Sound Noise Reduction Based on Deep Neural Networks
Keywords:
De-Noising, Noise Removal Of Audio Signal, Long Short Term Memory (LSTM), Recurrent Neural Network.Abstract
Audio transmittance is a generation that is now rapidly growing as a connectivity option for everyone around the world, demanding to experience the frictionless transfer of audio messages. Audio transmittance has a wide range of capabilities compared to other connectivity technologies. But we are living in the noisy world, hence while transmitting audio signal; we don’t only transmit audio, different types of noise gets transmitted with our audio signal as well which will lead to an unclear communication The basic purpose of this model is specifically focused on detecting and restoring noisy audio signals which consists various background noise. The removal of noise from the audio signal will enhance the information carrying capacity of the signal during audio communication. For the removal of noise from audio signal, a stacked Long Short Term Memory (LSTM) model is proposed. ‘Edinburgh DataShare’ dataset has been used to train the model. During the evaluation of model, the Huber loss of 0.0205 has been evaluated in 50 epochs which shows that the LSTM network was successfully implemented for noise removal of audio signal. Hence on the basis of result, we can conclude that that Stacked LSTM network works well in noise removal of audio signals
References
- S. Boll, “Suppression of acoustic noise in speech using spectral subtraction,” IEEE Transactions on acoustics, speech, and signal processing, vol. 27, no. 2, pp. 113–120, 1979.
- H. G. Hirsch and C. Ehrlicher, “Noise estimation techniques for robust speech recognition,” in Proc. ICASSP, 1995, vol.1, pp. 153–156.
- T. Gerkmann and R.C. Hendriks, “Unbiased MMSE-based noise power estimation with low complexity and low tracking delay,” IEEE Transactions on Audio, Speech, and Language Processing, vol. 20, no. 4, pp. 1383–1393, 2012.
- Y. Ephraim and D. Malah, “Speech enhancement using a minimum mean-square error log-spectral amplitude estimator,” IEEE Transactions on Acoustics, Speech, and Signal Processing, vol. 33, no. 2, pp. 443–445, 1985.
- A. Maas, Q.V. Le, T.M. O’Neil, O. Vinyals, P. Nguyen, and A.Y. Ng, “Recurrent neural networks for noise reduction in robust ASR,” in Proc.INTERSPEECH, 2012.
- D. Liu, P. Smaragdis, and M. Kim, “Experiments on deep learning for speech denoising,” in Proc. Fifteenth Annual Conference of the International Speech Communication Association, 2014.
- Y. Xu, J. Du, L.-R. Dai, and C.-H. Lee, “A regression approach to speech enhancement based on deep neural networks,” IEEE Transactions on Audio, Speech and Language Processing, vol. 23, no. 1, pp. 7–19, 2015.
- A. Narayanan and D. Wang, “Ideal ratio mask estimation using deep neural networks for robust speech recognition,” in Proc. ICASSP, 2013, pp. 7092–7096.
- S. Mirsamadi and I. Tashev, “Causal speech enhancement combining data-driven learning and suppression rule estimation.,” in Proc. INTERSPEECH, 2016, pp. 2870–2874.
- https://grubbrr.com/the-importance-of-backgroundnoise-suppression-in-ai/
- https://developer.nvidia.com/blog/nvidia-real-timenoise-suppression-deep-learning/
- Jean-Marc Valin, “ A Hybrid DSP / Deep Learning Approach to Real-Time Full-Band Speech Enhancement,” arXiv:1709.08243v3 [cs.SD] 31 May 2018
- Urmila Shrawankar and Vilas Thakare, “Noise Estimation and Noise Removal Techniques for Speech Recognition in Adverse Environment,”.
Downloads
Published
Issue
Section
License
Copyright (c) IJSRSET
This work is licensed under a Creative Commons Attribution 4.0 International License.