Whisper Speech Enhancement Using Joint Variational Autoencoder for Improved Speech Recognition