<ul data-eligibleForWebStory="true">Physiological signals pose challenges due to artifacts and non-stationarity, making traditional analysis methods inadequate.A new wavelet-based approach is introduced for multi-scale time-frequency representation of physiological signals.Novel pretrained models for EMG and ECG signals demonstrate superior performance and establish new benchmarks for downstream tasks.A unified multi-modal framework is created by integrating pretrained EEG model to address challenges like low SNR and device variations.The wavelet-based architecture enhances the analysis of diverse physiological signals, offering a foundation for future biomedical applications.