Abstract: Information hiding has been a research problem since 1990’s. Random silent gap of speech signal makes less hiding space for speech than space for other audio signals such as music although information hiding in speech has very good prospective in Internet or any other wire/wireless channel. This paper provides an information hiding method in MFCC for speech signal. Because short time energy of speech is more invariable and can ensure the synchrony of frame in hiding and extracting, it makes extracting algorithm restore the secret from hided speech signal correctly. This method can be used in application of Internet or LAN channel with high speed transmission.

Key words: speech, information hiding, speech recognition, MFCC