Pith. sign in

TalkNet: Fully-Convolutional Non-Autoregressive Speech Synthesis Model

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
abstract

We propose TalkNet, a convolutional non-autoregressive neural model for speech synthesis. The model consists of two 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 generates a mel-spectrogram from the expanded text. To train a grapheme duration predictor, we add the grapheme duration to the training dataset using a pre-trained Connectionist Temporal Classification (CTC)-based speech recognition model. The explicit duration prediction eliminates word skipping and repeating. Experiments on the LJSpeech dataset show that the speech quality nearly matches auto-regressive models. The model is very compact -- it has 10.8M parameters, almost 3x less than the present state-of-the-art text-to-speech models. The non-autoregressive architecture allows for fast training and inference.

citation-role summary

background 1

citation-polarity summary

fields

cs.CV 1

years

2025 1

verdicts

CONDITIONAL 1

roles

background 1

polarities

unclear 1

representative citing papers

Multi-human Interactive Talking Dataset

cs.CV · 2025-08-05 · conditional · novelty 6.0

The paper contributes a 12-hour multi-person conversational video dataset with pose and speaking annotations, plus a baseline model for generating full-body talking videos of two to four people.

citing papers explorer

Showing 1 of 1 citing paper.

  • Multi-human Interactive Talking Dataset cs.CV · 2025-08-05 · conditional · none · ref 3 · internal anchor

    The paper contributes a 12-hour multi-person conversational video dataset with pose and speaking annotations, plus a baseline model for generating full-body talking videos of two to four people.