Speech-dependent Data Augmentation for Own Voice Reconstruction with Hearable Microphones in Noisy Environments | Synapse