arXiv subjects

Duc Le

Publications and source records attributed to Duc Le.

50 records · Page 3Linked to original sources

arXiv2020

Emformer: Efficient Memory Transformer Based Acoustic Model For Low Latency Streaming Speech Recognition

This paper proposes an efficient memory transformer Emformer for low latency streaming speech recognition. In Emformer, the long-range history context is distilled into an augmented memory bank to reduce self-attention's computation complexity. A cache mechanism saves the computation for the key and value in self-attention for the left context. Emformer applies a parallelized block processing in training to support low latency models. We carry out experiments on benchmark LibriSpeech data. Under average latency of 960 ms, Emformer gets WER $2.50\%$ on test-clean and $5.62\%$ on test-other. Comparing with a strong baseline augmented memory transformer (AM-TRF), Emformer gets $4.6$ folds training speedup and $18\%$ relative real-time factor (RTF) reduction in decoding with relative WER reduction $17\%$ on test-clean and $9\%$ on test-other. For a low latency scenario with an average latency of 80 ms, Emformer achieves WER $3.01\%$ on test-clean and $7.09\%$ on test-other. Comparing with the LSTM baseline with the same latency and model size, Emformer gets relative WER reduction $9\%$ and $16\%$ on test-clean and test-other, respectively.

cs.SD↗

arXiv2020

Deep Shallow Fusion for RNN-T Personalization

End-to-end models in general, and Recurrent Neural Network Transducer (RNN-T) in particular, have gained significant traction in the automatic speech recognition community in the last few years due to their simplicity, compactness, and excellent performance on generic transcription tasks. However, these models are more challenging to personalize compared to traditional hybrid systems due to the lack of external language models and difficulties in recognizing rare long-tail words, specifically entity names. In this work, we present novel techniques to improve RNN-T's ability to model rare WordPieces, infuse extra information into the encoder, enable the use of alternative graphemic pronunciations, and perform deep fusion with personalized language models for more robust biasing. We show that these combined techniques result in 15.4%-34.5% relative Word Error Rate improvement compared to a strong RNN-T baseline which uses shallow fusion and text-to-speech augmentation. Our work helps push the boundary of RNN-T personalization and close the gap with hybrid systems on use cases where biasing and entity recognition are crucial.

cs.CL↗

arXiv2020

Improving RNN Transducer Based ASR with Auxiliary Tasks

End-to-end automatic speech recognition (ASR) models with a single neural network have recently demonstrated state-of-the-art results compared to conventional hybrid speech recognizers. Specifically, recurrent neural network transducer (RNN-T) has shown competitive ASR performance on various benchmarks. In this work, we examine ways in which RNN-T can achieve better ASR accuracy via performing auxiliary tasks. We propose (i) using the same auxiliary task as primary RNN-T ASR task, and (ii) performing context-dependent graphemic state prediction as in conventional hybrid modeling. In transcribing social media videos with varying training data size, we first evaluate the streaming ASR performance on three languages: Romanian, Turkish and German. We find that both proposed methods provide consistent improvements. Next, we observe that both auxiliary tasks demonstrate efficacy in learning deep transformer encoders for RNN-T criterion, thus achieving competitive results - 2.0%/4.2% WER on LibriSpeech test-clean/other - as compared to prior top performing models.

cs.CL↗

arXiv2020

Alignment Restricted Streaming Recurrent Neural Network Transducer

There is a growing interest in the speech community in developing Recurrent Neural Network Transducer (RNN-T) models for automatic speech recognition (ASR) applications. RNN-T is trained with a loss function that does not enforce temporal alignment of the training transcripts and audio. As a result, RNN-T models built with uni-directional long short term memory (LSTM) encoders tend to wait for longer spans of input audio, before streaming already decoded ASR tokens. In this work, we propose a modification to the RNN-T loss function and develop Alignment Restricted RNN-T (Ar-RNN-T) models, which utilize audio-text alignment information to guide the loss computation. We compare the proposed method with existing works, such as monotonic RNN-T, on LibriSpeech and in-house datasets. We show that the Ar-RNN-T loss provides a refined control to navigate the trade-offs between the token emission delays and the Word Error Rate (WER). The Ar-RNN-T models also improve downstream applications such as the ASR End-pointing by guaranteeing token emissions within any given range of latency. Moreover, the Ar-RNN-T loss allows for bigger batch sizes and 4 times higher throughput for our LSTM model architecture, enabling faster training and convergence on GPUs.

cs.CL↗

arXiv2020

Improved Neural Language Model Fusion for Streaming Recurrent Neural Network Transducer

Recurrent Neural Network Transducer (RNN-T), like most end-to-end speech recognition model architectures, has an implicit neural network language model (NNLM) and cannot easily leverage unpaired text data during training. Previous work has proposed various fusion methods to incorporate external NNLMs into end-to-end ASR to address this weakness. In this paper, we propose extensions to these techniques that allow RNN-T to exploit external NNLMs during both training and inference time, resulting in 13-18% relative Word Error Rate improvement on Librispeech compared to strong baselines. Furthermore, our methods do not incur extra algorithmic latency and allow for flexible plug-and-play of different NNLMs without re-training. We also share in-depth analysis to better understand the benefits of the different NNLM fusion methods. Our work provides a reliable technique for leveraging unpaired text data to significantly improve RNN-T while keeping the system streamable, flexible, and lightweight.

cs.CL↗

arXiv2020

Evidence for the topological order in a kagome antiferromagnet

A Z2 quantum spin liquid hosts one of the simplest topological orders and exhibits many exotic properties due to long-range quantum entanglements. Its elementary excitations are anyons such as spinons carrying fractionalized spin quantum number and visons carrying emergent Z2 gauge flux. However, experimental detection of these anyons remains elusive. The difficulties lie not only in the fact that there exists few candidates for Z2 quantum spin liquids but also in that visons are magnetically inert hence immune to available experimental techniques. Here we have studied the spin excitations and specific heats of kagome-lattice antiferromagnet Cu$_4$(OH)$_6$FBr and Cu$_3$Zn(OH)$_6$FBr, which consists of two-dimensional Cu$^{2+}$ kagome layers with either Cu$^{2+}$ or Zn$^{2+}$ ions in between. By combining the first principle calculations and inelastic neutron scattering data in the former, we show that the dominate couplings in Cu$_4$(OH)$_6$FBr are between the nearest neighbor spins within the kagome planes, and the kagome and interlayer spin systems are essentially decoupled above the antiferromagnetic transition temperature. The intrinsic spin excitations and specific heats of the kagome layers for Cu$_3$Zn(OH)$_6$FBr are thus derived by removing the contributions from the residual interlayer Cu$^{2+}$ magnetic impurities. Accordingly, the kagome spin system exhibits spin continuum with momentum-dependent spin gap and a large magnetic entropy at low temperature that is insensitive to magnetic field, which can be understood as the evidences of spinons and visons in this kagome quantum spin liquid candidate. Our results suggest the existence of the Z2 anyons in the material, and therefore provide a comprehensive set of evidences for the Z2 topological order in kagome quantum spin liquid and bring their choreographed entanglement dances to the stage of real materials.

cond-mat.str-el↗

arXiv2020

Classification of Huntington Disease using Acoustic and Lexical Features

Speech is a critical biomarker for Huntington Disease (HD), with changes in speech increasing in severity as the disease progresses. Speech analyses are currently conducted using either transcriptions created manually by trained professionals or using global rating scales. Manual transcription is both expensive and time-consuming and global rating scales may lack sufficient sensitivity and fidelity. Ultimately, what is needed is an unobtrusive measure that can cheaply and continuously track disease progression. We present first steps towards the development of such a system, demonstrating the ability to automatically differentiate between healthy controls and individuals with HD using speech cues. The results provide evidence that objective analyses can be used to support clinical diagnoses, moving towards the tracking of symptomatology outside of laboratory and clinical environments.

eess.AS↗

arXiv2020

Weak-Attention Suppression For Transformer Based Speech Recognition

Transformers, originally proposed for natural language processing (NLP) tasks, have recently achieved great success in automatic speech recognition (ASR). However, adjacent acoustic units (i.e., frames) are highly correlated, and long-distance dependencies between them are weak, unlike text units. It suggests that ASR will likely benefit from sparse and localized attention. In this paper, we propose Weak-Attention Suppression (WAS), a method that dynamically induces sparsity in attention probabilities. We demonstrate that WAS leads to consistent Word Error Rate (WER) improvement over strong transformer baselines. On the widely used LibriSpeech benchmark, our proposed method reduced WER by 10%$ on test-clean and 5% on test-other for streamable transformers, resulting in a new state-of-the-art among streaming models. Further analysis shows that WAS learns to suppress attention of non-critical and redundant continuous acoustic frames, and is more likely to suppress past frames rather than future ones. It indicates the importance of lookahead in attention-based ASR models.

eess.AS↗

arXiv2020

Transformer-based Acoustic Modeling for Hybrid Speech Recognition

We propose and evaluate transformer-based acoustic models (AMs) for hybrid speech recognition. Several modeling choices are discussed in this work, including various positional embedding methods and an iterated loss to enable training deep transformers. We also present a preliminary study of using limited right context in transformer models, which makes it possible for streaming applications. We demonstrate that on the widely used Librispeech benchmark, our transformer-based AM outperforms the best published hybrid result by 19% to 26% relative when the standard n-gram language model (LM) is used. Combined with neural network LM for rescoring, our proposed approach achieves state-of-the-art results on Librispeech. Our findings are also confirmed on a much larger internal dataset.

cs.CL↗

arXiv2020

G2G: TTS-Driven Pronunciation Learning for Graphemic Hybrid ASR

Grapheme-based acoustic modeling has recently been shown to outperform phoneme-based approaches in both hybrid and end-to-end automatic speech recognition (ASR), even on non-phonemic languages like English. However, graphemic ASR still has problems with rare long-tail words that do not follow the standard spelling conventions seen in training, such as entity names. In this work, we present a novel method to train a statistical grapheme-to-grapheme (G2G) model on text-to-speech data that can rewrite an arbitrary character sequence into more phonetically consistent forms. We show that using G2G to provide alternative pronunciations during decoding reduces Word Error Rate by 3% to 11% relative over a strong graphemic baseline and bridges the gap on rare name recognition with an equivalent phonetic setup. Unlike many previously proposed methods, our method does not require any change to the acoustic model training procedure. This work reaffirms the efficacy of grapheme-based modeling and shows that specialized linguistic knowledge, when available, can be leveraged to improve graphemic ASR.

eess.AS↗

arXiv2019

Transformer-Transducer: End-to-End Speech Recognition with Self-Attention

We explore options to use Transformer networks in neural transducer for end-to-end speech recognition. Transformer networks use self-attention for sequence modeling and comes with advantages in parallel computation and capturing contexts. We propose 1) using VGGNet with causal convolution to incorporate positional information and reduce frame rate for efficient inference 2) using truncated self-attention to enable streaming for Transformer and reduce computational complexity. All experiments are conducted on the public LibriSpeech corpus. The proposed Transformer-Transducer outperforms neural transducer with LSTM/BLSTM networks and achieved word error rates of 6.37 % on the test-clean set and 15.30 % on the test-other set, while remaining streamable, compact with 45.7M parameters for the entire system, and computationally efficient with complexity of O(T), where T is input sequence length.

eess.AS↗

arXiv2019

From Senones to Chenones: Tied Context-Dependent Graphemes for Hybrid Speech Recognition

There is an implicit assumption that traditional hybrid approaches for automatic speech recognition (ASR) cannot directly model graphemes and need to rely on phonetic lexicons to get competitive performance, especially on English which has poor grapheme-phoneme correspondence. In this work, we show for the first time that, on English, hybrid ASR systems can in fact model graphemes effectively by leveraging tied context-dependent graphemes, i.e., chenones. Our chenone-based systems significantly outperform equivalent senone baselines by 4.5% to 11.1% relative on three different English datasets. Our results on Librispeech are state-of-the-art compared to other hybrid approaches and competitive with previously published end-to-end numbers. Further analysis shows that chenones can better utilize powerful acoustic models and large training data, and require context- and position-dependent modeling to work well. Chenone-based systems also outperform senone baselines on proper noun and rare word recognition, an area where the latter is traditionally thought to have an advantage. Our work provides an alternative for end-to-end ASR and establishes that hybrid systems can be improved by dropping the reliance on phonetic knowledge.

eess.AS↗

arXiv2017

Uncovering Architectural Design Decisions

Over the past three decades, considerable effort has been devoted to the study of software architecture. A major portion of this effort has focused on the originally proposed view of four "C"s---components, connectors, configurations, and constraints---that are the building blocks of a system's architecture. Despite being simple and appealing, this view has proven to be incomplete and has required further elaboration. To that end, researchers have more recently tried to approach architectures from another important perspective---that of design decisions that yield a system's architecture. These more recent efforts have lacked a precise understanding of several key questions, however: (1) What is an architectural design decision (definition)? (2) How can architectural design decisions be found in existing systems (identification)? (3) What system decisions are and are not architectural (classification)? (4) How are architectural design decisions manifested in the code (reification)? (5) How can important architectural decisions be preserved and/or changed as desired (evolution)? This paper presents a technique targeted at answering these questions by analyzing information that is readily available about software systems. We applied our technique on over 100 different versions of two widely adopted open- source systems, and found that it can accurately uncover the architectural design decisions embodied in the systems.

cs.SE↗

arXiv2011

Solitonic lattice and Yukawa forces in the rare earth orthoferrite TbFeO3

The control of domains in ferroic devices lies at the heart of their potential for technological applications. Multiferroic materials offer another level of complexity as domains can be either or both of a ferroelectric and magnetic nature. Here we report the discovery of a novel magnetic state in the orthoferrite TbFeO3 using neutron diffraction under an applied magnetic field. This state has a very long incommensurate period ranging from 340 Angstrom at 3K to 2700 Angstrom at the lowest temperatures and exhibits an anomalously large number of higher-order harmonics, allowing us to identify it with the periodic array of sharp domain walls of Tb spins separated by many lattice constants. The Tb domain walls interact by exchanging spin waves propagating through the Fe magnetic sublattice. The resulting Yukawa-like force, familiar from particle physics, has a finite range that determines the period of the incommensurate state.

cond-mat.str-el↗