DeepAI
Log In Sign Up

PoCoNet: Better Speech Enhancement with Frequency-Positional Embeddings, Semi-Supervised Conversational Data, and Biased Loss

08/11/2020
by   Umut Isik, et al.
16

Neural network applications generally benefit from larger-sized models, but for current speech enhancement models, larger scale networks often suffer from decreased robustness to the variety of real-world use cases beyond what is encountered in training data. We introduce several innovations that lead to better large neural networks for speech enhancement. The novel PoCoNet architecture is a convolutional neural network that, with the use of frequency-positional embeddings, is able to more efficiently build frequency-dependent features in the early layers. A semi-supervised method helps increase the amount of conversational training data by pre-enhancing noisy datasets, improving performance on real recordings. A new loss function biased towards preserving speech quality helps the optimization better match human perceptual opinions on speech quality. Ablation experiments and objective and human opinion metrics show the benefits of the proposed improvements.

READ FULL TEXT

page 1

page 2

page 3

page 4

09/11/2021

Incorporating Real-world Noisy Speech in Neural-network-based Speech Enhancement Systems

Supervised speech enhancement relies on parallel databases of degraded s...
08/24/2020

AMRConvNet: AMR-Coded Speech Enhancement Using Convolutional Neural Networks

Speech is converted to digital signals using speech coding for efficient...
11/16/2021

Unsupervised Speech Enhancement with speech recognition embedding and disentanglement losses

Speech enhancement has recently achieved great success with various deep...
11/08/2020

Frequency Gating: Improved Convolutional Neural Networks for Speech Enhancement in the Time-Frequency Domain

One of the strengths of traditional convolutional neural networks (CNNs)...
12/09/2021

A Training Framework for Stereo-Aware Speech Enhancement using Deep Neural Networks

Deep learning-based speech enhancement has shown unprecedented performan...
02/16/2018

Constrained Convolutional-Recurrent Networks to Improve Speech Quality with Low Impact on Recognition Accuracy

For a speech-enhancement algorithm, it is highly desirable to simultaneo...
04/07/2022

Leveraging Real Conversational Data for Multi-Channel Continuous Speech Separation

Existing multi-channel continuous speech separation (CSS) models are hea...