Speech enhancement with weighted denoising auto-encoder
نویسندگان
چکیده
A novel speech enhancement method with Weighted Denoising Auto-encoder (WDA) is proposed in this paper. A weighted reconstruction loss function is introduced to the conventional Denoising Auto-encoder (DA), and makes it suitable for the task of speech enhancement. First, the proposed WDA is used to model the relationship between the noisy and clean power spectrums of speech signal. Then, the estimated clean power spectrum is used in the a Posteriori SNR Controlled Recursive Averaging (PCRA) approach for the estimation of the a priori SNR. Finally, the enhanced speech is obtained by Wiener filter operating in the frequency domain. From the test results under ITU-T G.160, in comparison with the reference method, the proposed method could achieve similar amount of noise reduction in both white and colored noise, and the distortion on the level of speech signal is smaller. Also, the objective speech quality is improved in all the test conditions.
منابع مشابه
Perception Optimized Deep Denoising AutoEncoders for Speech Enhancement
Speech Enhancement is a challenging and important area of research due to the many applications that depend on improved signal quality. It is a pre-processing step of speech processing systems and used for perceptually improving quality of speech for humans. With recent advances in Deep Neural Networks (DNN), deep Denoising Auto-Encoders have proved to be very successful for speech enhancement....
متن کامل多樣訊雜比之訓練語料於降噪自動編碼器其語音強化功能之初步研究 (A Preliminary Study of Various SNR-level Training Data in the Denoising Auto-encoder (DAE) Technique for Speech Enhancement) [In Chinese]
Speech enhancement (SE) that reduces the noise effect plays an important role in the current widespread audio applications such as speech recognition, speech-based information retrieval and voice control. Among the various speech enhancement techniques, denoising auto-encoder (DAE) employs the well-known deep learning process to learn the transformation from noisy data to the respective clean n...
متن کاملDeep Denoising Auto-encoder for Statistical Speech Synthesis
This paper proposes a deep denoising auto-encoder technique to extract better acoustic features for speech synthesis. The technique allows us to automatically extract low-dimensional features from high dimensional spectral features in a non-linear, data-driven, unsupervised way. We compared the new stochastic feature extractor with conventional mel-cepstral analysis in analysis-by-synthesis and...
متن کاملThree ways to adapt a CTS recognizer to unseen reverberated speech in BUT system for the ASpIRE challenge
This paper describes several strategies tested in BUT’s submission to the IARPA ASpIRE challenge. The ASpIRE task was to develop an automatic speech recognition (ASR) system for wide-band noisy reverberant speech, while only clean CTS (Fisher) data was allowed for ASR training. To solve this task, we have started with augmenting Fisher data with artificially noised and reverberated versions. Th...
متن کاملRevisiting Denoising Auto-encoders
Denoising auto-encoders (DAE)s were proposed as a simple yet powerful way to obtain representations in an unsupervised manner by learning a map that approximates the clean inputs from their corrupted versions. However, the original objective function proposed for DAEs does not guarantee that denoising happens only at the encoding stages. We argue that a better representation can be obtained if ...
متن کامل