Automatic recognition of speech without any audio information | IEEE Conference Publication | IEEE Xplore