Controllable Neural Text-to-Speech Synthesis Using Intuitive Prosodic Features

Tuomo Raitio, Ramya Rasipuram, Dan Castellani


Modern neural text-to-speech (TTS) synthesis can generate speech that is indistinguishable from natural speech. However, the prosody of generated utterances often represents the average prosodic style of the database instead of having wide prosodic variation. Moreover, the generated prosody is solely defined by the input text, which does not allow for different styles for the same sentence. In this work, we train a sequence-to-sequence neural network conditioned on acoustic speech features to learn a latent prosody space with intuitive and meaningful dimensions. Experiments show that a model conditioned on sentence-wise pitch, pitch range, phone duration, energy, and spectral tilt can effectively control each prosodic dimension and generate a wide variety of speaking styles, while maintaining similar mean opinion score (4.23) to our Tacotron baseline (4.26).


 DOI: 10.21437/Interspeech.2020-2861

Cite as: Raitio, T., Rasipuram, R., Castellani, D. (2020) Controllable Neural Text-to-Speech Synthesis Using Intuitive Prosodic Features. Proc. Interspeech 2020, 4432-4436, DOI: 10.21437/Interspeech.2020-2861.


@inproceedings{Raitio2020,
  author={Tuomo Raitio and Ramya Rasipuram and Dan Castellani},
  title={{Controllable Neural Text-to-Speech Synthesis Using Intuitive Prosodic Features}},
  year=2020,
  booktitle={Proc. Interspeech 2020},
  pages={4432--4436},
  doi={10.21437/Interspeech.2020-2861},
  url={http://dx.doi.org/10.21437/Interspeech.2020-2861}
}