Intelligibility Of Noisy Speech Research Articles

Single channel speech enhancement using time-frequency attention mechanism based nested U-net model

Deep-learning models have used attention mechanisms to improve quality and intelligibility of noisy speech, demonstrating the effectiveness of attention mechanisms. We rely on either spatial or temporal-based attention mechanisms, resulting in severe information loss. In this paper, a time-frequency attention mechanism with a nested U-network (TFANUNet) is proposed for single-channel speech enhancement. By using TFA, learns which channel, frequency and time information is more significant for speech enhancement. Basically, the proposed model is an encoder-decoder model, where each layer in the encoder and decoder is followed by a nested dense residual dilated DensNet (NDRD) based multi-scale context aggression block. NDRD involves multiple dilated convolutions with different dilatation factors to explore the large receptive area at different scales simultaneously. NDRD avoids the aliasing problem in DenseNet. We integrated the TFA and NDRD blocks into the proposed model to enable refined feature set extraction without information loss and utterance-level context aggregation, respectively. Under seen and unseen noise conditions, the proposed TFAD3MNet model produces an average of 87.02% and 85.04% of STOI values, and 3.19 and 3.01 averaged PESQ values. The trainable parameters of proposed model are 2.09 million,which is very less compared to baselines. TFANUNet model results outperform baselines in terms of STOI and PESQ.

Performance of single-channel speech enhancement algorithms on Mandarin listeners with different immersion conditions in New Zealand English

Speech enhancement (SE) is a widely used technology to improve the quality and intelligibility of noisy speech. So far, SE algorithms were designed and evaluated on native listeners only, but not on non-native listeners who are known to be more disadvantaged when listening in noisy environments. This paper investigates the performance of five widely used single-channel SE algorithms on early-immersed New Zealand English (NZE) listeners and native Mandarin listeners with different immersion conditions in NZE under negative input signal-to-noise ratio (SNR) by conducting a subjective listening test in NZE sentences. The performance of the SE algorithms in terms of speech intelligibility in the three participant groups was investigated. The result showed that the early-immersed group always achieved the highest intelligibility. The late-immersed group outperformed the non-immersed group for higher input SNR conditions, possibly due to the increasing familiarity with the NZE accent, whereas this advantage disappeared at the lowest tested input SNR conditions. The SE algorithms tested in this study failed to improve and rather degraded the speech intelligibility, indicating that these SE algorithms may not be able to reduce the perception gap between early-, late- and non-immersed listeners, nor able to improve the speech intelligibility under negative input SNR in general. These findings have implications for the future development of SE algorithms tailored to Mandarin listeners, and for understanding the impact of language immersion on speech perception in noise.

Speech Enhancement: A Review of Different Deep Learning Methods

Speech enhancement methods differ depending on the degree of degradation and noise in the speech signal, so research in the field is still difficult, especially when dealing with residual and background noise, which is highly transient. Numerous deep learning networks have been developed that provide promising results for improving the perceptual quality and intelligibility of noisy speech. Innovation and research in speech enhancement have been opened up by the power of deep learning techniques with implications across a wide range of real time applications. By reviewing the important datasets, feature extraction methods, deep learning models, training algorithms and evaluation metrics for speech enhancement, this paper provides a comprehensive overview. We begin by tracing the evolution of speech enhancement research, from early approaches to recent advances in deep learning architectures. By analyzing and comparing the approaches to solving speech enhancement challenges, we categorize them according to their strengths and weaknesses. Moreover, we discuss the challenges and future directions of deep learning in speech enhancement, including the demand for parameter-efficient models for speech enhancement. The purpose of this paper is to examine the development of the field, compare and contrast different approaches, and highlight future directions as well as challenges for further research.

Multivariate fMRI responses in superior temporal cortex predict visual contributions to, and individual differences in, the intelligibility of noisy speech

Humans have the unique ability to decode the rapid stream of language elements that constitute speech, even when it is contaminated by noise. Two reliable observations about noisy speech perception are that seeing the face of the talker improves intelligibility and the existence of individual differences in the ability to perceive noisy speech. We introduce a multivariate BOLD fMRI measure that explains both observations. In two independent fMRI studies, clear and noisy speech was presented in visual, auditory and audiovisual formats to thirty-seven participants who rated intelligibility. An event-related design was used to sort noisy speech trials by their intelligibility. Individual-differences multidimensional scaling was applied to fMRI response patterns in superior temporal cortex and the dissimilarity between responses to clear speech and noisy (but intelligible) speech was measured. Neural dissimilarity was less for audiovisual speech than auditory-only speech, corresponding to the greater intelligibility of noisy audiovisual speech. Dissimilarity was less in participants with better noisy speech perception, corresponding to individual differences. These relationships held for both single word and entire sentence stimuli, suggesting that they were driven by intelligibility rather than the specific stimuli tested. A neural measure of perceptual intelligibility may aid in the development of strategies for helping those with impaired speech perception.

Comparison of ideal mask-based speech enhancement algorithms for speech mixed with white noise at low mixture signal-to-noise ratios

The literature shows that the intelligibility of noisy speech can be improved by applying an ideal binary or soft gain mask in the time-frequency domain for signal-to-noise ratios (SNRs) between -10 and +10 dB. In this study, two mask-based algorithms are compared when applied to speech mixed with white Gaussian noise (WGN) at lower SNRs, that is, SNRs from -29 to -5 dB. These comprise an Ideal Binary Mask (IBM) with a Local Criterion (LC) set to 0 dB and an Ideal Ratio Mask (IRM). The performance of three intrusive Short-Time Objective Intelligibility (STOI) variants-STOI, STOI+, and Extended Short-Time Objective Intelligibility (ESTOI)-is compared with that of other monaural intelligibility metrics that can be used before and after mask-based processing. The results show that IRMs can be used to obtain near maximal speech intelligibility (>90% for sentence material) even at very low mixture SNRs, while IBMs with LC = 0 provide limited intelligibility gains for SNR < -14 dB. It is also shown that, unlike STOI, STOI+ and ESTOI are suitable metrics for speech mixed with WGN at low SNRs and processed by IBMs with LC = 0 even when speech is high-pass filtered to flatten the spectral tilt before masking.

End-to-End Deep Convolutional Recurrent Models for Noise Robust Waveform Speech Enhancement.

Because of their simple design structure, end-to-end deep learning (E2E-DL) models have gained a lot of attention for speech enhancement. A number of DL models have achieved excellent results in eliminating the background noise and enhancing the quality as well as the intelligibility of noisy speech. Designing resource-efficient and compact models during real-time processing is still a key challenge. In order to enhance the accomplishment of E2E models, the sequential and local characteristics of speech signal should be efficiently taken into consideration while modeling. In this paper, we present resource-efficient and compact neural models for end-to-end noise-robust waveform-based speech enhancement. Combining the Convolutional Encode-Decoder (CED) and Recurrent Neural Networks (RNNs) in the Convolutional Recurrent Network (CRN) framework, we have aimed at different speech enhancement systems. Different noise types and speakers are used to train and test the proposed models. With LibriSpeech and the DEMAND dataset, the experiments show that the proposed models lead to improved quality and intelligibility with fewer trainable parameters, notably reduced model complexity, and inference time than existing recurrent and convolutional models. The quality and intelligibility are improved by 31.61% and 17.18% over the noisy speech. We further performed cross corpus analysis to demonstrate the generalization of the proposed E2E SE models across different speech datasets.

SINGLE CHANNEL SPEECH ENHANCEMENT USING EVOLUTIONARY ALGORITHM WITH LOG-MMSE

Additive noise degrades speech quality and intelligibility. Speech enhancement reduces this noise to make speech more pleasant and intelligible. It plays a significant role in speech recognition or speech-operated systems. In this paper, we propose a single-channel speech enhancement method in which the log-minimum mean square error method (log-MMSE) and modified accelerated particle swarm optimization algorithm are used to design a filter for improving the quality and intelligibility of noisy speech. Accelerated particle swarm optimization (APSO) algorithm is modified in which a single dimension of particle position is changed in a single iteration while obtaining the particle’s new position. Using this algorithm, a filter is designed with multiple passbands and notches for speech enhancement. The modified algorithm converges faster compared with standard particle swarm optimization algorithm (PSO) and APSO giving optimum filter coefficients. The designed filter is used to enhance the speech. The proposed speech enhancement method improves the perceptual estimation of speech quality (PESQ) by 17.05% for 5dB babble noise, 33.92 % for 5dB car noise, 14.96 % for 5dB airport noise, and 39.13 % for 5dB exhibition noise. The average output PESQ for these four types of noise is improved compared to conventional methods of speech enhancement. There is an average of 7.58 dB improvement in segmental SNR for these noise types. The proposed method improves speech intelligibility with minimum speech distortion.

Fusing Bone-conduction and Air-conduction Sensors for Complex-Domain Speech Enhancement.

Speech enhancement aims to improve the listening quality and intelligibility of noisy speech in adverse environments. It proves to be challenging to perform speech enhancement in very low signal-to-noise ratio (SNR) conditions. Conventional speech enhancement utilizes air-conduction (AC) microphones, which are sensitive to background noise but capable of capturing full-band signals. On the other hand, bone-conduction (BC) sensors are unaffected by acoustic noise, but recorded speech has limited bandwidth. This study proposes an attention-based fusion method to combine the strengths of AC and BC signals and perform complex spectral mapping for speech enhancement. Experiments on the EMSB dataset demonstrate that the proposed approach effectively leverages the advantages of AC and BC sensors, and outperforms a recent time-domain baseline in all conditions. We also show that the sensor fusion method is superior to single-sensor counterparts, especially in low SNR conditions. As the amount of BC data is very limited, we additionally propose a semi-supervised technique to utilize both parallelly and unparallely recorded AC and BC speech signals. With additional AC speech from the AISHELL-1 dataset, we achieve similar performance to supervised learning with only 50% parallel data.

Subjective Evaluation of a Noise-Reduced Training Target for Deep Neural Network-Based Speech Enhancement

Speech enhancement systems aim to improve the quality and intelligibility of noisy speech. In this study, we compare two speech enhancement systems based on deep neural networks. The speech intelligibility and quality of both systems were evaluated subjectively by a speech recognition test based on Hagerman sentences and a translation of the ITU-T P.835 recommendation, respectively. Results were compared with the objective measures STOI and POLQA. Neither STOI nor POLQA reliably predicted subjective results. While STOI anticipated improvement, subjective results for both models showed degradation of speech intelligibility. POLQA results were overall hardly affected, while the subjective results showed significant changes in overall quality, both positive and negative, in many of the tests. One of the systems was trained to remove all noise; a strategy that is common in speech enhancement systems found in the literature. The other system was trained to only reduce the noise such that the signal-to-noise ratio increased with 10 dB. The latter system subjectively outperformed the system that attempted to remove noise completely. From this, we conclude that objective evaluation cannot replace subjective evaluation until a measure that reliably predicts intelligibility and quality for deep-neural-network-based systems has been identified. Results further indicate that it may be beneficial to move away from more aggressive noise removal strategies toward noise reduction strategies that cause less speech distortion.