Talker age estimation using machine learning

Abstract

As a person ages, the acoustic characteristics of their voice change. Understanding how the sound of a voice changes with age may give insight into physiological changes related to vocal function. Previous work has shown changes in acoustical parameters with chronological age as well as differences between perceived age and chronological age. However, much of this previous work was done using cross-sectional speech samples, which will show changes with age but may average out important individual variability with regard to aging differences. The current study used a longitudinal recording sample gathered from a corpus of speeches from an individual spanning about 50 years (48 to 97 years of age). This study investigates how the voice changes with age using both chronological age and perceived age as independent variables; perceived age data were obtained in a previous direct age estimation study. Using the longitudinal recordings, a range of voice and speech acoustic parameters were extracted. These acoustic parameters were fitted to a supervised learning model to predict chronological age and perceived age. Differences between the chronological age and perceived age models as well as the usefulness of the various acoustic parameters will be discussed.

Machine-Learning Analysis of Voice Samples Recorded through Smartphones: The Combined Effect of Ageing and Gender

“…Indeed, by comparing females included in the YA and OA groups as well as males included in the YA and OA groups, in separate analyses, we have examined the pure effect of ageing on voice. Our findings fully agree with previous reports demonstrating the effect of ageing on the human voice [ 24 , 25 , 26 , 27 , 28 , 33 , 34 , 35 , 36 , 37 , 38 ]. Early studies based on the qualitative/perceptual evaluation of voice recordings have demonstrated that physiologic ageing leads to several changes in specific characteristics of the human voice [ 1 ].…”

“…In our study, by applying the ROC curve analysis, we demonstrated in detail the high accuracy of our machine learning analysis in demonstrating age-related changes in the human voice. Our results fit in well with previous studies applying automatic classifiers based on machine learning analysis [ 24 , 25 , 26 , 27 , 28 , 33 , 34 , 35 , 36 , 37 , 38 ]. More in detail, our machine learning algorithm has achieved higher results than those obtained on the INTERSPEECH 2010 age and gender sub-challenge feature set [ 33 , 34 ].…”

“…The advantages of applying those algorithms consist of obtaining smaller dataset of features (only 20 features in our study), easier math handled and with shorter computation time. Moreover, all the previous studies considered only MFCC-, f0-, pitch-, energy-, jitter-, and shimmer-related features [ 24 , 25 , 26 , 27 , 28 , 33 , 34 , 35 , 36 , 37 ], with only a study considering non-traditional features including RASTA-PLP coefficients [ 38 ]. In addition to the traditional frequency-, jitter-, shimmer-, energy-, spectral, and cepstral-related features, we have also included MFCC and RASTA-PLP coefficients and three additional representative features (HNR, SNR, and CPPs).…”

“…The most recent approach used to assess physiologic ageing in healthy subjects consists of the objective voice analysis based on machine learning algorithms [ 24 , 25 , 26 , 27 , 28 ]. Machine-learning is a novel and robust method commonly applied to classify complex variables obtained from large datasets [ 29 , 30 , 31 ].…”

“…More in detail, machine learning can be applied to predict outcomes from recurring patterns of features within various types of multidimensional data sets [ 32 ]. Several authors have applied automatic classifiers based on machine learning analysis on voice recordings to classify healthy subjects according to their age and gender [ 24 , 25 , 26 , 27 , 28 , 33 , 34 , 35 , 36 , 37 , 38 ]. More recently, to further improve the overall accuracy of the machine learning analysis, several studies have included an increasing number of voice features in the datasets [ 24 , 25 , 26 , 27 , 28 ] and compared the performance of different machine learning algorithms [ 37 , 38 ].…”