Semi-black-box Attacks Against Speech Recognition Systems Using Adversarial Samples | IEEE Conference Publication | IEEE Xplore