To improve suicide prevention strategies, Concordia PhD student Alaa Nfissi has spearheaded a remarkable AI tool model for Speech Emotion Recognition (SER), using Artificial Intelligence tools. The next-generation model was introduced at IEEE’s 2024 18th International Conference on Semantic Computing and has elevated the capabilities of mental health interventions to new heights.
Automated Speech Emotion Recognition (SER)
Earlier on the suicide hotline counselors use speech patterns as their main indicator of callers’ emotional states. On the other hand, this procedure is slow and can be subject to human mistakes. Nfissi’s advanced deep-learning model automates this procedure by quickly picking up necessary emotional signals from callers’ voices.
Nfissi’s model uses a diverse database of actual hotline calls that encompasses a range of emotions, employing neural networks as well as gated recurrent units that analyze the speech waveforms. Unlike the previous model, which was only capable of uniform segment lengths, Nfissi’s model adapts to variable lengths, boosting its accuracy and flexibility.
Unrivalled accuracy and instant intervention
Nfissi’s model accurately detects main states of emotions such as fear, sadness, anger, and neutrality. Proven results illustrate the effectiveness of the model, it has confidently located the emotions in over 70 % of cases, with an especially high percentage of successful cases in the professionally recorded segments.
This discovery resonates beyond what is perceived in purely academic circles. Nfissi designs a dashboard that, in real-time, provides the hotline counselors the ability to act swiftly and effectively during crisis intervention. Through the use of this technology, such an instant emotional insight is promised thus, preparing to change the suicide prevention scheme for good and, hence, saving many lives.
A personal mission: Overcoming the gaps in mental health support.
For Nfissi, this research is very close to the heart. Having accordingly explored the details of suicide hotline intervention, he realizes the expediency of rendering efficient support to those in crisis. With his method, he intends to fill the vacancies in counselor training and simplify the intervention process so that every person who calls can receive the help they require.
Working in collaboration and motivated by great minds at the Concordia Institute for Information Systems and Engineering, Université TÉLUQ, and CRISE, Nfissi’s work echoes the collective spirit of applying AI for the improvement of care in mental health.
Alaa Nfissi ‘s breakthrough research which is catching on in suicidal prevention holds great prospects for AI-based SER that provides timely, accurate and compassionate support for people in distress. While the world confronts escalated mental health issues, as the Nfissi model displayed, such innovations illuminate the light at the end of the tunnel, and open the door to a new beginning, where each voice is heard, and each life is appreciated.
