Audio de-noising by spectral subtraction technique implemented on reconfigurable hardware

Biswas, Tanmay; Pal, Chandrajit; Mandal, Sudhindu Bikash; Chakrabarti, Amlan

doi:10.1109/ic3.2014.6897179

Cited by 7 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DWT is the most popular technique for improving speech, which can estimate parameters like SNR and MSE [11]. On the other hand, in a Spectral Subtraction lter, the noise spectrum can be estimated and then subtracted from speech, and a clean signal can be preserved [12]. So, these two methods are useful Speech Enhancement methods for improving the SNR value for noisy speech signals in hearing aids.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Discrete Wavelet Transform and Spectral Subtraction Based Speech Enhancement Algorithm for Hearing Aid Application

Iqbal,

Zhang,

Geng

et al. 2024

Preprint

View full text Add to dashboard Cite

Hearing aids are small electronic devices intended to help those with hearing loss improve their hearing ability with the use of advanced audio signal processing techniques and technologies. Usually, in hearing aids, a set of speech enhancement methods are utilized to improve speech signal quality in low signal-to-noise ratio (SNR) environments. In speech processing, Discrete Wavelet Transform (DWT) algorithm and Spectral Subtraction (SS) filter are some of the most commonly used methods today when handling background noise in hearing aids. However, these speech enhancement systems have drawbacks like in DWT, the selective thresholding face problems when applying to distinct types of noise with different frequencies and time scales, and spectral subtraction has a problem with music noise, which affects the denoising performance. In this paper, the Spectral Subtraction (SS) filter and DWT speech enhancement methods are combined to deal with problem faced by DWT selective thresholding techniques against noise of different frequencies and time scales and the musical noise faced by the spectral subtraction filter. We first used spectral subtraction to reduce average noise intensity and then added DWT thresholding for reducing the background noise further. The contemporary methods evaluated using English speech signals taken from CSTR VCTK Corpus database. The objective metrics used to evaluate proposed speech enhancement systems are MSE, SNR, PESQ, and STOI. The proposed speech enhancement algorithm is applied for noise reduction and then compared with conventional algorithms. Experimental results prove that using of the proposed speech enhancement algorithm reduces background noise, improves the SNR of an original speech signal, and improves the final noise reduction performance.

show abstract

Section: Introductionmentioning

confidence: 99%

“…: Analysis of DWT-BSN based on evolution metrics in an audio clip is sampled and then subtracted from the entire frequency recording domain is the process of Spectral Subtraction (SS) technique[12,14,41]. Using SS as a speech enhancement technique and producing higher-quality speech data with improved SNR is the primary goal of our proposed work.…”

mentioning

confidence: 99%

Discrete Wavelet Transform and Spectral Subtraction Based Speech Enhancement Algorithm for Hearing Aid Application

Iqbal,

Zhang,

Geng

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…One of the earliest forms of denoising speech systems was spectral subtraction, which estimated the noise spectrum from the signal and subtracted it from the noisy signal to create clean speech [3]. Spectral subtraction was effective at reducing noise to an extent but introduces additional speech artifacts [4]. Another method of denoising speech are Wiener filter algorithms, which filter out any noise to provide an estimate of the clean signal.…”

Section: Introductionmentioning

confidence: 99%

Denoising Speech Signals with Hifi-Coulomb-GANs

Satheesh¹,

Muthu-Manivannan²

2022

J Stud Res

View full text Add to dashboard Cite

Recorded speech signals often contain noise that affects the quality of the signal and reduces intelligibility. Several studies have used Generative Adversarial Networks (GANs) to remove noise artifacts and improve speech intelligibility. However, GANs can suffer from gradient vanishing or gradient explosion that can reduce their effectiveness in denoising. To mitigate gradient vanishing, we applied the CoulombGAN architecture to speech denoising using a model structure similar to Hifi-GAN, the current state of the art speech denoiser. We call this new model Hifi-CoGAN. We used a WaveNet generator to denoise signals, a PostNet for general cleanup, and a Multi-Resolution Discriminator to evaluate the signal quality relative to the clean signal. Our results show that Hifi-CoGAN was able to outperform Hifi-GAN in many of the narrowband signals (signals with a limited range of frequencies) in terms of the Short-Term Objective Intelligibility (STOI) and Perceptual Evaluation of Speech Quality (PESQ) metrics. However, the model did not perform as well as Hifi-GAN with wideband noise signals (signals with a wider range of frequencies) such as white noise, so future work must be done to improve the model for these noise signals.

show abstract

“…Spectral subtraction technique based audio de-noising is explained in [4]. An efficient architecture in hardware for the algorithm of spectral subtraction is applied to speech improvement is used.…”

Section: Introductionmentioning

confidence: 99%

Audio Denoising Based on Short Time Fourier Transform

Ashwin

Manoharan

2018

IJEECS

View full text Add to dashboard Cite

<p>This paper presents a novel audio de-noising scheme in a given speech signal. The recovery of original from the communication channel without any noise is a difficult task. Many de-noising techniques have been proposed for the removal of noises from a digital signal. In this paper, an audio de-noising technique based on Short Time Fourier Transform (STFT) is implemented. The proposed architecture uses a novel approach to estimate environmental noise from speech adaptively. Here original speech signals are given as input signal. Using AWGN, noises are added to the signal. Then noised signals are de-noised using STFT techniques. Finally Signal to Noise Ratio (SNR), Peak Signal to Noise Ratio (PSNR) values for noised and de-noised signals are obtained.</p>

show abstract

Audio de-noising by spectral subtraction technique implemented on reconfigurable hardware

Cited by 7 publications

References 8 publications

Discrete Wavelet Transform and Spectral Subtraction Based Speech Enhancement Algorithm for Hearing Aid Application

Discrete Wavelet Transform and Spectral Subtraction Based Speech Enhancement Algorithm for Hearing Aid Application

Denoising Speech Signals with Hifi-Coulomb-GANs

Audio Denoising Based on Short Time Fourier Transform

Contact Info

Product

Resources

About