End-to-end LPCNet: A Neural Vocoder With Fully-Differentiable LPC Estimation

TL;DR

This paper introduces an end-to-end neural vocoder that integrates LPC coefficient inference into the model, eliminating the need for explicit LP analysis and enabling flexible conditioning features while maintaining low computational complexity.

Contribution

It presents a novel end-to-end LPCNet model that learns to infer LP coefficients directly from input features, simplifying the pipeline and broadening applicability.

Findings

End-to-end LPCNet matches or exceeds original LPCNet quality.

The model operates without explicit LP analysis.

It supports various conditioning features.

Abstract

Neural vocoders have recently demonstrated high quality speech synthesis, but typically require a high computational complexity. LPCNet was proposed as a way to reduce the complexity of neural synthesis by using linear prediction (LP) to assist an autoregressive model. At inference time, LPCNet relies on the LP coefficients being explicitly computed from the input acoustic features. That makes the design of LPCNet-based systems more complicated, while adding the constraint that the input features must represent a clean speech spectrum. We propose an end-to-end version of LPCNet that lifts these limitations by learning to infer the LP coefficients from the input features in the frame rate network. Results show that the proposed end-to-end approach equals or exceeds the quality of the original LPCNet model, but without explicit LP analysis. Our open-source end-to-end model still benefits from LPCNet's low complexity, while allowing for any type of conditioning features.