Telephone systems commonly transmit narrowband speech with an audio bandwidth limited to the traditional telephone band of 300–3400 Hz. The narrow bandwidth degrades the quality and intelligibility of speech. A significant improvement is provided by the transmission of wideband speech, which typically covers the frequency range of 50–7000 Hz. Currently, wideband speech services are increasingly deployed in cellular telephone networks, but the transition from narrowband to wideband speech transmission is expected to take a long time. The quality of narrowband speech can be improved by artificial bandwidth extension (ABE) that attempts to reconstruct the missing frequency content artificially from the narrowband speech signal. ABE can be implemented in the receiving terminal device and used with the existing narrowband speech coding and transmission techniques. ABE is expected to gain importance during the transition to wideband telephony as a means to reduce the perceived quality difference between narrowband and wideband speech. This thesis contributes to the development and evaluation of ABE, especially for the application in mobile phones. Two new ABE methods are proposed: the filter bank-based ABE (FB-ABE) for the high-frequency range above the traditional telephone band and the lowband ABE (LB-ABE) for the low-frequency range below the telephone band. The feasibility of implementation in mobile devices is taken into account in the design of the methods. The methods were evaluated primarily by subjective listening tests that simulated realistic telephone speech including the use of speech codecs. A subjective evaluation of ABE in three languages was also organized and ABE was found to improve the speech quality in all the evaluated languages. Furthermore, conversational evaluations were arranged for simulated telephone connections involving real-time ABE processing. The ABE methods presented in this thesis are shown to improve, in general, the perceived quality and preference over narrowband speech.
|Translated title of the contribution||Puhelinkaistan keinotekoisten laajennusmenetelmien kehittäminen ja evaluointi|
|Publication status||Published - 2013|
|MoE publication type||G5 Doctoral dissertation (article)|
- speech enhancement
- artificial bandwidth extension
- quality evaluation