP.
I received my Ph.D. degree in Tsinghua University under the supervision of Prof. Jia Liu in 2020, with the dissertation Research on Speaker Embedding Extraction Methods based on Deep Learning. I received my M.E. degree in Tsinghua University in 2018 and the B.E. degree in Wuhan University in 2012. I was a visiting student in Cambridge University Engineering Department from July to December 2018.



  1. Yi Liu, Liang He, Jia Liu, Michael T. Johnson. "Introducing phonetic information to speaker embedding for speaker verification." EURASIP Journal on Audio, Speech, and Music Processing. 2019.


  1. Yi Liu, Liang He, Jia Liu. "Large Margin Softmax Loss for Speaker Verification." Interspeech, 2019. [PDF][code]
  2. Yi Liu, Liang He, Weiwei Liu, Jia Liu. "Exploring a Unified Attention-Based Pooling Framework for Speaker Verification." ISCSLP, 2018. [PDF]
  3. Yi Liu, Liang He, Jia Liu, Michael T. Johnson. "Speaker Embedding Extraction with Phonetic Information." INTERSPEECH, 2018. [PDF] [code]
  4. Yi Liu, Liang He, Wei-Qiang Zhang, Jia Liu, Michael T. Johnson. "Investigation of Frame Alignments for GMM-based Digit-prompted Speaker Verification." APSIPA ASC 2018.[PDF]
  5. Yi Liu, Liang He, Yao Tian, Zhuzi Chen, Jia Liu, Michael T. Johnson. "Comparison of Multiple Features and Modeling Methods for Text-dependent Speaker Verification." ASRU, 2017.[PDF]
  6. Yi Liu, Yao Tian, Liang He, Jia Liu. "Investigating Various Diarization Algorithms for Speaker in the Wild (SITW) Speaker Recognition Challenge." INTERSPEECH, 2016.[PDF]
  7. Yi Liu, Yao Tian, Liang He, Jia Liu, Michael T. Johnson. "Simultaneous Utilization of Spectral Magnitude and Phase Information to Extract Supervectors for Speaker Verification Anti-spoofing." INTERSPEECH, 2015.[PDF]
  8. Yi Liu, Liang He, and Jia Liu. "Improved multitaper PNCC feature for robust speaker verification." International Symposium on Chinese Spoken Language Processing (ISCSLP), 2014. [PDF]
  9. Xianhong Chen, Liang He, Can Xu, Yi Liu, Tianyu Liang and Jia Liu. "VB-HMM Speaker Diarization with Enhanced and Refined Segment Representation." Odyssey, 2018.
  10. Liang He, Yao Tian, Yi Liu, Fang Dong, WeiQiang Zhang, Jia Liu, "A study of variational method for text-independent speaker recognition." ISCSLP, 2016.[PDF]
  11. Liang He, Yao Tian, Yi Liu, Jiaming Xu, Weiwei Liu, Cai Meng, Jia Liu. "THU-EE system description for NIST LRE 2015." INTERSPEECH, 2016. [PDF]
  12. Yao Tian, Liang He, Yi Liu, Jia Liu. "Investigation of Senone-based Long-Short Term Memory RNNs for Spoken Language Recognition." Odyssey, 2016. [PDF]


  • Speaker Embedding Extraction with Phonetic Information (based on Kaldi) [github]
  • Neural speaker recognition/verification system using Kaldi and Tensorflow [github]


    2016.8 - 2017.5 Intern researcher at Sogou. Working on speaker recognition system.
    2015.6 - 2015.9 Intern researcher at Big-data innovotion center of CreditEase. I developed an i-vector-based speaker diarization algorithm for telephony recordings.

    Technical Blogs (in Chinese)

    1. 知乎:CNN(卷积神经网络)、RNN(循环神经网络)、DNN(深度神经网络)的内部网络结构有什么区别?
    2. 知乎:为什么 Deep Learning 最先在语音识别和图像处理领域取得突破?
    3. 知乎:未来语音技术或者语音智能助手的发展方向是什么?

    My CV is available Here

    Follow me on GitHub, LinkedIn and Google Scholar