In this work, we propose a training algorithm for an audio-visual automatic speech recognition (AV-ASR) system using deep recurrent neural network (RNN).First, we train a deep RNN acoustic model with a Connectionist Temporal Classification (CTC) objective function. The frame labels obtained from the acoustic model are then used to perform a non-linear dimensionality reduction of the visual features using a deep bottleneck network. Audio and visual features are fused and used to train a fusion RNN. The use of bottleneck features for visual modality helps the model to converge properly during training. Our system is evaluated on GRID corpus. Our results show that presence of visual modality gives significant improvement in character error rate (CER) at various levels of noise even when the model is trained without noisy data. We also provide a comparison of two fusion methods: feature fusion and decision fusion.
Segmental conditional random fields (SCRFs) and connectionist temporal classification (CTC) are two sequence labeling methods used for end-to-end training of speech recognition models. Both models define a transcription probability by marginalizing decisions about latent segmentation alternatives to derive a sequence probability: the former uses a globally normalized joint model of segment labels and durations, and the latter classifies each frame as either an output symbol or a "continuation" of the previous label. In this paper, we train a recognition model by optimizing an interpolation between the SCRF and CTC losses, where the same recurrent neural network (RNN) encoder is used for feature extraction for both outputs. We find that this multitask objective improves recognition accuracy when decoding with either the SCRF or CTC models. Additionally, we show that CTC can also be used to pretrain the RNN encoder, which improves the convergence rate when learning the joint model.
DeepMine is a speech database in Persian and English designed to build and evaluate text-dependent, text-prompted, and text-independent speaker verification, as well as Persian speech recognition systems. It contains more than 1850 speakers and 540 thousand recordings overall, more than 480 hours of speech are transcribed. It is the first public large-scale speaker verification database in Persian, the largest public text-dependent and text-prompted speaker verification database in English, and the largest public evaluation dataset for text-independent speaker verification. It has a good coverage of age, gender, and accents. We provide several evaluation protocols for each part of the database to allow for research on different aspects of speaker verification. We also provide the results of several experiments that can be considered as baselines: HMM-based i-vectors for text-dependent speaker verification, and HMM-based as well as state-of-the-art deep neural network based ASR. We demonstrate that the database can serve for training robust ASR models.
This paper presents novel Weighted Finite-State Transducer (WFST) topologies to implement Connectionist Temporal Classification (CTC)-like algorithms for automatic speech recognition. Three new CTC variants are proposed: (1) the "compact-CTC", in which direct transitions between units are replaced with <epsilon> back-off transitions; (2) the "minimal-CTC", that only adds <blank> self-loops when used in WFST-composition; and (3) "selfless-CTC", that disallows self-loop for non-blank units. The new CTC variants have several benefits, such as reducing decoding graph size and GPU memory required for training while keeping model accuracy.
For speech recognition, deep neural networks (DNNs) have significantly improved the recognition accuracy in most of benchmark datasets and application domains. However, compared to the conventional Gaussian mixture models, DNN-based acoustic models usually have much larger number of model parameters, making it challenging for their applications in resource constrained platforms, e.g., mobile devices. In this paper, we study the application of the recently proposed highway network to train small-footprint DNNs, which are {\it thinner} and {\it deeper}, and have significantly smaller number of model parameters compared to conventional DNNs. We investigated this approach on the AMI meeting speech transcription corpus which has around 70 hours of audio data. The highway neural networks constantly outperformed their plain DNN counterparts, and the number of model parameters can be reduced significantly without sacrificing the recognition accuracy.
Speech and speaker recognition systems are employed in a variety of applications, from personal assistants to telephony surveillance and biometric authentication. The wide deployment of these systems has been made possible by the improved accuracy in neural networks. Like other systems based on neural networks, recent research has demonstrated that speech and speaker recognition systems are vulnerable to attacks using manipulated inputs. However, as we demonstrate in this paper, the end-to-end architecture of speech and speaker systems and the nature of their inputs make attacks and defenses against them substantially different than those in the image space. We demonstrate this first by systematizing existing research in this space and providing a taxonomy through which the community can evaluate future work. We then demonstrate experimentally that attacks against these models almost universally fail to transfer. In so doing, we argue that substantial additional work is required to provide adequate mitigations in this space.
Alzheimer's disease (AD) is a progressive neurodegenerative disease and recently attracts extensive attention worldwide. Speech technology is considered a promising solution for the early diagnosis of AD and has been enthusiastically studied. Most recent works concentrate on the use of advanced BERT-like classifiers for AD detection. Input to these classifiers are speech transcripts produced by automatic speech recognition (ASR) models. The major challenge is that the quality of transcription could degrade significantly under complex acoustic conditions in the real world. The detection performance, in consequence, is largely limited. This paper tackles the problem via tailoring and adapting pre-trained neural-network based ASR model for the downstream AD recognition task. Only bottom layers of the ASR model are retained. A simple fully-connected neural network is added on top of the tailored ASR model for classification. The heavy BERT classifier is discarded. The resulting model is light-weight and can be fine-tuned in an end-to-end manner for AD recognition. Our proposed approach takes only raw speech as input, and no extra transcription process is required. The linguistic information of speech is implicitly encoded in the tailored ASR model and contributes to boosting the performance. Experiments show that our proposed approach outperforms the best manual transcript-based RoBERTa by an absolute margin of 4.6% in terms of accuracy. Our best-performing models achieve the accuracy of 83.2% and 78.0% in the long-audio and short-audio competition tracks of the 2021 NCMMSC Alzheimer's Disease Recognition Challenge, respectively.
Self-attention has been a huge success for many downstream tasks in NLP, which led to exploration of applying self-attention to speech problems as well. The efficacy of self-attention in speech applications, however, seems not fully blown yet since it is challenging to handle highly correlated speech frames in the context of self-attention. In this paper we propose a new neural network model architecture, namely multi-stream self-attention, to address the issue thus make the self-attention mechanism more effective for speech recognition. The proposed model architecture consists of parallel streams of self-attention encoders, and each stream has layers of 1D convolutions with dilated kernels whose dilation rates are unique given stream, followed by a self-attention layer. The self-attention mechanism in each stream pays attention to only one resolution of input speech frames and the attentive computation can be more efficient. In a later stage, outputs from all the streams are concatenated then linearly projected to the final embedding. By stacking the proposed multi-stream self-attention encoder blocks and rescoring the resultant lattices with neural network language models, we achieve the word error rate of 2.2% on the test-clean dataset of the LibriSpeech corpus, the best number reported thus far on the dataset.
Traditional machine learning algorithms are designed to learn in isolation, i.e. address single tasks. The core idea of transfer learning (TL) is that knowledge gained in learning to perform one task (source) can be leveraged to improve learning performance in a related, but different, task (target). TL leverages and transfers previously acquired knowledge to address the expense of data acquisition and labeling, potential computational power limitations, and the dataset distribution mismatches. Although significant progress has been made in the fields of image processing, speech recognition, and natural language processing (for classification and regression) for TL, little work has been done in the field of scientific machine learning for functional regression and uncertainty quantification in partial differential equations. In this work, we propose a novel TL framework for task-specific learning under conditional shift with a deep operator network (DeepONet). Inspired by the conditional embedding operator theory, we measure the statistical distance between the source domain and the target feature domain by embedding conditional distributions onto a reproducing kernel Hilbert space. Task-specific operator learning is accomplished by fine-tuning task-specific layers of the target DeepONet using a hybrid loss function that allows for the matching of individual target samples while also preserving the global properties of the conditional distribution of target data. We demonstrate the advantages of our approach for various TL scenarios involving nonlinear PDEs under conditional shift. Our results include geometry domain adaptation and show that the proposed TL framework enables fast and efficient multi-task operator learning, despite significant differences between the source and target domains.
Recent studies have shown that deep neural networks (DNNs) perform significantly better than shallow networks and Gaussian mixture models (GMMs) on large vocabulary speech recognition tasks. In this paper, we argue that the improved accuracy achieved by the DNNs is the result of their ability to extract discriminative internal representations that are robust to the many sources of variability in speech signals. We show that these representations become increasingly insensitive to small perturbations in the input with increasing network depth, which leads to better speech recognition performance with deeper networks. We also show that DNNs cannot extrapolate to test samples that are substantially different from the training examples. If the training data are sufficiently representative, however, internal features learned by the DNN are relatively stable with respect to speaker differences, bandwidth differences, and environment distortion. This enables DNN-based recognizers to perform as well or better than state-of-the-art systems based on GMMs or shallow networks without the need for explicit model adaptation or feature normalization.