Digital encoding, decoding and processing of speech signals.

摘要

The invention provides a method of digital encoding of speech signals which has advantages over adaptive coding methods in that it facilitates processing of the encoded signals for example by pause removal, speech rate control, storing and forwarding etc, but which does not require as high a bit rate for a given speech quality as pulse code modulation. …In the method of the invention the speech signals are digitised during successive time intervals to produce pulse code modulated samples and these samples are standardised for each interval by subtracting from each the minimum sample value and dividing each difference by a scale factor proportional to the sample range. The standardised samples, together with values for the minimum sample value and the scale factor, are coded into code words, each representing one time interval. An example of a code word is shown in Fig. 7. …During pauses, only a single code word is generated, which, of course, represents the background noise level, and the code word includes the coded number of time intervals of silence representing the length of the pause. …A mu-law encoder/decoder may be used for digitising the speech signals. …In the examples given a 41 millisecond time interval is proposed, divided into 246 sampling intervals, each represented by 4-bit code, the sample codes and other information, including check digits, being packed into a code word onsisting of a 128 8-bit bytes.