Single-channel speech enhancement using implicit Wiener filter for high-quality speech communication

Traditional speech enhancement algorithms have been widely studied due to simplicity and convenience, especially spectral subtraction algorithm. However, traditional speech enhancement algorithms always utilize the available noisy speech phase spectrum as the optimal estimation of clean speech phase spectrum, which is inaccurate at low signal-noise ratio (SNR). Aiming at improving performance of typical spectral subtraction speech enhancement algorithm under the circumstances of low SNR, a multiband spectral subtraction speech enhancement algorithm with phase spectrum compensation is proposed in this paper. The proposed method compensates noisy speech phase spectrum based on the SNR and estimation of noise obtained in the current frame and reduces musical noise generated by spectral subtraction. Simulation results show that the proposed algorithm has a better performance than traditional algorithms, especially at low SNR.

The spectral subtraction method is a conventional approach for single channel speech enhancement. The basic principle of this method is to estimate the short-time spectral magnitude of speech by subtracting estimated noise from the noisy speech spectrum and to combines it with the phase of the noisy speech. Besides reducing the noise, this method generates an unnatural and unpleasant noise, called remnant noise. This paper proposes a novel algorithm to reduce the remnant noise, and thus improving the overall quality of the enhanced speech. In this algorithm, the output of multi-band spectral subtraction (MBSS) method is used as the input signal again for next iteration process. After the MBSS method, the additive noise is changed to remnant noise. The remnant noise is re-estimated at each iteration. The new estimated noise, furthermore, is been used to process the next MBSS. This procedure is iterated a small number of times. The simulation results as well as informal subjective evaluations prove that the speech enhanced by proposed algorithm is more pleasant to listeners than the conventional MBSS algorithm. This reveals that the proposed algorithm reduces remnant noise satisfactorily and produces good speech quality with improved signal-to-noise ratio.

Combined speech enhancement and auditory modelling for robust distributed speech recognition

Refinement of Enhanced Speech Using Hybrid-Median Filter and Harmonic Regeneration Approach

Speech enhancement with Geometric Advent of Spectral subtraction using connected time-frequency regions noise estimation aims to de-noise or reduce background noise from the noisy speech for better quality, pleasantness and improved intelligibility. Numerous enhancement methods are proposed including spectral subtraction, subspace, statistical with different noise estimations. The traditional spectral subtraction techniques are reasonably simple to implement and suffer from musical noise. This study addresses the new approach for speech enhancement which has minimized the insufficiencies in traditional spectral subtraction algorithms using MCRA. This approach with noise estimation has been evolved with PESQ, the ITU-T standard; Frequency weighted segmental SNR and weighted spectral slope. The analysis shows that Geometric approach with time-frequency connected regions has improved results than old-fashioned spectral subtraction algorithms. The normal hearing tests has suggested that new approach has lower audible musical noise.

Monaural speech segregation from complex concurrent noise is an extremely challenging problem; binary mask is a method to solve this problem, however, the performance of binary mask is limited by remaining the noise in the result. In this paper, an algorithm integrated Spectral Subtraction and binary masking for speech separation and enhancement was proposed. It follows the framework of computational auditory scene analysis (CASA). The energy of time-frequency (T-F) unit was used as the clue to generate the binary mask; then the spectral subtraction algorithm was used to eliminate noise energy in original speech and an interim speech was obtained, after covered the binary mask on the interim speech, the target speech can be achieved. Systematic evaluation shows that the combined algorithm can stably improve the SNR and voice quality for noisy speech. It performs better than existing binary masking systems in most situations, especially when the noise and the speech have the similar power spectrum.

Many studies on deep learning-based speech enhancement (SE) utilizing the computational auditory scene analysis method typically employs the ideal binary mask or the ideal ratio mask to reconstruct the enhanced speech signal. However, many SE applications in real scenarios demand a desirable balance between denoising capability and computational cost. In this study, first, an improvement over the ideal ratio mask to attain more superior SE performance is proposed through introducing an efficient adaptive correlation-based factor for adjusting the ratio mask. The proposed method exploits the correlation coefficients among the noisy speech, noise and clean speech to effectively re-distribute the power ratio of the speech and noise during the ratio mask construction phase. Second, to make the supervised SE system more computationally-efficient, quantization techniques are considered to reduce the number of bits needed to represent floating numbers, leading to a more compact SE model. The proposed quantized correlation mask is utilized in conjunction with a 4-layer deep neural network (DNN-QCM) comprising dropout regulation, pre-training and noise-aware training to derive a robust and high-order mapping in enhancement, and to improve generalization capability in unseen conditions. Results show that the quantized correlation mask outperforms the conventional ratio mask representation and the other SE algorithms used for comparison. When compared to a DNN with ideal ratio mask as its learning targets, the DNN-QCM provided an improvement of approximately 6.5% in the short-time objective intelligibility score and 11.0% in the perceptual evaluation of speech quality score. The introduction of the quantization method can reduce the neural network weights to a 5-bit representation from a 32-bit, while effectively suppressing stationary and non-stationary noise. Timing analyses also show that with the techniques incorporated in the proposed DNN-QCM system to increase its compactness, the training and inference time can be reduced by 15.7% and 10.5%, respectively.

The estimation of the fundamental frequency (F0) in speech is a very important task that has been addressed by many researchers. F0 estimation can be used to separate two kind of frames from an utterance, those where the vocal folds vibrate (voiced sounds) and those where not (unvoiced sounds). The methods used to estimate F0 are affected by the presence of additive noise in recordings made in non-controlled environments, however, there are different techniques to mitigate the effect of such noise and Speech Enhancement (SE) has proven to be one of the most effective ones. This article presents results of the evaluation of the effects of noise and SE algorithms on the detection of F0 and the signal segmentation in voiced/unvoiced segments. We performed experiments with signals artificially contaminated with two different kinds of noise, White Gaussian Noise (WGN) and background noise recorded at a cafeteria (Cafeteria babble), subsequently, the signals are processed with SE algorithms of four different classes: Wiener Filter, Spectral Subtraction, Statistical-Model Based and Sub-space algorithms. Two different kind of error metrics are considered: Gross Pitch Error and Voicing Determination Error. The results show that only the sub-space approach improves the performance in the detection of F0 and the signal segmentation in voiced/unvoicd segments.

Automatic separation of speech from noise and segregation of overlapping co-channel speech are two of the most challenging problems in speech processing. In previous work, we have developed algorithms to both enhance noisy speech and segregate overlapping speech streams. Our single-channel speech enhancement and speech segregation algorithms have shown better performance than other reported algorithms for automatic speech recognition. Additionally, objective evaluation scores of perceptual quality have shown a significant improvement following processing by our algorithms. In this study, we focus on subjective evaluation of these algorithms for human listeners. We investigate the intelligibility and quality of speech from our algorithm on normal-hearing listeners, cochlear implant users, and hearing impaired subjects. Our preliminary results indicate a significant improvement in the perceptual quality of the speech signal after being processed by our algorithm and suggest that the proposed algorithms can be used as a pre-processing block within the signal processing in hearing aid devices.

We conducted an investigation to explore how neurotypical (NT) listeners perceive the emotional tone of voice in sentences spoken by individuals with high-functioning autism spectrum disorders (ASD) and NT speakers. The investigation included both male and female speakers from both groups. In Study 1, NT listeners were asked to identify the emotional prosody (anger, fear, happiness, surprise or neutral) conveyed by the speakers. Results revealed that emotional expressions produced by male ASD speakers were generally less accurately recognized compared to male NT speakers. In contrast, emotions expressed by female ASD speakers were more accurately categorized compared to female NT speakers, except when expressing fear. This suggests that female ASD speakers may not express emotional prosody in the same way as their male counterparts. In Study 2, a subset of produced materials was rated for valence, voice modulation, and voice control to supplement Study 1 results: Female ASD speakers sounded less negative when expressing fear compared to female NT speakers. Male ASD speakers were perceived as less positive than NT speakers when expressing happiness. Voice modulation also differed between groups, showing a tendency for ASD speakers to follow different display rules for both positive emotions (happiness and surprise) tested. Finally, male ASD speakers were rated to use voice cues less appropriately compared to NT male speakers, an effect less pronounced for female ASD speakers. Together, the results imply that difficulties in social interactions among individuals with high-functioning ASD could be due to non-prototypical voice use of male ASD speakers and emphasize that female individuals do not show the same effects.

Noise in media is any undesirable signal that masks relevant information content. The addition of noise to real-world data in any context is practically inevitable. Noise reduction algorithms in the past have addressed the problem but lacked adaptability to various real-world applications while also being time and resource extensive. Spectral subtraction provides a hybrid approach to noise reduction that incorporates versatility and efficient resource usage. This research tested the performance of two spectral subtraction noise reduction algorithms (stationary and non-stationary) across five categories of real-world noise (speech only, speech with natural noise, music, animal sounds, and noise only). The research question under study was how stationary and non-stationary spectral subtraction algorithms differ in their noise reduction performance when subjected to the various categories of noise. The testing was done based on normalized cross-correlation, which is the similarity between the noise-reduced audio and the original recording in each case. Non-stationary spectral subtraction performed better in samples where human speech was the target: speech only and speech with natural noise. Stationary spectral subtraction performed better when denoising music and animal sounds. This anomaly in performance between the two algorithms was only noted in categories with no human speech. These results exemplify the performance and versatility of different spectral subtraction algorithms. The category-specific results can be used to employ specific spectral subtraction algorithms for specific tasks for optimum performance.

In this paper, a hidden Markov model (HMM)-based cue parameters estimation method for single-channel speech enhancement is proposed, in which the cue parameters of binaural cue coding (BCC) are applied to single-channel speech enhancement system successfully. First, the clean speech and noise signals are considered as the left and right channels of stereo signal, respectively; and the noisy speech is treated as the down-mixed mono signal of BCC method. According to the clean speech and noise data set and the corresponding noisy speech data set, the clean cue parameters and pre-enhanced cue parameters are extracted, respectively. Then the cue HMM is trained offline, which exploits the a priori information about the clean cue parameters and the pre-enhanced cue parameters for speech enhancement. Next, using the trained cue HMM, the clean cue parameters are estimated from noisy speech online. Finally, following the synthesis principle of BCC cue parameters, the speech estimator is constructed for enhancing noisy speech. The test results demonstrate that, for the segmental signal-noise-ratio (SNR), the log spectral distortion and PESQ measures, the proposed method performs better than the reference methods.

Conventional multiband speech enhancement involves splitting the spectrum into various frequency bins and performing speech enhancement in each band independently. However, owing to the pole‐interaction problem in the spectral domain, estimation of clean speech from the formants, suppressed by the influence of the formants in the neighbouring bands, may result in poor quality. To reduce the influence of stronger formants over the neighbouring bands, in the current work, clean speech is estimated by filtering unprocessed speech in the temporal domain into various equivalent rectangular bandwidth based subbands followed by discrete cosine transform (DCT) based spectral speech enhancement in each band using spectral subtraction/minimum mean square error (MMSE). To further enhance speech, a spectral subtraction‐based approach that incorporates band‐specific weighting factor obtained using respective band signal‐to‐noise ratio (SNR), and an MMSE estimator that calculates apriori speech presence/absence probability based on local and global apriori SNR rather than a fixed/equiprobable value are proposed. The performance of the algorithms is evaluated using perceptual evaluation of speech quality and composite speech quality measure. It is observed that DCT‐derived spectrum based temporal‐domain multiband speech enhancement algorithm outperforms the existing techniques for car, babble, train, white, and factory noise in the 0–10 dB SNR levels.

Speech enhancement and improved recognition accuracy by integrating wavelet transform and spectral subtraction algorithm

In this study, the morpheme of the dialogues in Korean textbooks for married immigrants was analyzed and the nouns in the dialogues were examined from the gender perspective. By comparing the dialogues between the pre-revised textbook 『Korean language for Married immigrant』 and the post-revised textbook 『Joyful Korean Language』 and 『Accurate Korean Language』, it was confirmed that there was a positive change between the two textbooks, but there were still parts that needed improvement. The speech ratio of male and female speakers improved, but the speech amount of female speakers still accounted for a large proportion. By analyzing the subject of the dialogue through the overall upper-frequency noun, the subject of the dialogue in the textbook after the revision was diversified. As a result of comparing the upper-frequency nouns according to gender, the bias of the conversation topic disappeared. However, the conversation on the subject of 'food' has become more biased than in the past, which can lead to stereotypes about gender roles, so it is suggested that it needs to be carefully reviewed and corrected. In addition, when noun vocabulary with markedness used by male and female speakers was listed in the order of upper-frequency, male speakers used more advanced words than female speakers in both textbooks before and after the revision. This seems to be due to the establishment of the social status and role of male and female speakers, and it was suggested that improvement was needed in that unequal perceptions were reflected in the setting of the social roles of male and female speakers.