We wrote yesterday about speech as a potential marker for obstructive sleep apnea, one possible reason for disordered sleep. There is further research on the speech sequellae of disordered sleep of any etiology. One study, from a bioengineering perspective, appeared last July (Moon J, et al., Annu Int Conf IEEE Eng Med Biol Soc. 2022 Jul;2022:1981-1984. doi: 10.1109/EMBC48229.2022.9870900), on how prolonged sleepiness appears to be detectable generally, through automated analysis of acoustic patterns of speech, which are different from those of normal speech. This change is independent of the language being spoken. To date, there have been no studies exactly like it, examining linguistic-independent sleepy speech detection.
In this study, two different languages, English and German, were trialed, where the former was used to train and validate and the latter to test the effectiveness of machine and deep learning models. Specifically, the team trained ResNet50, a deep learning model, and five machine learning models with relevant vocal features. Speech data segments from three English-speaking subjects were used for training the model and segments from an English-speaking subject were used for validation. ResNet50 and the five different machine-learning models were then tested against speech data segments from one German-speaking subject. Deep learning far outperformed all of the machine learning approaches. The accuracy, sensitivity, specificity, and geometric mean values were found to be 0.96, 0.92, 0.99, and 0.95, respectively, using ResNet50 on the test data.
Preliminary results therefore suggest that sleepiness can be accurately detected independently from linguistic speech.
MyoNews from BreatheWorksTM is a report on trends and developments in oromyofunctional disorder and therapy. These updates are not intended as diagnosis, treatment, cure or prevention of any disease or syndrome.