Automatic speech recognition (ASR) based approach for speech therapy of aphasic patients: A review

Intelligent speech technologies for transcription, disease diagnosis, and medical equipment interactive control in smart hospitals: A review

“…(1) Use IST to recognize the doctors' voices and reduce their time spent in non-medical related work, which was studied by researchers from an early stage [ 62 , 63 ]. (2) IST is also utilized to process the patients’ speech signals to assist doctors in diagnosing and evaluating diseases [ 16 ]. This application has made significant breakthroughs in recent years with the development of ML and is also a hotspot of current research [ 64 ].…”

“…There are many review articles on speech technologies in medical applications, such as medical reporting [ 13 ], clinical documentation [ 14 ], speech impairment assessment [ 15 ], and speech therapy [ 16 ], healthcare [ 10 , 17 ]. However, we still require the review of state-of-the-art IST applications in smart hospitals.…”

Voice Communication in Noisy Environments in a Smart House Using Hybrid LMS+ICA Algorithm

“…Speech contains information that is usually obtained by processing a speech signal captured by a microphone using sampling, quantization, coding [ 38 ], parametrization, preprocessing, segmentation, centring, pre-emphasis, and window weighting [ 39 , 40 ]. The next step is speech recognition with statistical approach for continuous speech recognition [ 41 ] with different approaches [ 42 ] for speech recognition system’s [ 43 ] using the perceptual linear prediction (PLP) of speech [ 44 ], for example, Audio-to-Visual Conversion in Mpeg-4 [ 45 ], acoustic modeling and feature extraction [ 46 ], speech activity detectors [ 47 ] or joint training of hybrid neural networks for acoustic modeling in automatic speech recognition [ 48 ], the RASTA method (RelAtive SpecTrAl) [ 38 ], and the Mel-frequency cepstral analysis (MFCC), for example, dimensionality reduction of a pathological voice quality assessment system [ 49 ], content-based clinical depression detection in adolescents [ 50 ], speech recognition in an intelligent wheelchair [ 51 ], speech recognition by using the from speech signals of spoken words [ 52 ], the hidden Markov models (HMM) [ 53 ], and artificial neural networks (ANN) [ 54 ], for example, feed-forward Neural Network (NN) with back propagation algorithm and a Radial Basis Functions Neural Networks [ 55 ], an automatic speech recognition (ASR) based approach for speech therapy of aphasic patients [ 56 ], fast adaptation of deep neural network based on discriminant codes for speech recognition [ 57 ], implementation of dnn-hmm acoustic models for phoneme recognition [ 58 ], combination of features in a hybrid HMM/MLP and a HMM/GMM speech recognition system [ 59 ], and hybrid continuous speech recognition systems by HMM, MLP and SVM [ ...…”

“…artificial neural networks (ANN) [ 54 ], for example, feed-forward Neural Network (NN) with back propagation algorithm and a Radial Basis Functions Neural Networks [ 55 ], an automatic speech recognition (ASR) based approach for speech therapy of aphasic patients [ 56 ], fast adaptation of deep neural network based on discriminant codes for speech recognition [ 57 ], implementation of dnn-hmm acoustic models for phoneme recognition [ 58 ], combination of features in a hybrid HMM/MLP and a HMM/GMM speech recognition system [ 59 ], and hybrid continuous speech recognition systems by HMM, MLP and SVM [ 60 ]. …”

Improving Aphasic Speech Recognition by Using Novel Semi-Supervised Learning Methods on AphasiaBank for English and Spanish

“…Nonetheless, there are still challenges related to automatic speech recognition (ASR) that must be solved worldwide in order to extend these therapy applications, since they basically depend on adequate engines that should properly recognize aphasic speech. ASR systems are usually trained with the voices of people without any speech pathology, and their performance degrades when they are applied to aphasic speech [23][24][25][26][27]. Furthermore, ASR systems are usually language-dependent and have to be trained with hundreds or thousands of hours of transcribed speech.…”