A Melody-Unsupervision Model for Singing Voice Synthesis
This addresses a data preparation bottleneck for researchers and developers in singing voice synthesis, though it is incremental as it builds on existing text-to-speech models.
The paper tackles the problem of requiring time-consuming manual temporal alignment of melody and lyric labels for training singing voice synthesis models by proposing a melody-unsupervision model that only needs audio-and-lyrics pairs without alignment during training, and it generates singing voice audio from melody and lyrics inputs, showing capability to be trained with speech audio and text labels.
Recent studies in singing voice synthesis have achieved high-quality results leveraging advances in text-to-speech models based on deep neural networks. One of the main issues in training singing voice synthesis models is that they require melody and lyric labels to be temporally aligned with audio data. The temporal alignment is a time-exhausting manual work in preparing for the training data. To address the issue, we propose a melody-unsupervision model that requires only audio-and-lyrics pairs without temporal alignment in training time but generates singing voice audio given a melody and lyrics input in inference time. The proposed model is composed of a phoneme classifier and a singing voice generator jointly trained in an end-to-end manner. The model can be fine-tuned by adjusting the amount of supervision with temporally aligned melody labels. Through experiments in melody-unsupervision and semi-supervision settings, we compare the audio quality of synthesized singing voice. We also show that the proposed model is capable of being trained with speech audio and text labels but can generate singing voice in inference time.