While deep learning technologies are now capable of generating realistic images confusing humans, the research efforts are turning to the synthesis of images for more concrete and application-specific purposes. Facial image generation based on vocal characteristics from speech is one of such important yet challenging tasks. It is the key enabler to influential use cases of image generation, especially for business in public security and entertainment. Existing solutions to the problem of speech2face renders limited image quality and fails to preserve facial similarity due to the lack of quality dataset for training and appropriate integration of vocal features. In this paper, we investigate these key technical challenges and propose Speech Fusion to Face, or SF2F in short, attempting to address the issue of facial image quality and the poor connection between vocal feature domain and modern image generation models. By adopting new strategies on data model and training, we demonstrate dramatic performance boost over state-of-the-art solution, by doubling the recall of individual identity, and lifting the quality score from 15 to 19 based on the mutual information score with VGGFace classifier.
An end-to-end (E2E) speech recognition model implicitly learns a biased internal language model (ILM) during training. To fused an external LM during inference, the scores produced by the biased ILM need to be estimated and subtracted. In this paper we propose two novel approaches to estimate the biased ILM based on Listen-Attend-Spell (LAS) models. The simpler method is to replace the context vector of the LAS decoder at every time step with a learnable vector. The other more advanced method is to use a simple feed-forward network to directly map query vectors to context vectors, making the generation of the context vectors independent of the LAS encoder. Both the learnable vector and the mapping network are trained on the transcriptions of the training data to minimize the perplexity while all the other parameters of the LAS model is fixed. Experiments show that the ILMs estimated by the proposed methods achieve the lowest perplexity. In addition, they also significantly outperform the shallow fusion method and two previously proposed Internal Language Model Estimation (ILME) approaches on multiple datasets.
This paper argues that training GANs on local and non-local dependencies in speech data offers insights into how deep neural networks discretize continuous data and how symbolic-like rule-based morphophonological processes emerge in a deep convolutional architecture. Acquisition of speech has recently been modeled as a dependency between latent space and data generated by GANs in Begu\v{s} (arXiv:2006.03965), who models learning of a simple local allophonic distribution. We extend this approach to test learning of local and non-local phonological processes that include approximations of morphological processes. We further parallel outputs of the model to results of a behavioral experiment where human subjects are trained on the data used for training the GAN network. Four main conclusions emerge: (i) the networks provide useful information for computational models of language acquisition even if trained on a comparatively small dataset of an artificial grammar learning experiment; (ii) local processes are easier to learn than non-local processes, which matches both behavioral data in human subjects and typology in the world's languages. This paper also proposes (iii) how we can actively observe the network's progress in learning and explore the effect of training steps on learning representations by keeping latent space constant across different training steps. Finally, this paper shows that (iv) the network learns to encode the presence of a prefix with a single latent variable; by interpolating this variable, we can actively observe the operation of a non-local phonological process. The proposed technique for retrieving learning representations has general implications for our understanding of how GANs discretize continuous speech data and suggests that rule-like generalizations in the training data are represented as an interaction between variables in the network's latent space.
In recent years, end-to-end (E2E) based automatic speech recognition (ASR) systems have achieved great success due to their simplicity and promising performance. Neural Transducer based models are increasingly popular in streaming E2E based ASR systems and have been reported to outperform the traditional hybrid system in some scenarios. However, the joint optimization of acoustic model, lexicon and language model in neural Transducer also brings about challenges to utilize pure text for language model adaptation. This drawback might prevent their potential applications in practice. In order to address this issue, in this paper, we propose a novel model, factorized neural Transducer, by factorizing the blank and vocabulary prediction, and adopting a standalone language model for the vocabulary prediction. It is expected that this factorization can transfer the improvement of the standalone language model to the Transducer for speech recognition, which allows various language model adaptation techniques to be applied. We demonstrate that the proposed factorized neural Transducer yields 15% to 20% WER improvements when out-of-domain text data is used for language model adaptation, at the cost of a minor degradation in WER on a general test set.
In this paper, we present a method for learning discrete linguistic units by incorporating vector quantization layers into neural models of visually grounded speech. We show that our method is capable of capturing both word-level and sub-word units, depending on how it is configured. What differentiates this paper from prior work on speech unit learning is the choice of training objective. Rather than using a reconstruction-based loss, we use a discriminative, multimodal grounding objective which forces the learned units to be useful for semantic image retrieval. We evaluate the sub-word units on the ZeroSpeech 2019 challenge, achieving a 27.3\% reduction in ABX error rate over the top-performing submission, while keeping the bitrate approximately the same. We also present experiments demonstrating the noise robustness of these units. Finally, we show that a model with multiple quantizers can simultaneously learn phone-like detectors at a lower layer and word-like detectors at a higher layer. We show that these detectors are highly accurate, discovering 279 words with an F1 score of greater than 0.5.
In recent years, studies on automatic speech recognition (ASR) have shown outstanding results that reach human parity on short speech segments. However, there are still difficulties in standardizing the output of ASR such as capitalization and punctuation restoration for long-speech transcription. The problems obstruct readers to understand the ASR output semantically and also cause difficulties for natural language processing models such as NER, POS and semantic parsing. In this paper, we propose a method to restore the punctuation and capitalization for long-speech ASR transcription. The method is based on Transformer models and chunk merging that allows us to (1), build a single model that performs punctuation and capitalization in one go, and (2), perform decoding in parallel while improving the prediction accuracy. Experiments on British National Corpus showed that the proposed approach outperforms existing methods in both accuracy and decoding speed.
Stuttering is a speech impediment affecting tens of millions of people on an everyday basis. Even with its commonality, there is minimal data and research on the identification and classification of stuttered speech. This paper tackles the problem of detection and classification of different forms of stutter. As opposed to most existing works that identify stutters with language models, our work proposes a model that relies solely on acoustic features, allowing for identification of several variations of stutter disfluencies without the need for speech recognition. Our model uses a deep residual network and bidirectional long short-term memory layers to classify different types of stutters and achieves an average miss rate of 10.03%, outperforming the state-of-the-art by almost 27%
Using phonological speech vocoding, we propose a platform for exploring relations between phonology and speech processing, and in broader terms, for exploring relations between the abstract and physical structures of a speech signal. Our goal is to make a step towards bridging phonology and speech processing and to contribute to the program of Laboratory Phonology. We show three application examples for laboratory phonology: compositional phonological speech modelling, a comparison of phonological systems and an experimental phonological parametric text-to-speech (TTS) system. The featural representations of the following three phonological systems are considered in this work: (i) Government Phonology (GP), (ii) the Sound Pattern of English (SPE), and (iii) the extended SPE (eSPE). Comparing GP- and eSPE-based vocoded speech, we conclude that the latter achieves slightly better results than the former. However, GP - the most compact phonological speech representation - performs comparably to the systems with a higher number of phonological features. The parametric TTS based on phonological speech representation, and trained from an unlabelled audiobook in an unsupervised manner, achieves intelligibility of 85% of the state-of-the-art parametric speech synthesis. We envision that the presented approach paves the way for researchers in both fields to form meaningful hypotheses that are explicitly testable using the concepts developed and exemplified in this paper. On the one hand, laboratory phonologists might test the applied concepts of their theoretical models, and on the other hand, the speech processing community may utilize the concepts developed for the theoretical phonological models for improvements of the current state-of-the-art applications.
In this paper, we discuss the initial attempts at boosting understanding human language based on deep-learning models with quantum computing. We successfully train a quantum-enhanced Long Short-Term Memory network to perform the parts-of-speech tagging task via numerical simulations. Moreover, a quantum-enhanced Transformer is proposed to perform the sentiment analysis based on the existing dataset.
The most common way for humans to communicate is by speech. But perhaps a language system cannot know what it is communicating without a connection to the real world by image perception. In fact, humans perceive these multiple sources of information together to build a general concept. However, constructing a machine that can alleviate these modalities together in a supervised learning fashion is difficult, because a parallel dataset is required among speech, image, and text modalities altogether that is often unavailable. A machine speech chain based on sequence-to-sequence deep learning was previously proposed to achieve semi-supervised learning that enabled automatic speech recognition (ASR) and text-to-speech synthesis (TTS) to teach each other when they receive unpaired data. In this research, we take a further step by expanding the speech chain into a multimodal chain and design a closely knit chain architecture that connects ASR, TTS, image captioning (IC), and image retrieval (IR) models into a single framework. ASR, TTS, IC, and IR components can be trained in a semi-supervised fashion by assisting each other given incomplete datasets and leveraging cross-modal data augmentation within the chain.