Your Mental Health,Empowered by AI

Detect stress through voice, sensors, and psychology with cutting-edge multimodal AI technology.

What is SafeSpace AI?

The future of stress detection through intelligent multimodal analysis

SafeSpace AI is a revolutionary real-time stress detection system that combines accuracy, interpretability, and deployability. Recognizing that stress is complex and varies between individuals and contexts, our platform takes a comprehensive multimodal approach.

By integrating physiological signals, voice-based emotional cues, and self-reported assessments, SafeSpace captures a fuller picture of stress than traditional single-modality systems.

Privacy First

Edge processing ensures data security

Real-time

Instant stress level detection

Multimodal Analysis

Real-time processing of multiple data streams

Our Multimodal Approach

Three powerful data streams working in harmony for unprecedented accuracy

Physiological Signals

WESAD dataset integration with EDA, ECG, temperature, and accelerometry for comprehensive autonomic arousal detection and real-time biosignal processing.

Real-time biosignal processing

Voice Analysis

RAVDESS and IEMOCAP datasets for advanced emotional cue detection through sophisticated pitch jitter analysis and harmonic pattern recognition.

Advanced speech processing

Self-Assessment

DASS-21 questionnaire integration for capturing subjective stress experiences and psychological context with clinical validation standards.

Clinical validation

Explainable AI

Transparent decision-making process

SHAP Analysis
LIME Integration
Gradients

Explainable AI Integration

A central goal of SafeSpace is not only to apply explainable AI (XAI), but to compare techniques such as SHAP, LIME, and Integrated Gradients to understand which best aligns with our system structure.

SHAP (SHapley Additive exPlanations)

Advanced feature importance analysis with game-theoretic foundations

LIME

Local interpretable model explanations for transparent decision-making

Integrated Gradients

Deep learning interpretability through attribution methods

Research Impact & Innovation

Integrating multimodal data and XAI increases both accuracy and trust. SafeSpace is a leap toward stress detection that is as understandable as it is effective.

95%+
Detection Accuracy
<100ms
Response Latency
3
Modalities Integrated

Meet the Minds BehindSafeSpace AI

Dr. Rajendra Kumar Roul - undefined

Dr. Rajendra Kumar Roul

Machine LearningDSA+1
Arsh Chatrath - undefined

Arsh Chatrath

Machine LearningNLP+1
Kabir Oberoi - undefined

Kabir Oberoi

Edge ComputingIoT+1
Devansh - undefined

Devansh

DSPAudio Processing+1
Durvish - undefined

Durvish

Voice AnalysisAudio ML+1
Lakshita - undefined

Lakshita

User ExperienceInterface Design+1
Khushpreet - undefined

Khushpreet

ReactNode.js+1
Samya Aggarwal - undefined

Samya Aggarwal

DjangoFastAPI+1