Search

We discuss quantum two-block codes, a large class of CSS codes constructed from two commuting square matrices.Interesting families of such codes are generalized-bicycle (GB) codes and two-block group-algebra (2BGA) codes, where a cyclic group is replaced with an arbitrary finite group, generally non-abelian. We present code construction and give several expressions for code dimension, applicable depending on whether the constituent group is cyclic, abelian, or non-abelian. This gives a simple criterion for an essentially non-abelian 2BGA code guaranteed not to be permutation-equivalent to such a code based on an abelian group. We also give a lower bound on the distance which, in particular, applies to the case when a 2BGA code reduces to a hypergraph-product code constructed from a pair of classical group codes.

Generalized bicycle (GB) codes is a class of quantum error-correcting codes constructed from a pair of binary circulant matrices. Unlike for other simple quantum code ans\"atze, unrestricted GB codes may have linear distance scaling. In addition, low-density parity-check GB codes have a naturally overcomplete set of low-weight stabilizer generators, which is expected to improve their performance in the presence of syndrome measurement errors. For such GB codes with a given maximum generator weight $w$, we constructed upper distance bounds by mapping them to codes local in $D\le w-1$ dimensions, and lower existence bounds which give $d\ge {\cal O}({n}^{1/2})$. We have also done an exhaustive enumeration of GB codes for certain prime circulant sizes in a family of two-qubit encoding codes with row weights 4, 6, and 8; the observed distance scaling is consistent with $A(w){n}^{1/2}+B(w)$, where $n$ is the code length and $A(w)$ is increasing with $w$., Comment: 12 pages, 5 figures

Mismatch between enrollment and test conditions causes serious performance degradation on speaker recognition systems. This paper presents a statistics decomposition (SD) approach to solve this problem. This approach decomposes the PLDA score into three components that corresponding to enrollment, prediction and normalization respectively. Given that correct statistics are used in each component, the resultant score is theoretically optimal. A comprehensive experimental study was conducted on three datasets with different types of mismatch: (1) physical channel mismatch, (2) speaking behavior mismatch, (3) near-far recording mismatch. The results demonstrated that the proposed SD approach is highly effective, and outperforms the ad-hoc multi-condition training approach that is commonly adopted but not optimal in theory.

This paper describes system setup of our submission to speaker diarisation track (Track 4) of VoxCeleb Speaker Recognition Challenge 2020. Our diarisation system consists of a well-trained neural network based speech enhancement model as pre-processing front-end of input speech signals. We replace conventional energy-based voice activity detection (VAD) with a neural network based VAD. The neural network based VAD provides more accurate annotation of speech segments containing only background music, noise, and other interference, which is crucial to diarisation performance. We apply agglomerative hierarchical clustering (AHC) of x-vectors and variational Bayesian hidden Markov model (VB-HMM) based iterative clustering for speaker clustering. Experimental results demonstrate that our proposed system achieves substantial improvements over the baseline system, yielding diarisation error rate (DER) of 10.45%, and Jacard error rate (JER) of 22.46% on the evaluation set., Comment: 5 pages, 2 figures, A report about our diarisation system for VoxCeleb Challenge, Interspeech conference workshop

Decision making is an important component in a speaker verification system. For the conventional GMM-UBM architecture, the decision is usually conducted based on the log likelihood ratio of the test utterance against the GMM of the claimed speaker and the UBM. This single-score decision is simple but tends to be sensitive to the complex variations in speech signals (e.g. text content, channel, speaking style, etc.). In this paper, we propose a decision making approach based on multiple scores derived from a set of cohort GMMs (cohort scores). Importantly, these cohort scores are not simply averaged as in conventional cohort methods; instead, we employ a powerful discriminative model as the decision maker. Experimental results show that the proposed method delivers substantial performance improvement over the baseline system, especially when a deep neural network (DNN) is used as the decision maker, and the DNN input involves some statistical features derived from the cohort scores., Comment: APSIPA ASC 2016