This paper investigates the performance of Deep Learning for speech emotion classification when the speech is compounded with noise. It reports on the classification accuracy and concludes with the future directions for achieving greater robustness for emotion recognition from noisy speech.