High Quality Speech Compression Research Articles

Adaptive nonuniform polar quantization application to high quality speech compression

The logarithmic companding technique has shown to be extremely useful in speech quantization with rate of 8 bits/sample. However, for lower bit rates it is not the ideal solution for high quality speech coding. Because of that, in this paper we establish source coding scheme which enables better spectrum efficiency for input that has a large dynamic range. Since our wish is also to improve signal quality in comparison with quality defined with standards G.711 and G.712, we opt for adaptive technique application to the speech coding. Our research shows that proper design of forward gain-adaptive polar quantization can enable compression of about 1 bit/sample as well as significantly better quality than in case of using coder designed according to standard G.711. Furthermore, performances can be sustained over the whole speech dynamic range. Also, if the requisite speech quality is not supposed to be lower than G.712 standard quality, the achieved compression can be almost 1.5 bits/sample. Besides, we propose knew simple encoding rule which can additionally reduce bit rate for 0.1 bit/sample.

Application of Wavelet LPC Excitation Model for Speech Compression

This paper presents an application of linear predictive coding (LPC) excitation wavelet models for low bit- rate, high-quality speech compression. The compression scheme exploits the model properties, especially magnitude dependent sensitivity, scale dependent sensitivity, and limited frame length. We use the wavelet model in an open-loop dither based codebook scheme. With t his approach, the compression yields a signal-to-noise ratio of at least 11 dB at rates of 5 kbit/s and.

Classified vector excitation speech coding

Code excited linear prediction (CELP) has been shown [1] to be capable of achieving high quality speech compression at very low bit rates. However, the original method suffers from the two drawbacks that it requires a large amount of computation, and the Gaussian codebook used is not able to reproduce a sharp impulse in the coded residual waveform. The authors present a classified vector excitation coding (CVXC) scheme based on selective modelling of the LPC residual which attempts to overcome both these drawbacks by using separate codebooks for separate excitation types. A further feature of our coder which helps preserve the nature of pulsive residuals, is the use of pairwise nearest neighbour (PNN) matching [2] to generate the excitation codebooks.