Wide-Band Audio Coding Based on Frequency-Domain Linear Prediction

Abstract

We revisit an original concept of speech coding in which the signal is separated into the carrier modulated by the signal envelope. A recently developed technique, called frequency-domain linear prediction (FDLP), is applied for the efficient estimation of the envelope. The processing in the temporal domain allows for a straightforward emulation of the forward temporal masking. This, combined with an efficient nonuniform sub-band decomposition and application of noise shaping in spectral domain instead of temporal domain (a technique to suppress artifacts in tonal audio signals), yields a codec that does not rely on the linear speech production model but rather uses well-accepted concept of frequency-selective auditory perception. As such, the codec is not only specific for coding speech but also well suited for coding other important acoustic signals such as music and mixed content. The quality of the proposed codec at 66 kbps is evaluated using objective and subjective quality assessments. The evaluation indicates competitive performance with the MPEG codecs operating at similar bit rates.

Perceptually Motivated Sub-band Decomposition for FDLP Audio Coding

“…Since the QMF decomposition in the base line system is implemented in a tree-like structure (6-stage binary tree [4]), the merging is equivalent to tying some branches at any particular stage to form a nonuniform band. This tying operation tries to follow critical band decomposition in the human auditory system.…”

“…As described in [4], the full-band input signal is decomposed into frequency sub-bands. In each sub-band, FDLP is applied and Line Spectral Frequencies (LSFs) approximating the sub-band temporal envelopes are quantized using Vector Quantization (VQ).…”

“…In this case, the objective quality is reduced significantly. The block of quantization, described in [4], was not modified during these experiments.…”

“…This version of the codec achieves good quality of reconstructed signal compared to the older version using Gaussian band decomposition [4]. The performance advantage was mainly due to the increased frequency resolution in lower sub-bands and critical sub-sampling resulting in the minimal number of FDLP residual parameters to be transmitted.…”

“…Unlike [2], this technique applies FDLP to approximate relatively long segments of the Hilbert envelopes in individual frequency sub-bands. This speech compression technique was later extended to high quality audio coding [4], where an input audio signal is decomposed into N frequency sub-bands. Temporal envelopes of these sub-bands are then approximated using FDLP applied over relatively long time segments (e.g.…”