{"ID":80569,"CreatedAt":"2026-02-27T13:00:40Z","UpdatedAt":"2026-02-27T13:00:40Z","DeletedAt":null,"paper_url":"https://paperswithcode.com/paper/a-convolutional-neural-network-model-based-on","arxiv_id":"1901.10629","title":"A Convolutional Neural Network model based on Neutrosophy for Noisy Speech Recognition","abstract":"Convolutional neural networks are sensitive to unknown noisy condition in the\ntest phase and so their performance degrades for the noisy data classification\ntask including noisy speech recognition. In this research, a new convolutional\nneural network (CNN) model with data uncertainty handling; referred as NCNN\n(Neutrosophic Convolutional Neural Network); is proposed for classification\ntask. Here, speech signals are used as input data and their noise is modeled as\nuncertainty. In this task, using speech spectrogram, a definition of\nuncertainty is proposed in neutrosophic (NS) domain. Uncertainty is computed\nfor each Time-frequency point of speech spectrogram as like a pixel. Therefore,\nuncertainty matrix with the same size of spectrogram is created in NS domain.\nIn the next step, a two parallel paths CNN classification model is proposed.\nSpeech spectrogram is used as input of the first path and uncertainty matrix\nfor the second path. The outputs of two paths are combined to compute the final\noutput of the classifier. To show the effectiveness of the proposed method, it\nhas been compared with conventional CNN on the isolated words of Aurora2\ndataset. The proposed method achieves the average accuracy of 85.96 in noisy\ntrain data. It is more robust against Car, Airport and Subway noises with\naccuracies 90, 88 and 81 in test sets A, B and C, respectively. Results show\nthat the proposed method outperforms conventional CNN with the improvement of\n6, 5 and 2 percentage in test set A, test set B and test sets C, respectively.\nIt means that the proposed method is more robust against noisy data and handle\nthese data effectively.","url_abs":"http://arxiv.org/abs/1901.10629v2","url_pdf":"http://arxiv.org/pdf/1901.10629v2.pdf","authors":"[\"Elyas Rashno\", \"Ahmad Akbari\", \"Babak Nasersharif\"]","published":"2019-01-27T00:00:00Z","tasks":"[\"General Classification\", \"Noisy Speech Recognition\", \"speech-recognition\", \"Speech Recognition\"]","methods":"[]","has_code":false}