5.2.1 Perceptual weighting

26.4453GPPCodec for Enhanced Voice Services (EVS)Detailed algorithmic descriptionRelease 15TS

The encoding parameters, such as adaptive codebook delay and gain, algebraic codebook index and gain are searched by minimizing the error between the input signal and the synthesized signal in a perceptually weighted domain. Perceptual weighting is performed by filtering the signal through a perceptual weighting filter, derived from the LP filter coefficients. The perceptually is similar to the weighting also used in open-loop pitch analysis. However, an adaptive perceptual weighting is used in case of LP-based excitation coding.

The traditional perceptual weighting filter has inherent limitations in modelling the formant structure and the required spectral tilt concurrently. The spectral tilt is more pronounced in wideband signals due to the wide dynamic range between low and high frequencies. A solution to this problem is to introduce a pre-emphasis filter at the input and enhance the high frequency content in case of wideband signals. The LP filter coefficients are then found by means of LP analysis on the pre-emphasized signal. Subsequently, they are used to form a perceptual weighting filter. Its transfer function is the same as the LP filter transfer function but with the denominator having fixed coefficients (similar to the pre-emphasis filter). In this way, the weighting in formant regions is decoupled from the spectral tilt as shown below. Finally, the pre emphasized signal is filtered through the perceptual filter to obtain a perceptually weighted signal, which is used further.

The perceptual weighting filter has the following form




and is equal to 0.68.

Because is computed based on the pre-emphasized signal , the tilt of the filter is less pronounced compared to the case when is computed based on the original signal (as the pre-emphasized signal itself exhibits less spectral tilt than the original wideband signal). Since de-emphasis is performed in the decoder, it can be shown that the quantization error spectrum is shaped by a filter having a transfer function . Thus, the spectrum of the quantization error is shaped by a filter whose transfer function is , with computed based on the pre-emphasized signal. The perceptual weighting is performed on a frame basis while the LP filter coefficients are calculated on a subframe basis using the principle of LSP interpolation, described in subclause For a subframe of size = 64, the weighted speech is given by