Abstract | ||
---|---|---|
This paper presents a new method to apply variable bit-rate predictive quantization of the variable model order LPC parameters. In addition, the method is employed to interpolate the parameters within the analysis frame. The LPC model order selection algorithm of this work is based on the characteristics of the input signal and on the performance of the LPC model. Hence, the variable bit-rate LPC quantization is source controlled. The number of quantized parameters needs to be identical in successive frames to be able to apply the predictive quantization and to interpolate parameters inside the frame. Therefore, the order of the LPC model of the previous frame needs to be expanded or reduced to be the same as the current frame LPC model. The advantage of variable model order LPC quantization is the lowered average bit-rate compared to fixed rate while the speech quality remains the same. |
Year | DOI | Venue |
---|---|---|
1998 | 10.1109/ICASSP.1998.674364 | PROCEEDINGS OF THE 1998 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING, VOLS 1-6 |
Keywords | Field | DocType |
decoding,testing,parameter estimation,speech synthesis,linear predictive coding,variable bit rate,speech codec,interpolation,speech intelligibility,predictive models,speech coding,performance,cost function,quantization | Speech coding,Interpolation,Artificial intelligence,Estimation theory,Codec2,Linear predictive coding,Pattern recognition,Algorithm,Speech recognition,Quantization (physics),Harmonic Vector Excitation Coding,Quantization (signal processing),Mathematics | Conference |
ISSN | Citations | PageRank |
1520-6149 | 0 | 0.34 |
References | Authors | |
5 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Pasi Ojala | 1 | 31 | 9.21 |
Ari Lakaniemi | 2 | 41 | 5.23 |