Abstract
There is a marked difference between speech heard in direct face-to-face conversation and speech heard over the telephone. A good example is provided by an FM radio broadcast when a listener calls in to chat with the disk jockey on the air. This disparity in quality led us to develop an algorithm of generating wideband speech from ordinary telephone speech. The algorithm has two novel points. The first is spectrum envelope reconstruction based on codebook mapping. The other is speech signal reconstruction using the speech waveform specified by the reconstructed spectrum envelope. Because the algorithm makes it possible to generate high quality speech without the use of any additional transmitted information, it is important and useful in the area of speech coding. Evaluation by the acoustic distance measure and by listening tests confirmed the good performance of the algorithm.
Original language | English |
---|---|
Pages (from-to) | 104-109 |
Number of pages | 6 |
Journal | NTT Review |
Volume | 7 |
Issue number | 3 |
Publication status | Published - May 1 1995 |
Externally published | Yes |
ASJC Scopus subject areas
- Computer Networks and Communications
- Electrical and Electronic Engineering