This research introduces a unified classifier that can identify multiple demographic characteristics from speech samples, enhancing personalization and security applications.

• Predicts age, gender, native language, education level, and country from voice
• Leverages pre-trained WavLM models for advanced speech feature extraction
• Enables more accurate speaker identification for security applications
• Creates opportunities for personalized experiences in language learning and accessibility

For security professionals, this technology offers enhanced authentication capabilities, digital forensics tools, and improved fraud detection through voice biometrics.

Demographic Attributes Prediction from Speech Using WavLM Embeddings