arXiv subjects

Helen Meng

Publications and source records attributed to Helen Meng.

At least 217 records · Page 12Linked to original sources

arXiv2021

Non-Autoregressive Transformer ASR with CTC-Enhanced Decoder Input

Non-autoregressive (NAR) transformer models have achieved significantly inference speedup but at the cost of inferior accuracy compared to autoregressive (AR) models in automatic speech recognition (ASR). Most of the NAR transformers take a fixed-length sequence filled with MASK tokens or a redundant sequence copied from encoder states as decoder input, they cannot provide efficient target-side information thus leading to accuracy degradation. To address this problem, we propose a CTC-enhanced NAR transformer, which generates target sequence by refining predictions of the CTC module. Experimental results show that our method outperforms all previous NAR counterparts and achieves 50x faster decoding speed than a strong AR baseline with only 0.0 ~ 0.3 absolute CER degradation on Aishell-1 and Aishell-2 datasets.

cs.SD↗

arXiv2021

Towards Multi-Scale Style Control for Expressive Speech Synthesis

This paper introduces a multi-scale speech style modeling method for end-to-end expressive speech synthesis. The proposed method employs a multi-scale reference encoder to extract both the global-scale utterance-level and the local-scale quasi-phoneme-level style features of the target speech, which are then fed into the speech synthesis model as an extension to the input phoneme sequence. During training time, the multi-scale style model could be jointly trained with the speech synthesis model in an end-to-end fashion. By applying the proposed method to style transfer task, experimental results indicate that the controllability of the multi-scale speech style model and the expressiveness of the synthesized speech are greatly improved. Moreover, by assigning different reference speeches to extraction of style on each scale, the flexibility of the proposed method is further revealed.

cs.SD↗

arXiv2021

Adversarial defense for automatic speaker verification by cascaded self-supervised learning models

Automatic speaker verification (ASV) is one of the core technologies in biometric identification. With the ubiquitous usage of ASV systems in safety-critical applications, more and more malicious attackers attempt to launch adversarial attacks at ASV systems. In the midst of the arms race between attack and defense in ASV, how to effectively improve the robustness of ASV against adversarial attacks remains an open question. We note that the self-supervised learning models possess the ability to mitigate superficial perturbations in the input after pretraining. Hence, with the goal of effective defense in ASV against adversarial attacks, we propose a standard and attack-agnostic method based on cascaded self-supervised learning models to purify the adversarial perturbations. Experimental results demonstrate that the proposed method achieves effective defense performance and can successfully counter adversarial attacks in scenarios where attackers may either be aware or unaware of the self-supervised learning models.

eess.AS↗

arXiv2021

Replay and Synthetic Speech Detection with Res2net Architecture

Existing approaches for replay and synthetic speech detection still lack generalizability to unseen spoofing attacks. This work proposes to leverage a novel model structure, so-called Res2Net, to improve the anti-spoofing countermeasure's generalizability. Res2Net mainly modifies the ResNet block to enable multiple feature scales. Specifically, it splits the feature maps within one block into multiple channel groups and designs a residual-like connection across different channel groups. Such connection increases the possible receptive fields, resulting in multiple feature scales. This multiple scaling mechanism significantly improves the countermeasure's generalizability to unseen spoofing attacks. It also decreases the model size compared to ResNet-based models. Experimental results show that the Res2Net model consistently outperforms ResNet34 and ResNet50 by a large margin in both physical access (PA) and logical access (LA) of the ASVspoof 2019 corpus. Moreover, integration with the squeeze-and-excitation (SE) block can further enhance performance. For feature engineering, we investigate the generalizability of Res2Net combined with different acoustic features, and observe that the constant-Q transform (CQT) achieves the most promising performance in both PA and LA scenarios. Our best single system outperforms other state-of-the-art single systems in both PA and LA of the ASVspoof 2019 corpus.

eess.AS↗

arXiv2021

Bayesian Transformer Language Models for Speech Recognition

State-of-the-art neural language models (LMs) represented by Transformers are highly complex. Their use of fixed, deterministic parameter estimates fail to account for model uncertainty and lead to over-fitting and poor generalization when given limited training data. In order to address these issues, this paper proposes a full Bayesian learning framework for Transformer LM estimation. Efficient variational inference based approaches are used to estimate the latent parameter posterior distributions associated with different parts of the Transformer model architecture including multi-head self-attention, feed forward and embedding layers. Statistically significant word error rate (WER) reductions up to 0.5\% absolute (3.18\% relative) and consistent perplexity gains were obtained over the baseline Transformer LMs on state-of-the-art Switchboard corpus trained LF-MMI factored TDNN systems with i-Vector speaker adaptation. Performance improvements were also obtained on a cross domain LM adaptation task requiring porting a Transformer LM trained on the Switchboard and Fisher data to a low-resource DementiaBank elderly speech corpus.

cs.CL↗

arXiv2021

Neural Architecture Search For LF-MMI Trained Time Delay Neural Networks

Deep neural networks (DNNs) based automatic speech recognition (ASR) systems are often designed using expert knowledge and empirical evaluation. In this paper, a range of neural architecture search (NAS) techniques are used to automatically learn two types of hyper-parameters of state-of-the-art factored time delay neural networks (TDNNs): i) the left and right splicing context offsets; and ii) the dimensionality of the bottleneck linear projection at each hidden layer. These include the DARTS method integrating architecture selection with lattice-free MMI (LF-MMI) TDNN training; Gumbel-Softmax and pipelined DARTS reducing the confusion over candidate architectures and improving the generalization of architecture selection; and Penalized DARTS incorporating resource constraints to adjust the trade-off between performance and system complexity. Parameter sharing among candidate architectures allows efficient search over up to $7^{28}$ different TDNN systems. Experiments conducted on the 300-hour Switchboard corpus suggest the auto-configured systems consistently outperform the baseline LF-MMI TDNN systems using manual network design or random architecture search after LHUC speaker adaptation and RNNLM rescoring. Absolute word error rate (WER) reductions up to 1.0\% and relative model size reduction of 28\% were obtained. Consistent performance improvements were also obtained on a UASpeech disordered speech recognition task using the proposed NAS approaches.

eess.AS↗

arXiv2021

Creation and Evaluation of a Pre-tertiary Artificial Intelligence (AI) Curriculum

Contributions: The Chinese University of Hong Kong (CUHK)-Jockey Club AI for the Future Project (AI4Future) co-created an AI curriculum for pre-tertiary education and evaluated its efficacy. While AI is conventionally taught in tertiary level education, our co-creation process successfully developed the curriculum that has been used in secondary school teaching in Hong Kong and received positive feedback. Background: AI4Future is a cross-sector project that engages five major partners - CUHK Faculty of Engineering and Faculty of Education, Hong Kong secondary schools, the government and the AI industry. A team of 14 professors with expertise in engineering and education collaborated with 17 principals and teachers from 6 secondary schools to co-create the curriculum. This team formation bridges the gap between researchers in engineering and education, together with practitioners in education context. Research Questions: What are the main features of the curriculum content developed through the co-creation process? Would the curriculum significantly improve the students perceived competence in, as well as attitude and motivation towards AI? What are the teachers perceptions of the co-creation process that aims to accommodate and foster teacher autonomy? Methodology: This study adopted a mix of quantitative and qualitative methods and involved 335 student participants. Findings: 1) two main features of learning resources, 2) the students perceived greater competence, and developed more positive attitude to learn AI, and 3) the co-creation process generated a variety of resources which enhanced the teachers knowledge in AI, as well as fostered teachers autonomy in bringing the subject matter into their classrooms.

cs.AI↗

arXiv2021

Emotion controllable speech synthesis using emotion-unlabeled dataset with the assistance of cross-domain speech emotion recognition

Neural text-to-speech (TTS) approaches generally require a huge number of high quality speech data, which makes it difficult to obtain such a dataset with extra emotion labels. In this paper, we propose a novel approach for emotional TTS synthesis on a TTS dataset without emotion labels. Specifically, our proposed method consists of a cross-domain speech emotion recognition (SER) model and an emotional TTS model. Firstly, we train the cross-domain SER model on both SER and TTS datasets. Then, we use emotion labels on the TTS dataset predicted by the trained SER model to build an auxiliary SER task and jointly train it with the TTS model. Experimental results show that our proposed method can generate speech with the specified emotional expressiveness and nearly no hindering on the speech quality.

eess.AS↗

arXiv2021

Unstructured Knowledge Access in Task-oriented Dialog Modeling using Language Inference, Knowledge Retrieval and Knowledge-Integrative Response Generation

Dialog systems enriched with external knowledge can handle user queries that are outside the scope of the supporting databases/APIs. In this paper, we follow the baseline provided in DSTC9 Track 1 and propose three subsystems, KDEAK, KnowleDgEFactor, and Ens-GPT, which form the pipeline for a task-oriented dialog system capable of accessing unstructured knowledge. Specifically, KDEAK performs knowledge-seeking turn detection by formulating the problem as natural language inference using knowledge from dialogs, databases and FAQs. KnowleDgEFactor accomplishes the knowledge selection task by formulating a factorized knowledge/document retrieval problem with three modules performing domain, entity and knowledge level analyses. Ens-GPT generates a response by first processing multiple knowledge snippets, followed by an ensemble algorithm that decides if the response should be solely derived from a GPT2-XL model, or regenerated in combination with the top-ranking knowledge snippet. Experimental results demonstrate that the proposed pipeline system outperforms the baseline and generates high-quality responses, achieving at least 58.77% improvement on BLEU-4 score.

cs.CL↗

arXiv2020

Unsupervised Cross-Lingual Speech Emotion Recognition Using DomainAdversarial Neural Network

By using deep learning approaches, Speech Emotion Recog-nition (SER) on a single domain has achieved many excellentresults. However, cross-domain SER is still a challenging taskdue to the distribution shift between source and target domains.In this work, we propose a Domain Adversarial Neural Net-work (DANN) based approach to mitigate this distribution shiftproblem for cross-lingual SER. Specifically, we add a languageclassifier and gradient reversal layer after the feature extractor toforce the learned representation both language-independent andemotion-meaningful. Our method is unsupervised, i. e., labelson target language are not required, which makes it easier to ap-ply our method to other languages. Experimental results showthe proposed method provides an average absolute improve-ment of 3.91% over the baseline system for arousal and valenceclassification task. Furthermore, we find that batch normaliza-tion is beneficial to the performance gain of DANN. Thereforewe also explore the effect of different ways of data combinationfor batch normalization.

eess.AS↗

arXiv2020

Syntactic representation learning for neural network based TTS with syntactic parse tree traversal

Syntactic structure of a sentence text is correlated with the prosodic structure of the speech that is crucial for improving the prosody and naturalness of a text-to-speech (TTS) system. Nowadays TTS systems usually try to incorporate syntactic structure information with manually designed features based on expert knowledge. In this paper, we propose a syntactic representation learning method based on syntactic parse tree traversal to automatically utilize the syntactic structure information. Two constituent label sequences are linearized through left-first and right-first traversals from constituent parse tree. Syntactic representations are then extracted at word level from each constituent label sequence by a corresponding uni-directional gated recurrent unit (GRU) network. Meanwhile, nuclear-norm maximization loss is introduced to enhance the discriminability and diversity of the embeddings of constituent labels. Upsampled syntactic representations and phoneme embeddings are concatenated to serve as the encoder input of Tacotron2. Experimental results demonstrate the effectiveness of our proposed approach, with mean opinion score (MOS) increasing from 3.70 to 3.82 and ABX preference exceeding by 17% compared with the baseline. In addition, for sentences with multiple syntactic parse trees, prosodic differences can be clearly perceived from the synthesized speeches.

cs.CL↗

arXiv2020

Audio-visual Multi-channel Recognition of Overlapped Speech

Automatic speech recognition (ASR) of overlapped speech remains a highly challenging task to date. To this end, multi-channel microphone array data are widely used in state-of-the-art ASR systems. Motivated by the invariance of visual modality to acoustic signal corruption, this paper presents an audio-visual multi-channel overlapped speech recognition system featuring tightly integrated separation front-end and recognition back-end. A series of audio-visual multi-channel speech separation front-end components based on \textit{TF masking}, \textit{filter\&sum} and \textit{mask-based MVDR} beamforming approaches were developed. To reduce the error cost mismatch between the separation and recognition components, they were jointly fine-tuned using the connectionist temporal classification (CTC) loss function, or a multi-task criterion interpolation with scale-invariant signal to noise ratio (Si-SNR) error cost. Experiments suggest that the proposed multi-channel AVSR system outperforms the baseline audio-only ASR system by up to 6.81\% (26.83\% relative) and 22.22\% (56.87\% relative) absolute word error rate (WER) reduction on overlapped speech constructed using either simulation or replaying of the lipreading sentence 2 (LRS2) dataset respectively.

eess.AS↗

arXiv2020

Improved End-to-End Dysarthric Speech Recognition via Meta-learning Based Model Re-initialization

Dysarthric speech recognition is a challenging task as dysarthric data is limited and its acoustics deviate significantly from normal speech. Model-based speaker adaptation is a promising method by using the limited dysarthric speech to fine-tune a base model that has been pre-trained from large amounts of normal speech to obtain speaker-dependent models. However, statistic distribution mismatches between the normal and dysarthric speech data limit the adaptation performance of the base model. To address this problem, we propose to re-initialize the base model via meta-learning to obtain a better model initialization. Specifically, we focus on end-to-end models and extend the model-agnostic meta learning (MAML) and Reptile algorithms to meta update the base model by repeatedly simulating adaptation to different dysarthric speakers. As a result, the re-initialized model acquires dysarthric speech knowledge and learns how to perform fast adaptation to unseen dysarthric speakers with improved performance. Experimental results on UASpeech dataset show that the best model with proposed methods achieves 54.2% and 7.6% relative word error rate reduction compared with the base model without finetuning and the model directly fine-tuned from the base model, respectively, and it is comparable with the state-of-the-art hybrid DNN-HMM model.

eess.AS↗

arXiv2020

Investigating Robustness of Adversarial Samples Detection for Automatic Speaker Verification

Recently adversarial attacks on automatic speaker verification (ASV) systems attracted widespread attention as they pose severe threats to ASV systems. However, methods to defend against such attacks are limited. Existing approaches mainly focus on retraining ASV systems with adversarial data augmentation. Also, countermeasure robustness against different attack settings are insufficiently investigated. Orthogonal to prior approaches, this work proposes to defend ASV systems against adversarial attacks with a separate detection network, rather than augmenting adversarial data into ASV training. A VGG-like binary classification detector is introduced and demonstrated to be effective on detecting adversarial samples. To investigate detector robustness in a realistic defense scenario where unseen attack settings may exist, we analyze various kinds of unseen attack settings' impact and observe that the detector is robust (6.27\% EER_{det} degradation in the worst case) against unseen substitute ASV systems, but it has weak robustness (50.37\% EER_{det} degradation in the worst case) against unseen perturbation methods. The weak robustness against unseen perturbation methods shows a direction for developing stronger countermeasures.

eess.AS↗

arXiv2020

Speaker Independent and Multilingual/Mixlingual Speech-Driven Talking Head Generation Using Phonetic Posteriorgrams

Generating 3D speech-driven talking head has received more and more attention in recent years. Recent approaches mainly have following limitations: 1) most speaker-independent methods need handcrafted features that are time-consuming to design or unreliable; 2) there is no convincing method to support multilingual or mixlingual speech as input. In this work, we propose a novel approach using phonetic posteriorgrams (PPG). In this way, our method doesn't need hand-crafted features and is more robust to noise compared to recent approaches. Furthermore, our method can support multilingual speech as input by building a universal phoneme space. As far as we know, our model is the first to support multilingual/mixlingual speech as input with convincing results. Objective and subjective experiments have shown that our model can generate high quality animations given speech from unseen languages or speakers and be robust to noise.

eess.AS↗

arXiv2020

Speech-XLNet: Unsupervised Acoustic Model Pretraining For Self-Attention Networks

Self-attention network (SAN) can benefit significantly from the bi-directional representation learning through unsupervised pretraining paradigms such as BERT and XLNet. In this paper, we present an XLNet-like pretraining scheme "Speech-XLNet" for unsupervised acoustic model pretraining to learn speech representations with SAN. The pretrained SAN is finetuned under the hybrid SAN/HMM framework. We conjecture that by shuffling the speech frame orders, the permutation in Speech-XLNet serves as a strong regularizer to encourage the SAN to make inferences by focusing on global structures through its attention weights. In addition, Speech-XLNet also allows the model to explore the bi-directional contexts for effective speech representation learning. Experiments on TIMIT and WSJ demonstrate that Speech-XLNet greatly improves the SAN/HMM performance in terms of both convergence speed and recognition accuracy compared to the one trained from randomly initialized weights. Our best systems achieve a relative improvement of 11.9% and 8.3% on the TIMIT and WSJ tasks respectively. In particular, the best system achieves a phone error rate (PER) of 13.3% on the TIMIT test set, which to our best knowledge, is the lowest PER obtained from a single system.

cs.CL↗

arXiv2020

Transferring Source Style in Non-Parallel Voice Conversion

Voice conversion (VC) techniques aim to modify speaker identity of an utterance while preserving the underlying linguistic information. Most VC approaches ignore modeling of the speaking style (e.g. emotion and emphasis), which may contain the factors intentionally added by the speaker and should be retained during conversion. This study proposes a sequence-to-sequence based non-parallel VC approach, which has the capability of transferring the speaking style from the source speech to the converted speech by explicitly modeling. Objective evaluation and subjective listening tests show superiority of the proposed VC approach in terms of speech naturalness and speaker similarity of the converted speech. Experiments are also conducted to show the source-style transferability of the proposed approach.

eess.AS↗

arXiv2020

Emotional Voice Conversion With Cycle-consistent Adversarial Network

Emotional Voice Conversion, or emotional VC, is a technique of converting speech from one emotion state into another one, keeping the basic linguistic information and speaker identity. Previous approaches for emotional VC need parallel data and use dynamic time warping (DTW) method to temporally align the source-target speech parameters. These approaches often define a minimum generation loss as the objective function, such as L1 or L2 loss, to learn model parameters. Recently, cycle-consistent generative adversarial networks (CycleGAN) have been used successfully for non-parallel VC. This paper investigates the efficacy of using CycleGAN for emotional VC tasks. Rather than attempting to learn a mapping between parallel training data using a frame-to-frame minimum generation loss, the CycleGAN uses two discriminators and one classifier to guide the learning process, where the discriminators aim to differentiate between the natural and converted speech and the classifier aims to classify the underlying emotion from the natural and converted speech. The training process of the CycleGAN models randomly pairs source-target speech parameters, without any temporal alignment operation. The objective and subjective evaluation results confirm the effectiveness of using CycleGAN models for emotional VC. The non-parallel training for a CycleGAN indicates its potential for non-parallel emotional VC.

eess.AS↗