# A Real-Time Wideband Neural Vocoder at 1.6 kb/s Using LPCNet

## TL;DR

This paper introduces a low-bitrate neural vocoder based on LPCNet that operates in real-time, achieving higher speech quality than traditional codecs at 1.6 kb/s by leveraging linear prediction and sparse networks.

## Contribution

The paper presents a novel neural vocoder using LPCNet that combines linear prediction and sparsity to enable real-time low-bitrate speech synthesis.

## Key findings

- LPCNet at 1.6 kb/s outperforms MELP in quality.
- Uncompressed LPCNet exceeds low-bitrate waveform codec quality.
- Real-time operation achieved on general-purpose hardware.

## Abstract

Neural speech synthesis algorithms are a promising new approach for coding speech at very low bitrate. They have so far demonstrated quality that far exceeds traditional vocoders, at the cost of very high complexity. In this work, we present a low-bitrate neural vocoder based on the LPCNet model. The use of linear prediction and sparse recurrent networks makes it possible to achieve real-time operation on general-purpose hardware. We demonstrate that LPCNet operating at 1.6 kb/s achieves significantly higher quality than MELP and that uncompressed LPCNet can exceed the quality of a waveform codec operating at low bitrate. This opens the way for new codec designs based on neural synthesis models.