We investigate deep neural network performance in the textindependent speaker recognition task. We demonstrate that using angular softmax activation at the last classification layer of a classification neural network instead of a simple softmax activation allows to train a more generalized discriminative speaker embedding extractor. Cosine similarity is an effective metric for speaker verification in this embedding space. We also address the problem of choosing an architecture for the extractor. We found that deep networks with residual frame level connections outperform wide but relatively shallow architectures. This paper also proposes several improvements for previous DNN-based extractor systems to increase the speaker recognition accuracy. We show that the discriminatively trained similarity metric learning approach outperforms the standard LDA-PLDA method as an embedding backend. The results obtained on Speakers in the Wild and NIST SRE 2016 evaluation sets demonstrate robustness of the proposed systems when dealing with close to real-life conditions.

Deep neural network based speaker embeddings become increasingly popular in the text-independent speaker recognition task. In contrast to a generatively trained i-vector extractor, a DNN speaker embedding extractor is usually trained discriminatively in the closed set classification scenario using softmax. The problem we addressed in the paper is choosing a dnn based speaker embedding backend solution for the speaker verification scoring. There are several options to perform speaker verification in the dnn embedding space. One of them is using a simple heuristic speaker similarity metric for scoring (e.g. cosine metric). Similarly with i-vector based systems, the standard Linear Discriminant Analisys (LDA) followed by the Probabilistic Linear Discriminant Analisys (PLDA) can be used for segregating speaker information. As an alternative, the discriminative metric learning approach can be considered. This work demonstrates that performance of deep speaker embeddings based systems can be improved by using Cosine Similarity Metric Learning (CSML) with the triplet loss training scheme. Results obtained on Speakers in the Wild and NIST SRE 2016 evaluation sets demonstrate superiority and robustness of CSML based systems.

Speaker recognition systems based on deep speaker embeddings have achieved significant performance in controlled conditions according to the results obtained for early NIST SRE (Speaker Recognition Evaluation) datasets. From the practical point of view, taking into account the increased interest in virtual assistants (such as Amazon Alexa, Google Home, Apple Siri, etc.), speaker verification on short utterances in uncontrolled noisy environment conditions is one of the most challenging and highly demanded tasks. This paper presents approaches aimed to achieve two goals: a) improve the quality of far-field speaker verification systems in the presence of environmental noise, reverberation and b) reduce the system quality degradation for short utterances. For these purposes, we considered deep neural network architectures based on TDNN (Time Delay Neural Network) and ResNet (Residual Neural Network) blocks. We experimented with state-of-the-art embedding extractors and their training procedures. Obtained results confirm that ResNet architectures outperform the standard x-vector approach in terms of speaker verification quality for both longduration and short-duration utterances. We also investigate the impact of speech activity detector, different scoring models, adaptation and score normalization techniques. The experimental results are presented for publicly available data and verification protocols for the VoxCeleb1, VoxCeleb2, and VOiCES datasets.

No abstract

This paper presents the Speech Technology Center (STC) speaker recognition (SR) systems submitted to the VOiCES From a Distance challenge 2019 1 . The challenge's SR task is focused on the problem of speaker recognition in single channel distant/far-field audio under noisy conditions. In this work we investigate different deep neural networks architectures for speaker embedding extraction to solve the task. We show that deep networks with residual frame level connections outperform more shallow architectures. Simple energy based speech activity detector (SAD) and automatic speech recognition (ASR) based SAD are investigated in this work. We also address the problem of data preparation for robust embedding extractors training. The reverberation for the data augmentation was performed using automatic room impulse response generator. In our systems we used discriminatively trained cosine similarity metric learning model as embedding backend. Scores normalization procedure was applied for each individual subsystem we used. Our final submitted systems were based on the fusion of different subsystems. The results obtained on the VOiCES development and evaluation sets demonstrate effectiveness and robustness of the proposed systems when dealing with distant/far-field audio under noisy conditions.

No abstract

No abstract

No abstract