Deep Learning Approaches for Automatic Sung Speech Recognition
Adapting Spoken Technologies to Sung Speech
This website contains a number of audio samples that complement the content of my doctoral thesis.
In my thesis, I evaluated whether new DNN techniques designed for spoken speech ASR and audio source separation can be adapted to work for sung speech recognition.
Roadmap:
- Abstract: The abstract of the thesis
- Chapters/Chapter 2: Includes YouTube links to the mishearing samples presented in Section 2.2.
- Chapters/Chapter 4: Includes six recordings from the DSing evaluation set.
- Chapters/Chapter 5: Includes the recordings use to generate the different figures of the chapter.
- Chapters/Chapter 6: Includes recognition results from mismatched and adapted ASR model.
- Publications: The list of publications.