There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality.

Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

I and i

In this paper we present an end-to-end speech recognition model with Transformer encoders that can be used in a streaming speech recognition system. Transformer computation blocks based on self-attention are used to encode both audio and label sequences independently. The activations from both audio and label encoders are combined with a feed-forward layer to compute a probability distribution over the label space for every combination of acoustic frame position and label history. This is similar to the Recurrent Neural Network Transducer (RNN-T) model, which uses RNNs for information encoding instead of Transformer encoders. The model is trained with the RNN-T loss well-suited to streaming decoding. We present results on the LibriSpeech dataset showing that limiting the left context for self-attention in the Transformer layers makes decoding computationally tractable for streaming, with only a slight degradation in accuracy. We also show that the full attention version of our model beats the-state-of-the art accuracy on the LibriSpeech benchmarks. Our results also show that we can bridge the gap between full attention and limited attention versions of our model by attending to a limited number of future frames.

Transformer Transducer: A Streamable Speech Recognition Model with Transformer Encoders and RNN-T Loss

We present Mockingjay as a new speech representation learning approach, where bidirectional Transformer encoders are pre-trained on a large amount of unlabeled speech. Previous speech representation methods learn through conditioning on past frames and predicting information about future frames. Whereas Mockingjay is designed to predict the current frame through jointly conditioning on both past and future contexts. The Mockingjay representation improves performance for a wide range of downstream tasks, including phoneme classification, speaker recognition, and sentiment classification on spoken content, while outperforming other approaches. Mockingjay is empirically powerful and can be fine-tuned with downstream models, with only 2 epochs we further improve performance dramatically. In a low resource setting with only 0.1% of labeled data, we outperform the result of Mel-features that uses all 100% labeled data.

Mockingjay: Unsupervised Speech Representation Learning with Deep Bidirectional Transformer Encoders

We introduce a self-supervised speech pre-training method called TERA, which stands for Transformer Encoder Representations from Alteration. Recent approaches often learn through the formulation of a single auxiliary task like contrastive prediction, autoregressive prediction, or masked reconstruction. Unlike previous approaches, we use a multi-target auxiliary task to pre-train Transformer Encoders on a large amount of unlabeled speech. The model learns through the reconstruction of acoustic frames from its altered counterpart, where we use a stochastic policy to alter along three dimensions: temporal, channel, and magnitude. TERA can be used to extract speech representations or fine-tune with downstream models. We evaluate TERA on several downstream tasks, including phoneme classification, speaker recognition, and speech recognition. TERA achieved strong performance on these tasks by improving upon surface features and outperforming previous methods. In our experiments, we show that through alteration along different dimensions, the model learns to encode distinct aspects of speech. We explore different knowledge transfer methods to incorporate the pre-trained model with downstream models. Furthermore, we show that the proposed method can be easily transferred to another dataset not used in pre-training.

TERA: Self-Supervised Learning of Transformer Encoder Representation for Speech

Deep learning is revolutionizing how machine translation systems are built today This book introduces the challenge of machine translation and evaluation - including historical, linguistic, and applied context -- then develops the core deep learning methods used for natural language applications Code examples in Python give readers a hands-on blueprint for understanding and implementing their own machine translation systems The book also provides extensive coverage of machine learning tricks, issues involved in handling various forms of data, model enhancements, and current challenges and methods for analysis and visualization Summaries of the current research in the field make this a state-of-the-art textbook for undergraduate and graduate classes, as well as an essential reference for researchers and developers interested in other applications of neural methods in the broader field of human language processing

Neural Machine Translation

Recently, end-to-end sequence-to-sequence models for speech recognition have gained significant interest in the research community. While previous architecture choices revolve around time-delay neural networks (TDNN) and long short-term memory (LSTM) recurrent neural networks, we propose to use self-attention via the Transformer architecture as an alternative. Our analysis shows that deep Transformer networks with high learning capacity are able to exceed performance from previous end-to-end approaches and even match the conventional hybrid systems. Moreover, we trained very deep models with up to 48 Transformer layers for both encoder and decoders combined with stochastic residual connections, which greatly improve generalizability and training efficiency. The resulting models outperform all previous end-to-end ASR approaches on the Switchboard benchmark. An ensemble of these models achieve 9.9% and 17.7% WER on Switchboard and CallHome test sets respectively. This finding brings our end-to-end models to competitive levels with previous hybrid systems. Further, with model ensembling the Transformers can outperform certain hybrid systems, which are more complicated in terms of both structure and training procedure.

Very Deep Self-Attention Networks for End-to-End Speech Recognition.

Sequence-to-Sequence (S2S) models recently started to show state-of-the-art performance for automatic speech recognition (ASR). With these large and deep models overfitting remains the largest problem, outweighing performance improvements that can be obtained from better architectures. One solution to the overfitting problem is increasing the amount of available training data and the variety exhibited by the training data with the help of data augmentation. In this paper we examine the influence of three data augmentation methods on the performance of two S2S model architectures. One of the data augmentation method comes from literature, while two other methods are our own development – a time perturbation in the frequency domain and sub-sequence sampling. Our experiments on Switchboard and Fisher data show state-of-the-art performance for S2S models that are trained solely on the speech training data and do not use additional text data.

Improving Sequence-To-Sequence Speech Recognition Training with On-The-Fly Data Augmentation

Very Deep Self-Attention Networks for End-to-End Speech Recognition

Latency is one of the main challenges in the task of simultaneous spoken language translation. While significant improvements in recent years have led to high quality automatic translations, their usefulness in real-time settings is still severely limited due to the large delay between the input speech and the delivered translation. In this paper, we present a novel scheme which reduces the latency of a large scale speech translation system drastically. Within this scheme, the transcribed text and its translation can be updated when more context is available, even after they are presented to the user. Thereby, this scheme allows us to display an initial transcript and its translation to the user with a very low latency. If necessary, both transcript and translation can later be updated to better, more accurate versions until eventually the final versions are displayed. Using this framework, we are able to reduce the latency of the source language transcript into half. For the translation, an average delay of 3.3s was achieved, which is more than twice as fast as our initial system.

Dynamic transcription for low-latency speech translation

This paper describes KIT’s submission to the IWSLT 2019 Speech Translation task on two sub-tasks corresponding to two different datasets. We investigate different end-to-end architectures for the speech recognition module, including our new transformer-based architectures. Overall, our modules in the pipe-line are based on the transformer architecture which has recently achieved great results in various fields. In our systems, using transformer is also advantageous compared to traditional hybrid systems in term of simplicity while still having competent results.

Thai-Son Nguyen

Papers

Very Deep Self-Attention Networks for End-to-End Speech Recognition.

Improving Sequence-To-Sequence Speech Recognition Training with On-The-Fly Data Augmentation

Very Deep Self-Attention Networks for End-to-End Speech Recognition

Dynamic transcription for low-latency speech translation

The IWSLT 2019 KIT Speech Translation System