TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction
arXiv:2104.08189
Abstract
We propose TalkNet, a non-autoregressive convolutional neural model for speech synthesis with explicit pitch and duration prediction. The model consists of three feed-forward convolutional networks. The first network predicts grapheme durations. An input text is expanded by repeating each symbol according to the predicted duration. The second network predicts pitch value for every mel frame. The third network generates a mel-spectrogram from the expanded text conditioned on predicted pitch. All networks are based on 1D depth-wise separable convolutional architecture. The explicit duration prediction eliminates word skipping and repeating. The quality of the generated speech nearly matches the best auto-regressive models - TalkNet trained on the LJSpeech dataset got MOS 4.08. The model has only 13.2M parameters, almost 2x less than the present state-of-the-art text-to-speech models. The non-autoregressive architecture allows for fast training and inference. The small model size and fast inference make the TalkNet an attractive candidate for embedded speech synthesis.
arXiv commentarXiv admin note: substantial text overlap with arXiv:2005.05514
References in corpus (8)
- Mixed Precision Training
- HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis
- Deep Voice: Real-time Neural Text-to-Speech
- NeMo: a toolkit for building AI applications using Neural Modules
- Non-Attentive Tacotron: Robust and Controllable Neural TTS Synthesis Including Unsupervised Duration Modeling
- Neural Speech Synthesis with Transformer Network
- Speech Synthesis with Neural Networks
- A High Quality Text-To-Speech System Composed of Multiple Neural Networks