This model, presented in a paper in Mobile Networks and Applications, was trained to recognize emotions in human
"Multi-information model of the joint algorithmdecision making is created through emotion recognition,” wrote Han Tian, Zhang Zhu, and Xu Jing in their paper. "The model is being used to analyze representative data about subjects and to help diagnose depression in subjects."
Tian and his colleagues trained their model to setDAIC-WOZ data, a set of audio and 3D facial expressions of patients diagnosed with a depressive disorder and people without depression. These audio recordings and facial expressions were collected during interviews conducted by a virtual agent who asked various questions about the mood and life of the interviewee.
“Based on the study of the speech characteristics of people withdepressive disorder, this article provides an in-depth study of diagnosing depression using speech based on speech data from the DAIC-WOZ dataset, Tian, Zhu, and Jian wrote in their study. - Firstly, speech information is pre-processed, including speech pre-emphasis, framing, endpoint detection, denoising, etc. Secondly, OpenSmile is used to extract characteristics of speech signals, and speech characteristics that can reflect functions are studied and analyzed in depth.
To extract important features from voicerecords, the team's model uses OpenSmile (open source speech and music interpretation by large space extraction). It is a set of tools often used by computer scientists to extract features from audio clips and classify those clips.
The researchers used this tool toextraction of individual features of speech and their combinations, which are usually found in the speech of patients diagnosed with depression. Subsequently, they used a technique known as Principal Component Analysis to reduce the set of extracted features.
Tian, Zhu and Jian rated their model in the seriestests in which they assessed her ability to detect depressed and non-depressed people from their voice recordings. Their scheme produced remarkable results, detecting depression with an accuracy of 87% in male patients and 87.5% in female patients.
In the future, deep learning algorithm,developed by this group of researchers may become an additional auxiliary tool for psychiatrists and physicians along with other well-established diagnostic tools. In addition, this research may inspire the development of similar AI tools for detecting signs of mental disorders based on speech.
Read more:
Two photos of the Earth with a difference of 50 years were compared at NASA: what scientists found
Scientists transplanted human “brain” into rats and told what happened in the end
ChatGPT interviewed at Google for $183,000 engineer