Artificial bandwidth extension of narrowband speech applied to CELP-Type speech coding

Abstract

In this paper, an artificial bandwidth extension (ABE) algorithm from narrowband to wideband is proposed in order to improve the quality of narrowband speech decoded by a code-excited linear prediction (CELP)-type speech coder. The proposed ABE algorithm is based on spectral band replication in the modified discrete cosine transform (MDCT) domain, without any additional bits. In particular, a patch index search for band replication is restricted so that the harmonic structure of the wideband speech is maintained after ABE. To this end, the proposed ABE algorithm first determines whether the current analysis frame of speech signals is voiced or unvoiced. Then, a harmonic or correlation-based spectral band replication approach is then applied for the voiced or unvoiced frame, respectively. The proposed ABE algorithm is finally embedded into the ITU-T Recommendation G.729 speech decoder as a post-processor. It is shown from a subjective evaluation test that the mean opinion score of the wideband speech signals extended by the proposed ABE algorithm is 75.5, which is around 14% higher than that of the narrowband speech signals. ©2013 International Information Institute.