Encoding and decoding apparatus of LSP (line spectrum pair) parameters

Abstract

An encoding and decoding apparatus quantizes LSP (Line Spectrum Pair) parameters, which are characteristics parameters of spectrum information included in a voice signal, with high accuracy and stability. This encoding and decoding apparatus comprises a first quantizer in which quantization is performed independently in the unit of one frame, and a second quantizer which uses a correlation between adjacent frames. An error comparator compares quantization errors produced by the first quantizer and quantization errors produced by the second quantizer to select the one quantizer whose quantization errors are less than that of the other quantizer. The first quantizer produces a highly accurate quantization with stability regardless a condition of an input voice signal, while the second quantizer produces a highly accurate quantization when input voice signal stays quasi-stationary. By switching these two quantizers, this apparatus can offer stable and highly accurate quantization regardless of the condition of the input voice signal.

1. An LSP parameter encoding apparatus comprising:
 (a) a first quantizing means for dividing a voice signal into frames of a predetermined length and quantizing LSP parameters of an input voice signal into a vector independently for each frame;  
 (b) a second quantizing means for quantizing LSP parameters into a vector based on a correlation between adjacent frames;  
 (c) an error comparing means for comparing quantization errors generated by the first quantizing means and the second quantizing means; and  
 (d) a switch for selecting the quantizing means which produces a smaller quantization error.  
 
     
     
       2. The LSP parameter encoding apparatus according to  claim 1  wherein the second quantizing means comprises:
 (a) a first step quantizing means for quantizing LSP parameters of a present frame into a vector independently for each frame;  
 (b) a second step quantizing means for quantizing a difference between LSP parameters of a present frame and a predicted value of the present frame, said predicted value being predicted based on a quantized value provided by said first step quantizing means and a quantized value of a previous frame.  
 
     
     
       3. An LSP parameter encoding apparatus according to  claim 1 , wherein said error comparing means determines which one of i) the first quantizing means and ii) the second quantizing means produces the smallest quantization error and controls said switch to select the quantizing means which produces the smallest quantization error. 
     
     
       4. An LSP parameter encoding apparatus according to  claim 1 , wherein said error comparing means controls said switch to select one of i) the first quantizing means when a correlation between adjacent frames is smaller than a predetermined value and ii) the second quantizing means when the correlation between the adjacent frames is greater than or equal to the predetermined value. 
     
     
       5. An LSP parameter encoding apparatus according to  claim 1 , wherein said error comparing means detects transmission errors of the input voice signal and controls said switch to select one of i) the first quantizing means and ii) the second quantizing means based on the detected transmission error. 
     
     
       6. An LSP parameter encoding apparatus according to  claim 1 , wherein said error comparing means compares a frequency of the detected error with a predetermined frequency threshold and controls said switch to select one of i) the first quantizing means and ii) the second quantizing means based on the frequency of the detected error. 
     
     
       7. An LSP parameter encoding apparatus according to  claim 2 , wherein said LSP parameters of a predicted value of the present frame are calculated with a linear prediction based on independently quantized LSP parameters in a frame unit in the first step quantizing means and quantized LSP parameters of a past frame are stored in a buffer. 
     
     
       8. An LSP parameter encoding apparatus according to  claim 2 , wherein said first step quantizing means of the second quantizing means comprises:
 i) a first code book for storing a predetermined LSP parameter code;  
 ii) first error-minimizing means for selecting an LSP parameter code which minimizes an error between a calculated LSP parameter of the present frame and the LSP parameter code from said first code book;  
 and said second step quantizing means of the second quantizing means comprises: 
 i) a second code book for storing predetermined prediction coefficients and further LSP parameter codes;  
 ii) decoding means for decoding the quantized value from said first step quantizing means;  
 iii) prediction means for linearly predicting a prediction value of the LSP parameter of the present frame based on 1) the decoded quantized value from said decoding means, 2) quantized LSP parameters of a past frame stored in a buffer, and 3) the predetermined prediction coefficient from said second code book; and  
 iv) second error-minimizing means for selecting a pair of i) the prediction coefficient and ii) the LSP parameter code which minimizes an error of a difference between the predicted LSP parameter of the present frame and an unquantized LSP parameter of the present frame and the LSP parameter code from said second code book.  
 
 
     
     
       9. An LSP parameter encoding apparatus according to  claim 2 , wherein said first step quantizing means of the second quantizing means comprises:
 i) a first code book for storing a predetermined LSP parameter code;  
 ii) first error-minimizing means for selecting the predetermined LSP parameter code which minimizes an error between the calculated LSP parameter of the present frame and the LSP parameter code from said first code book; and  
 second step quantizing means of the second quantizing means comprises: 
 i) a second code book for storing predetermined prediction coefficients and further LSP parameter codes;  
 ii) decoding means for decoding the quantized value obtained from said first step quantizing means  
 iii) prediction means for linearly predicting the prediction value of the LSP parameter of the present frame based on 1) the quantized value from said decoding means, 2) a quantized LSP parameter of a past frame stored in a buffer, and 3) a predetermined prediction coefficient; and  
 iv) second error-minimizing means for selecting the further LSP parameter code which minimizes an error of the difference between the predicted LSP parameter of the present frame and the unquantized LSP parameter of the present frame and LSP parameter code from said second code book.  
 
 
     
     
       10. An LSP parameter decoding apparatus which receives a coded signal from i) a first quantizing means of an encoder for quantizing LSP parameters of an input voice signal into first vectors in a one frame unit independently or ii) a second quantizing means of the encoder for quantizing LSP parameters into second vectors by using a correlation between adjacent frames, said second quantizing means including a first step quantizing means for quantizing LSP parameters of a present frame into the second vectors in the one frame unit independently, and a second step quantizing means for quantizing a difference between the LSP parameters of the present frame and a predicted value of the present frame, said predicted value based on a  quantized values by said first step quantizing means and a quantized value of the adjacent frames, said decoding apparatus comprising:
 error detecting means for detecting errors produced in quantization codes on a transmission line; and  
 decoding means for outputting a decoded quantized value when a code of a further frame next to an error frame is produced by said first quantizing means and for outputting a decoded value from said first step quantizing means of said second quantizing means when the code of the further frame next to the error frame is produced by said second quantizing means.  
 
     
     
       11. An LSP parameter decoding apparatus according to  claim 10 , further comprising:
 first output means for outputting the decoded output of the first quantizing means;  
 second output means for outputting a decoded first step output of said second quantizing means;  
 third output means for outputting a decoded second step output of said second quantizing means;  
 buffer means for storing a quantized LSP parameter of a past frame;  
 prediction means for predicting the LSP parameter of the present frame based on the LSP parameter from said second output means and the LSP parameter from said buffer means;  
 switching means for selecting one output of said decoding means;  
 switch control means for controlling said switching means in response to an output signal from said error detecting means;  
 means for one of i) connecting said first output means and said one output of said decoding means, and ii) connecting said third output means and said one output of said decoding means in response to received coded signals to output a decoded LSP parameter on a frame not detecting an error,  
 means for connecting said buffer means and said one output of said decoding means independently of received coded signals to output quantized LSP parameter of past frame on a frame detecting an error; and  
 means for connecting said first output means and said one output of said decoding means in response to the received coded signals to output a decoded LSP parameter on the further frame next to said frame detecting an error when the received coded signal is the coded signal from the first quantizing means of said encoder; and  
 means for connecting said second output means and said one output of said decoding means in response to the received coded signals to output a decoded LSP parameter by using only the first step codes on the next frame of said frame detected an error when the received coded signal is the coded signal from the second quantizing means of said encoder.  
 
     
     
       12. An LSP parameter encoding and decoding apparatus for use with input voice signals, said apparatus including an LSP parameter encoder and an LSP decoder connected at one end of a two-way communication path, said LSP encoder comprising:
 i) allocating means for dividing the input voice signal from said two-way communication path into frames of a predetermined length and allocating LSP parameters of the input voice signals in each of said frames;  
 ii) first quantizing means for independently quantizing said allocated LSP parameters into a vector in a one frame unit;  
 iii) second quantizing means for quantizing said allocated LSP parameters into vectors by using a correlation between adjacent frames;  
 iv) error comparing means for comparing quantization errors between the first quantizing means and the second quantizing means; and  
 v) switch means for selecting one of the first quantizing means and the second quantizing means based on which produces a smaller error, and  
 said LSP decoder comprising: 
 i) error detecting means for detecting transmission errors of an input code; and  
 ii) judging means for determining whether an error-detecting frequency detected by said error detecting means is less than a predetermined threshold;  
 
 wherein said switch means selects one of said first quantizing means and said second quantizing means based on which has fewer quantization errors when the error-detecting frequency detected by said judging means is less than the predetermined threshold, and said switch means is fixed to said first quantizing means when said error-detecting frequency is not less than the predetermined threshold.  
 
     
     
       13. An LSP parameter decoding apparatus which receives a coded signal from
   i )  a first quantizing means of an encoder for quantizing LSP parameters of an input voice signal into first vectors in a one frame unit independently or        ii )  a second quantizing means of the encoder for quantizing LSP parameters into second vectors by using a correlation between a present frame and a previous frame, said second quantizing means including a first step quantizing means for quantizing LSP parameters of the present frame into the second vectors in the one frame unit independently, and a second step quantizing means for quantizing a difference between the LSP parameters of the present frame and a predicted value of the present frame, said predicted value based on a quantized value by said first step quantizing means and a quantized value of the previous frame, said decoding apparatus comprising:      error detecting means for detecting errors produced in quantization codes on a transmission line; and        decoding means for outputting a decoded quantized value when a code of a further frame next to an error frame is produced by said first quantizing means and for outputting a decoded value from said first step quantizing means of said second quantizing means when the code of the further frame next to the error frame is produced by said second quantizing means.       
     
     
       14. An LSP parameter decoding apparatus which receives a coded signal from
   i )  a first quantizing means of an encoder for quantizing LSP parameters of an input voice signal into first vectors in a one frame unit independently or        ii )  a second quantizing means of the encoder for quantizing LSP parameters into second vectors by using a correlation between a present frame and a adjacent previous frame, said second quantizing means including a first step quantizing means for quantizing LSP parameters of the present frame into the second vectors in the one frame unit independently, and a second step quantizing means for quantizing a difference between the LSP parameters of the present frame and a predicted value of the present frame, said predicted value based on a quantized value by said first step quantizing means and a quantized value of the adjacent previous frame, said decoding apparatus comprising:      error detecting means for detecting errors produced in quantization codes on a transmission line; and        decoding means for outputting a decoded quantized value when a code of a further frame next to an error frame is produced by said first quantizing means and for outputting a decoded value from said first step quantizing means of said second quantizing means when the code of the further frame next to the error frame is produced by said second quantizing means.