Abstract

This paper describes Tacotron 2, a neural network architecture for speech synthesis directly from text. The system is composed of a recurrent sequence-to-sequence feature prediction network that maps character embeddings to mel-scale spectrograms, followed by a modified WaveNet model acting as a vocoder to synthesize time-domain waveforms from those spectrograms. Our model achieves a mean opinion score (MOS) of 4.53 comparable to a MOS of 4.58 for professionally recorded speech. To validate our design choices, we present ablation studies of key components of our system and evaluate the impact of using mel spectrograms as the conditioning input to WaveNet instead of linguistic, duration, and F0 features. We further show that using this compact acoustic intermediate representation allows for a significant reduction in the size of the WaveNet architecture.

Keywords

SpectrogramComputer scienceSpeech recognitionWaveformCharacter (mathematics)Speech synthesisFeature (linguistics)Artificial neural networkRepresentation (politics)Key (lock)Artificial intelligencePattern recognition (psychology)MathematicsTelecommunicationsLinguistics

Affiliated Institutions

Related Publications

Publication Info

Year
2018
Type
article
Pages
4779-4783
Citations
2462
Access
Closed

External Links

Download PDF (Free) View on DOI.org arXiv Semantic Scholar

Social Impact

Altmetric
PlumX Metrics

Social media, news, blog, policy document mentions

Citation Metrics

2462
OpenAlex
384
Influential
1507
CrossRef

Cite This

Jonathan Shen, Ruoming Pang, Ron J. Weiss et al. (2018). Natural TTS Synthesis by Conditioning Wavenet on MEL Spectrogram Predictions. 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , 4779-4783. https://doi.org/10.1109/icassp.2018.8461368

Identifiers

DOI
10.1109/icassp.2018.8461368
arXiv
1712.05884

Data Quality

Data completeness: 84%