GAN-Based Speech Enhancement for Low SNR Using Latent Feature Conditioning

Enhancing speech quality under adverse SNR conditions remains a significant challenge for discriminative deep neural network (DNN)-based approaches. In this work, we propose DisCoGAN, which is a time-frequency-domain generative adversarial network (GAN) conditioned by the latent features of a discriminative model pre-trained for speech enhancement in low SNR scenarios. Our proposed method achieves superior performance compared to state-of-the-arts discriminative methods and also surpasses end-to-end (E2E) trained GAN models. We also investigate the impact of various configurations for conditioning the proposed GAN model with the discriminative model and assess their influence on enhancing speech quality

Comparative Analysis Of Discriminative Deep Learning-Based Noise Reduction Methods In Low SNR Scenarios

In this study, we conduct a comparative analysis of deep learning-based noise reduction methods in low signal-to-noise ratio (SNR) scenarios. Our investigation primarily focuses on five key aspects: The impact of training data, the influence of various loss functions, the effectiveness of direct and indirect speech estimation techniques, the efficacy of masking, mapping, and deep filtering methodologies, and the exploration of different model capacities on noise reduction performance and speech quality. Through comprehensive experimentation, we provide insights into the strengths, weaknesses, and applicability of these methods in low SNR environments. The findings derived from our analysis are intended to assist both researchers and practitioners in selecting better techniques tailored to their specific applications within the domain of low SNR noise reduction.

Blind Acoustic Parameter Estimation Through Task-Agnostic Embeddings Using Latent Approximations

We present a method for blind acoustic parameter estimation from single-channel reverberant speech. The method is structured into three stages. In the first stage, a variational auto-encoder is trained to extract latent representations of acoustic impulse responses represented as mel-spectrograms. In the second stage, a separate speech encoder is trained to estimate low-dimensional representations from short segments of reverberant speech. Finally, the pre-trained speech encoder is combined with a small regression model and evaluated on two parameter regression tasks. Experimentally, the proposed method is shown to outperform a fully end-to-end trained baseline model.

On Improving Error Resilience of Neural End-to-End Speech Coders

Error resilient tools like Packet Loss Concealment (PLC) and Forward Error Correction (FEC) are essential to maintain a reliable speech communication for applications like Voice over Internet Protocol (VoIP), where packets are frequently delayed and lost. In recent times, end-to-end neural speech codecs have seen a significant rise, due to their ability to transmit speech signal at low bitrates but few considerations were made about their error resilience in a real system. Recently introduced Neural End-to-End Speech Codec (NESC) can reproduce high quality natural speech at low bitrates. We extend its robustness to packet losses by adding a low complexity network to predict the codebook indices in latent space. Furthermore, we propose a method to add an in-band FEC at an additional bitrate of 0.8 kbps. Both subjective and objective assessment indicate the effectiveness of proposed methods, and demonstrate that coupling PLC and FEC provide significant robustness against packet losses.

Simple and Efficient Quantization Techniques for Neural Speech Coding

Neural audio coding has emerged as a vivid research direction by promising good audio quality at very low bitrates unachievable by classical coding techniques. Here, end-to-end trainable autoencoder-like models represent the state of the art, where a discrete representation in the bottleneck of the autoencoder has to be learned that allows for efficient transmission of the input audio signal. This discrete representation is typically generated by applying a quantizer to the output of the neural encoder. In almost all state-of-the-art neural audio coding approaches, this quantizer is realized as a Vector Quantizer (VQ) and a lot of effort has been spent to alleviate drawbacks of this quantization technique when used together with a neural audio coder. In this paper, we propose simple alternatives to VQ, which are based on projected Scalar Quantization (SQ). These quantization techniques do not need any additional losses, scheduling parameters or codebook storage thereby simplifying the training of neural audio codecs. Furthermore, we propose a new causal network architecture for neural speech coding that shows good performance at very low computational complexity.

End-To-End Deep Learning-based Adaptation Control for Linear Acoustic Echo Cancellation

The attenuation of acoustic loudspeaker echoes remains to be one of the open challenges to achieve pleasant full-duplex hands free speech communication. In many modern signal enhancement interfaces, this problem is addressed by a linear acoustic echo canceler which subtracts a loudspeaker echo estimate from the recorded microphone signal. To obtain precise echo estimates, the parameters of the echo canceler, i.e., the filter coefficients, need to be estimated quickly and precisely from the observed loudspeaker and microphone signals. For this a sophisticated adaptation control is required to deal with high-power double-talk and rapidly track time-varying acoustic environments which are often faced with portable devices. In this paper, we address this problem by end-to-end deep learning. In particular, we suggest to infer the step-size for a least mean squares frequency-domain adaptive filter update by a Deep Neural Network (DNN). Two different step-size inference approaches are investigated. On the one hand broadband approaches, which use a single DNN to jointly infer step-sizes for all frequency bands, and on the other hand narrowband methods, which exploit individual DNNs per frequency band. The discussion of benefits and disadvantages of both approaches leads to a novel hybrid approach which shows improved echo cancellation while requiring only small DNN architectures. Furthermore, we investigate the effect of different loss functions, signal feature vectors, and DNN output layer architectures on the echo cancellation performance from which we obtain valuable insights into the general design and functionality of DNN-based adaptation control algorithms.

A Unifying View on Blind Source Separation of Convolutive Mixtures based on Independent Component Analysis

In many daily-life scenarios, acoustic sources recorded in an enclosure can only be observed with other interfering sources. Hence, convolutive Blind Source Separation (BSS) is a central problem in audio signal processing. Methods based on Independent Component Analysis (ICA) are especially important in this field as they require only few and weak assumptions and allow for blindness regarding the original source signals and the acoustic propagation path. Most of the currently used algorithms belong to one of the following three families: Frequency Domain ICA (FD-ICA), Independent Vector Analysis (IVA), and TRIple-N Independent component analysis for CONvolutive mixtures (TRINICON). While the relation between ICA, FD-ICA and IVA becomes apparent due to their construction, the relation to TRINICON is not well established yet. This paper fills this gap by providing an in-depth treatment of the common building blocks of these algorithms and their differences, and thus provides a common framework for all considered algorithms.

Microphone Utility Estimation in Acoustic Sensor Networks using Single-Channel Signal Features

In multichannel signal processing with distributed sensors, choosing the optimal subset of observed sensor signals to be exploited is crucial in order to maximize algorithmic performance and reduce computational load, ideally both at the same time. In the acoustic domain, signal cross-correlation is a natural choice to quantify the usefulness of microphone signals, i. e., microphone utility, for array processing, but its estimation requires that the uncoded signals are synchronized and transmitted between nodes. In resource-constrained environments like acoustic sensor networks, low data transmission rates often make transmission of all observed signals to the centralized location infeasible, thus discouraging direct estimation of signal cross-correlation. Instead, we employ characteristic features of the recorded signals to estimate the usefulness of individual microphone signals. In this contribution, we provide a comprehensive analysis of model-based microphone utility estimation approaches that use signal features and, as an alternative, also propose machine learning-based estimation methods that identify optimal sensor signal utility features. The performance of both approaches is validated experimentally using both simulated and recorded acoustic data, comprising a variety of realistic and practically relevant acoustic scenarios including moving and static sources.

Manifold learning-supported estimation of relative transfer functions for spatial filtering

Many spatial filtering algorithms used for voice capture in, e.g., teleconferencing applications, can benefit from or even rely on knowledge of Relative Transfer Functions (RTFs). Accordingly, many RTF estimators have been proposed which, however, suffer from performance degradation under acoustically adverse conditions or need prior knowledge on the properties of the interfering sources. While state-of-the-art RTF estimators ignore prior knowledge about the acoustic enclosure, audio signal processing algorithms for teleconferencing equipment are often operating in the same or at least a similar acoustic enclosure, e.g., a car or an office, such that training data can be collected. In this contribution, we use such data to train Variational Autoencoders (VAEs) in an unsupervised manner and apply the trained VAEs to enhance imprecise RTF estimates. Furthermore, a hybrid between classic RTF estimation and the trained VAE is investigated. Comprehensive experiments with real-world data confirm the efficacy for the proposed method.

End-To-End Deep Learning-Based Adaptation Control for Frequency-Domain Adaptive System Identification

We present a novel end-to-end deep learning-based adaptation control algorithm for frequency-domain adaptive system identification. The proposed method exploits a deep neural network to map observed signal features to corresponding step-sizes which control the filter adaptation. The parameters of the network are optimized in an end-to-end fashion by minimizing the average system distance of the adaptive filter. This avoids the need of explicit signal power spectral density estimation as required for model-based adaptation control and further auxiliary mechanisms to deal with model inaccuracies. The proposed algorithm achieves fast convergence and robust steady-state performance for scenarios characterized by non-white and non-stationary noise signals, time-varying environment changes and additional model inaccuracies.