The increasing prevalence of mental health issues, particularly depression and suicidal tendencies, necessitates innovative approaches for support and intervention. This study introduces an automated classification system that leverages a multi-modal approach, integrating text and audio inputs to refine the accuracy and efficiency in identifying individuals at risk. By employing natural language processing (NLP) techniques, the system analyzes text data collected from social media, online surveys, and screening questionnaires, allowing for the detection of linguistic patterns and emotional cues indicative of depression. In tandem, audio analysis captures vocal characteristics such as tone, pitch, and speech patterns, which serve as critical indicators of mental states. This holistic approach not only improves the classification accuracy of individuals exhibiting depressive symptoms but also facilitates early intervention strategies by the flagging of those displaying suicidal tendencies, clinicians and caregivers to identify risks more effectively. The promising results suggest that, when implemented responsibly, this approach could significantly enhance mental healthcare delivery, fostering a proactive rather than reactive stance in mental health management. This innovative framework represents a substantial step forward in advancing mental health support through the intersection of technology and psychological well-being.