Artificial Bandwidth Extension of Telephony Speech by Data Hiding
Siyue Chen, Henry Leung
Abstract
Siyue Chen, Henry Leung
Abstract
The current public switched telephone network (PSTN) is only able to deliver analog signals in a relatively narrow frequency band, about 300-3400 Hz. Such a bandwidth is so small that the intelligibility of speech frequently suffers from poor subjective quality. In order to improve the voice quality and intelligibility, it is proposed to use a data hiding technique to extend the PSTN channel bandwidth artificially. That is, the higher frequency components beyond the PSTN bandwidth are encoded and imperceptibly embedded into the narrowband signal. When the hidden signal is extracted at the receiver, wideband speech can be reconstructed with better perceptual quality. The advantage of the proposed scheme lies in that it is fully compatible with conventional end-user equipment, e.g., a plain ordinary telephone set (POTS). Experimental results show that the proposed scheme has a better performance than conventional bandwidth extension methods with speaker-independent training.
OpenAlex reports 25 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The current public switched telephone network (PSTN) is only able to deliver analog signals in a relatively narrow frequency band, about 300-3400 Hz. Such a bandwidth is so small that the intelligibility of speech frequently suffers from poor subjective quality. In order to improve the voice quality and intelligibility, it is proposed to use a data hiding technique to extend the PSTN channel bandwidth artificially. That is, the higher frequency components beyond the PSTN bandwidth are encoded and imperceptibly embedded into the narrowband signal. When the hidden signal is extracted at the receiver, wideband speech can be reconstructed with better perceptual quality. The advantage of the proposed scheme lies in that it is fully compatible with conventional end-user equipment, e.g., a plain ordinary telephone set (POTS). Experimental results show that the proposed scheme has a better performance than conventional bandwidth extension methods with speaker-independent training.
Key concepts: Bandwidth extension, Telephony, Computer science, Extension (predicate logic), Bandwidth (computing), Speech recognition, Telecommunications, Speech coding