Pattern recognition of hypernasality in voice of patients with Cleft and Lip Palate

Nieto, Roger Gomez; Marin-Hurtado, Jorge I.; Capacho-Valbuena, Luis Miguel; Suarez, Alexander Amaya; Bolaños, Elkyn Alexander Belalcázar

doi:10.1109/stsiva.2014.7010187

Cited by 10 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Mel-frequency cepstral coefficients (MFCCs) and other spectral transformations [37], [38], [36], [39], [40], [41], [42], [43], [44], glottal source related features (jitter and shimmer) [45], [46], difference between the low-pass and bandpass profile of the Teager Energy Operator (TEO) [47], [48], and non-linear features [49], [50] have all been proposed as model input features. Gaussian mixture models (GMM), support vector machines, and deep neural networks have been used in conjunction with these features for hypernasality evaluation from word and sentence level data [51], [52], [53], [54]. Recently, end-to-end neural networks taking MFCC frames as input and producing hypernasality assessments as output have also been proposed [55].…”

Section: A Related Workmentioning

confidence: 99%

Robust Estimation of Hypernasality in Dysarthria with Acoustic Model Likelihood Features

Saxon,

Tripathi,

Jiao

et al. 2019

Preprint

View full text Add to dashboard Cite

Hypernasality is a common characteristic symptom across many motor-speech disorders. For voiced sounds, hypernasality introduces an additional resonance in the lower frequencies and, for unvoiced sounds, there is reduced articulatory precision due to air escaping through the nasal cavity. However, the acoustic manifestation of these symptoms is highly variable, making hypernasality estimation very challenging, both for human specialists and automated systems. Previous work in this area relies on either engineered features based on statistical signal processing or machine learning models trained on clinical ratings. Engineered features often fail to capture the complex acoustic patterns associated with hypernasality, whereas metrics based on machine learning are prone to overfitting to the small disease-specific speech datasets on which they are trained.Here we propose a new set of acoustic features that capture these complementary dimensions. The features are based on two acoustic models trained on a large corpus of healthy speech. The first acoustic model aims to measure nasal resonance from voiced sounds, whereas the second acoustic model aims to measure articulatory imprecision from unvoiced sounds. To demonstrate that the features derived from these acoustic models are specific to hypernasal speech, we evaluate them across different dysarthria corpora. Our results show that the features generalize even when training on hypernasal speech from one disease and evaluating on hypernasal speech from another disease (e.g. training on Parkinson's disease, evaluation on Huntington's disease), and when training on neurologically disordered speech but evaluating on cleft palate speech.

show abstract

Section: A Related Workmentioning

confidence: 99%

Robust Estimation of Hypernasality in Dysarthria with Acoustic Model Likelihood Features

Saxon,

Tripathi,

Jiao

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…En la Figura 1 se muestra un esquema general de clasificación de audio con AP, donde la entrada de la red neuronal es una representación espectral del audio, en este caso un espectrograma calculado con la transformada de Fourier de término reducido (STFT por su sigla en inglés) sobre el audio multiplicado previamente por una ventana (windowing). En el contexto de este proyecto, se busca que la parte de la red que realiza la extracción de características transmita a las capas posteriores información relativa a ciertas medicas acústicas que se saben relacionadas con la calidad vocal, como shimmer, jitter y harmonics-to-noise ratio (HNR) [5][6][7].…”

Section: Materiales Y Métodosunclassified

Clasificación Automática de la Calidad Vocal

García

Destéfanis²

2019

AJEA

View full text Add to dashboard Cite

Se presenta un enfoque para la construcción de un clasificador extremo-a-extremo de la calidad vocal en escala GRBAS basado en redes neuronales profundas. En base a este enfoque se muestran tres redes neuronales. Las redes presentadas calculan la transformada de Fourier de término reducido (STFT), el cepstrum y shimmer de una señal de audio. Las redes neuronales que calculan la STFT y shimmer se logran entrenar correctamente, mientras que la que calcula el cepstrum no. Para este último caso, se plantea una solución alternativa al cepstrum, la autocovariance, que sí se puede entrenar. Se concluye que las redes neuronales desarrolladas son compatibles con el enfoque planteado porque permiten que el gradiente del error se propague hacia atrás, condición necesaria para entrenar el modelo completo.

show abstract

“…Shimmer value is associated to voice quality [2][3][4][5][6][7], state of mind [8][9][10][11][12][13], age [14] and gender [15] of people. There are many research works that use shimmer (among other measures) with goals ranging from pathologies detection [6,16,17] to the improvement of human-machine interfaces through the estimation of the intensionality of a spoken phrase [19].…”

Section: Introductionmentioning

confidence: 99%

Deep Neural Networks for Shimmer Approximation in Synthesized Audio Signal

García

Destéfanis

2018

Communications in Computer and Information Science

View full text Add to dashboard Cite

Shimmer is a classical acoustic measure of the amplitude perturbation of a signal. This kind of variation in the human voice allow to characterize some properties, not only of the voice itself, but of the person who speaks. During the last years deep learning techniques have become the state of the art for recognition tasks on the voice. In this work the relationship between shimmer and deep neural networks is analyzed. A deep learning model is created. It is able to approximate shimmer value of a simple synthesized audio signal (stationary and without formants) taking the spectrogram as input feature. It is concluded firstly, that for this kind of synthesized signal, a neural network like the one we proposed can approximate shimmer, and secondly, that the convolution layers can be designed in order to preserve the information of shimmer and transmit it to the following layers.

show abstract

Pattern recognition of hypernasality in voice of patients with Cleft and Lip Palate

Cited by 10 publications

References 6 publications

Robust Estimation of Hypernasality in Dysarthria with Acoustic Model Likelihood Features

Robust Estimation of Hypernasality in Dysarthria with Acoustic Model Likelihood Features

Clasificación Automática de la Calidad Vocal

Deep Neural Networks for Shimmer Approximation in Synthesized Audio Signal

Contact Info

Product

Resources

About