Adriana Vasilache

Overview of the EVS codec architecture

The recently standardized 3GPP codec for Enhanced Voice Services (EVS) offers new features and improvements for low-delay real-time communication systems. Based on a novel, switched low-delay speech/audio codec, the EVS codec contains various tools for better compression efficiency and higher quality for clean/noisy speech, mixed content and music, including support for wideband, super-wideband and full-band content. The EVS codec operates in a broad range of bitrates, is highly robust against packet loss and provides an AMR-WB interoperable mode for compatibility with existing systems. This paper gives an overview of the underlying architecture as well as the novel technologies in the EVS codec and presents listening test results showing the performance of the new codec in terms of compression and speech/audio quality.

Indexing and entropy coding of lattice codevectors

We present two methods of entropy coding for the lattice codevectors. We compare our entropy coding methods with one method previously presented in the literature from the point of view of rate-distortion as well as of the computation complexity and memory requirements. The results are presented for artificial Laplacian and Gaussian data, as well as for LSF parameters of speech signals. In the latter case, the multiple scale lattice VQ (MSLVQ) is used for quantization, which reduces the rate gain of the entropy coding method when compared with the fixed rate case, but allows a dynamic allocation of the bits in the whole speech coding scheme.

Multiple-scale leader-lattice VQ with application to LSF quantization

In this study we introduce two new quantization structures, namely the multiple scale leader-lattice vector quantization (MSLLVQ) using different lattice truncations and (MSLLVQ) using different unions of leader vectors. The design methods and the corresponding encoding algorithms are presented for each structure. We first focus on the search in a truncated lattice and in a leader class of a lattice and we then propose two new search methods for pyramidal (l1-) truncations and l1/2-truncations. The encoding algorithms have low memory and computational requirements. The new schemes outperform in terms of SNR other lattice quantization structures presented in the literature in the case of generalized Gaussian source densities with decay parameters 0.5, 1 and 2. In addition, when used for LSF quantization they can achieve a mean spectral distortion below 1 dB at only 19 bits, being superior to the G.729 speech coder.

Predictive multiple-scale lattice VQ for LSF quantization

This paper introduces a new lattice quantization scheme, the multiple-scale lattice vector quantization (MSLVQ), based on the truncation of the D/sub 10//sup +/ lattice. The codebook is composed of several copies of the truncated lattice scaled with different scaling factors. A fast nearest neighbor search is introduced. We compare the performance of predictive MSLVQ for quantization of line spectrum frequency (LSF) coefficients with the quantization technique used in the G.729 codec and show the better performance of our method in terms of spectral distortion. The MSLVQ scheme achieves the transparent quality at 21 bits/frame.

Vectorial Spectral Quantization for Audio Coding

The paper introduces a new coding methodology of the spectral modified discrete cosine transform (MDCT) coefficients of an audio signal. A lattice quantizer is used for each spectral sub-band, having the dimension equal to the size of the respective sub-band. The information that needs to be encoded consists of lattice codevector indexes, side information relative to the number the bits on which the indexes are represented and the integer exponents of the sub-band scaling factors. The nature of the side information, together with the parameterization of the quantization resolution allows the use of the method for a large range of bitrates e.g. for 44.1 kHz sampled mono files, from 128 kbits/s down to 16 kbits/s. Subjective listening tests show similar performance of the proposed method to the advanced audio coding (AAC) codec for high bitrates (128 kbits down to 64 kbits/s) and clearly better performance for lower bitrates

Robust indexing of lattices and permutation codes over binary symmetric channels

In this paper we develop two families of indexing methods for permutation codes and lattices, the lexicographical and binomial indexing families. The parameters of a method within a family can be optimized with respect to a given criterion in order to obtain a good indexing method on the considered set of vectors to be indexed. The proposed indexing methods can be straightforwardly applied for error resilient coding, but also to other applications as e.g. the magnetic recording which is based on (d,k) constrained codes. We have tested the performance of the proposed method at the optimization of the indexing method of a codebook with respect to the channel distortion for several types of lattices and other structured vector quantizers over binary symmetric channels.

Robust speech coding with EVS

This paper discusses the voice and audio quality characteristics of EVS, the recently standardized 3GPP codec. Especially frame erasure conditions were evaluated. Comparison to industry standard voice codecs: 3GPP AMR and AMR-WB as well as direct signals at varying bandwidths was made. Speech quality was evaluated with two subjective listening tests containing clean and noisy speech in Finnish language. Five different random frame erasure rates were evaluated: 0 %, 3 %, 6 %, 10 % and 15 %. Nine-scale subjective mean opinion score was calculated for all tested conditions.

Rate-distortion models for entropy constrained lattice quantization

In this paper we present new models for rate-distortion curves for entropy coded lattice codevectors. Exact models for both the rate and the distortion are proposed for the lattice Zn for generalized Gaussian sources. The resulting precision with respect to experimental values is improved by 50% over previously proposed models. In addition an approximate model for general lattices is proposed for Gaussian sources, its precision being verified against experimental values and shown to improve the estimation precision from 10% to 4%.

Flexible spectrum coding in the 3GPP EVS codec

This paper proposes a flexible encoding technique based on multi-stage multiple scale lattice vector quantization and block-constrained trellis coded vector quantization. It is used for the spectrum encoding, more precisely encoding of the LSF parameters, and incorporated in the recently standardized 3GPP EVS codec. The proposed method can handle multiple bit allocations and signal types with low complexity and low memory requirements.

Using context dependent distributions for coding prediction residuals of companded audio signals

We propose a context conditioning scheme for encoding the prediction residuals when compressing files containing companded signals. Our scheme encompasses decompanding of the signals, performing linear prediction in the decompanded domain, and then companding back the predicted value into a companded prediction (CP) value, which will differ from the true companded value by an amount called companded prediction residual (CPR). The proposed context conditioning scheme for encoding the CPR, uses a probability distribution conditional on a context made up of two quantities: (1) the predicted value and (2) a scale parameter of the background probability distribution function assumed for the decompanded domain residuals. Various context building schemes and various storing strategies can be used to obtain the necessary conditional coding distribution of the CPR, to be used with an arithmetic coder or range coder. The implementation in fixed point precision can be done very efficiently and with very low memory requirements.