Vector encoding method, in particular for voice signals

Details Vector encoding method, in particular for voice signals Vector encoding method, in particular for voice signals

1995-11-07

1998-03-17

MacDonald, Allen R.

395 228, 395 229, 395 226, 395 232, G10L 702

Patent

active

057296545

DESCRIPTION:

BRIEF SUMMARY
BACKGROUND OF THE INVENTION

The invention is based on a vector encoding method, in particular for voice signals, using a codebook for prediction coding.
In the encoding and decoding of voice signals, in particular for mobile radio applications, the voice signal is sampled and subdivided into intervals (time intervals). For each interval, predicted values are formed for different types of signal parameters. Such signal parameters are, for example, short-term parameters for characterizing the formant structure (resonances of the voicebox) and long-term parameters for characterizing the pitch structure (level of tone) of the voice signal (ANT November 1988, pages 93 to 105). In voice encoding by means of "analysis by synthesis", the model and excitation parameters are quantized, encoded and transmitted to the receiver. For further reducing the bit rate, vector quantization is used (see above; DE/EP 0 266 620 T1; EP 504 627 A2; EP 294 020 A2).
It is known from ANT Nachrichtentechnische Berichte, Issue 5, November 1988, pages 93 to 105, in particular page 102, second column, to page 103, second column, that up to 90% of the components of the codebook vector can be set to zero without impairing the voice quality. In addition, it is proposed there to restrict even further the number of excitation sequences for the filter for synthetic voice generation by choosing not only a fixed position of the pulses (coefficients) but in addition fixed amplitudes for them. There is chosen there a fixed excitation codebook whose vectors comprise four elements other than zero (vector components).


SUMMARY OF THE INVENTION

The object of the present invention is to develop a method of the above-described type in such a way as to provide codebook parameters which have a low error susceptibility, in particular for purposes of transmission over faulty channels, with reduced storage and processing effort.
According to the invention, the vector encoding method comprises storing combination patterns for excitation signals of an excitation filter for generating synthetic signals, in particularly voice signals, with n, advantageously 4, amplitude values other than 0 and having the same magnitudes but differing only in operational sign or addresses for storing the amplitude values in a first codebook; storing respectively associated autocorrelation coefficients in another codebook; and performing a logic operation on the autocorrelation coefficients with an autocorrelation function for pulse response of the excitation filter to give an energy of a filtered excitation signal. Only combination patterns having three autocorrelation coefficients other than zero are allowed and amplitude values other than 0 of the autocorrelation coefficients beyond a correlation maximum (AKF.sub.(0)) differ only in an operational sign thereof.
Preferred embodiments are explained in more detail in the description hereinbelow and claimed in the appended claims.
The method according to the invention is based on the following perceptions: not have to be stored. It suffices if their addresses, i.e. their positions are stored. This reduces the storage effort and the processing considerably. Combination patterns and their distribution can be transmitted separately. Since voice codes react more sensitively to pulse patterns (combination patterns of the autocorrelation coefficients) than to their distribution (arrangement of the vector components), a better voice quality is obtained in the case of faulty transmission. For this reason, the method according to the invention can be used advantageously, in particular for mobile radio.


BRIEF DESCRIPTION OF THE DRAWING

An exemplary embodiment of the invention is explained in more detail with reference to the drawings, in which:
FIG. 1 is a block diagram showing the principle of a method of RELP voice encoding,
FIG. 2 is a block diagram showing the principle of a method of analysis by synthesis,
FIG. 3 shows an example of the choice of autocorrelation coefficients according to the invention and
FIG. 4 shows part-codeboo

REFERENCES:
patent: 4486899 (1984-12-01), Fushikida
patent: 4860355 (1989-08-01), Copperi
patent: 4868867 (1989-09-01), Davidson et al.
patent: 4907276 (1990-03-01), Aldersberg
patent: 4969192 (1990-11-01), Chen et al.
patent: 5327520 (1994-07-01), Chen
patent: 5339164 (1994-08-01), Lim
patent: 5457783 (1995-10-01), Chhatwal
patent: 5621852 (1997-04-01), Lin

Affiliated with

Also associated with

Rating

Vector encoding method, in particular for voice signals does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Vector encoding method, in particular for voice signals, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Vector encoding method, in particular for voice signals will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-965475