Alexander Rudnicky

Li-Wei Chen, Shinji Watanabe, Alexander Rudnicky

Recent Text-to-Speech (TTS) systems trained on reading or acted corpora have achieved near human-level naturalness. The diversity of human speech, however, often goes beyond the coverage of these corpora. We believe the ability to handle such diversity is crucial for AI systems to achieve human-level communication. Our work explores the use of more abundant real-world data for building speech synthesizers. We train TTS systems using real-world speech from YouTube and podcasts. We observe the mismatch between training and inference alignments in mel-spectrogram based autoregressive models, leading to unintelligible synthesis, and demonstrate that learned discrete codes within multiple code groups effectively resolves this issue. We introduce our MQTTS system whose architecture is designed for multiple code generation and monotonic alignment, along with the use of a clean silence prompt to improve synthesis quality. We conduct ablation analyses to identify the efficacy of our methods. We show that MQTTS outperforms existing TTS systems in several objective and subjective measures.

Li-Wei Chen, Shinji Watanabe, Alexander Rudnicky

We present a unified system to realize one-shot voice conversion (VC) on the pitch, rhythm, and speaker attributes. Existing works generally ignore the correlation between prosody and language content, leading to the degradation of naturalness in converted speech. Additionally, the lack of proper language features prevents these systems from accurately preserving language content after conversion. To address these issues, we devise a cascaded modular system leveraging self-supervised discrete speech units as language representation. These discrete units provide duration information essential for rhythm modeling. Our system first extracts utterance-level prosody and speaker representations from the raw waveform. Given the prosody representation, a prosody predictor estimates pitch, energy, and duration for each discrete unit in the utterance. A synthesizer further reconstructs speech based on the predicted prosody, speaker representation, and discrete units. Experiments show that our system outperforms previous approaches in naturalness, intelligibility, speaker transferability, and prosody transferability. Code and samples are publicly available.

Mario Rodríguez-Cantelar, Chen Zhang, Chengguang Tang et al.

The advent and fast development of neural networks have revolutionized the research on dialogue systems and subsequently have triggered various challenges regarding their automatic evaluation. Automatic evaluation of open-domain dialogue systems as an open challenge has been the center of the attention of many researchers. Despite the consistent efforts to improve automatic metrics' correlations with human evaluation, there have been very few attempts to assess their robustness over multiple domains and dimensions. Also, their focus is mainly on the English language. All of these challenges prompt the development of automatic evaluation metrics that are reliable in various domains, dimensions, and languages. This track in the 11th Dialogue System Technology Challenge (DSTC11) is part of the ongoing effort to promote robust and multilingual automatic evaluation metrics. This article describes the datasets and baselines provided to participants and discusses the submission and result details of the two proposed subtasks.

Li-Wei Chen, Takuya Higuchi, Zakaria Aldeneh et al.

The success of large language models in text processing has inspired their adaptation to speech modeling. However, since speech is continuous and complex, it is often discretized for autoregressive modeling. Speech tokens derived from self-supervised models (known as semantic tokens) typically focus on the linguistic aspects of speech but neglect prosodic information. As a result, models trained on these tokens can generate speech with reduced naturalness. Existing approaches try to fix this by adding pitch features to the semantic tokens. However, pitch alone cannot fully represent the range of paralinguistic attributes, and selecting the right features requires careful hand-engineering. To overcome this, we propose an end-to-end variational approach that automatically learns to encode these continuous speech attributes to enhance the semantic tokens. Our approach eliminates the need for manual extraction and selection of paralinguistic features. Moreover, it produces preferred speech continuations according to human raters. Code, samples and models are available at https://github.com/b04901014/vae-gslm.

Chen Zhang, João Sedoc, Luis Fernando D'Haro et al.

The development of Open-Domain Dialogue Systems (ODS)is a trending topic due to the large number of research challenges, large societal and business impact, and advances in the underlying technology. However, the development of these kinds of systems requires two important characteristics:1) automatic evaluation mechanisms that show high correlations with human judgements across multiple dialogue evaluation aspects (with explainable features for providing constructive and explicit feedback on the quality of generative models' responses for quick development and deployment)and 2) mechanisms that can help to control chatbot responses,while avoiding toxicity and employing intelligent ways to handle toxic user comments and keeping interaction flow and engagement. This track at the 10th Dialogue System Technology Challenge (DSTC10) is part of the ongoing effort to promote scalable and toxic-free ODS. This paper describes the datasets and baselines provided to participants, as well as submission evaluation results for each of the two proposed subtasks.

Li-Wei Chen, Alexander Rudnicky

While Wav2Vec 2.0 has been proposed for speech recognition (ASR), it can also be used for speech emotion recognition (SER); its performance can be significantly improved using different fine-tuning strategies. Two baseline methods, vanilla fine-tuning (V-FT) and task adaptive pretraining (TAPT) are first presented. We show that V-FT is able to outperform state-of-the-art models on the IEMOCAP dataset. TAPT, an existing NLP fine-tuning strategy, further improves the performance on SER. We also introduce a novel fine-tuning method termed P-TAPT, which modifies the TAPT objective to learn contextualized emotion representations. Experiments show that P-TAPT performs better than TAPT, especially under low-resource settings. Compared to prior works in this literature, our top-line system achieved a 7.4\% absolute improvement in unweighted accuracy (UA) over the state-of-the-art performance on IEMOCAP. Our code is publicly available.

Li-Wei Chen, Alexander Rudnicky

In this paper, we present a novel architecture to realize fine-grained style control on the transformer-based text-to-speech synthesis (TransformerTTS). Specifically, we model the speaking style by extracting a time sequence of local style tokens (LST) from the reference speech. The existing content encoder in TransformerTTS is then replaced by our designed cross-attention blocks for fusion and alignment between content and style. As the fusion is performed along with the skip connection, our cross-attention block provides a good inductive bias to gradually infuse the phoneme representation with a given style. Additionally, we prevent the style embedding from encoding linguistic content by randomly truncating LST during training and using wav2vec 2.0 features. Experiments show that with fine-grained style control, our system performs better in terms of naturalness, intelligibility, and style transferability. Our code and samples are publicly available.

Takashi Maekaku, Xuankai Chang, Yuya Fujita et al.

We present a system for the Zero Resource Speech Challenge 2021, which combines a Contrastive Predictive Coding (CPC) with deep cluster. In deep cluster, we first prepare pseudo-labels obtained by clustering the outputs of a CPC network with k-means. Then, we train an additional autoregressive model to classify the previously obtained pseudo-labels in a supervised manner. Phoneme discriminative representation is achieved by executing the second-round clustering with the outputs of the final layer of the autoregressive model. We show that replacing a Transformer layer with a Conformer layer leads to a further gain in a lexical metric. Experimental results show that a relative improvement of 35% in a phonetic metric, 1.5% in the lexical metric, and 2.3% in a syntactic metric are achieved compared to a baseline method of CPC-small which is trained on LibriSpeech 460h data. We achieve top results in this challenge with the syntactic metric.

Tzu-Hsiang Lin, Alexander Rudnicky, Trung Bui et al.

Natural Language Image Editing (NLIE) aims to use natural language instructions to edit images. Since novices are inexperienced with image editing techniques, their instructions are often ambiguous and contain high-level abstractions that tend to correspond to complex editing steps to accomplish. Motivated by this inexperience aspect, we aim to smooth the learning curve by teaching the novices to edit images using low-level commanding terminologies. Towards this end, we develop a task-oriented dialogue system to investigate low-level instructions for NLIE. Our system grounds language on the level of edit operations, and suggests options for a user to choose from. Though compelled to express in low-level terms, a user evaluation shows that 25% of users found our system easy-to-use, resonating with our motivation. An analysis shows that users generally adapt to utilizing the proposed low-level language interface. In this study, we identify that object segmentation as the key factor to the user satisfaction. Our work demonstrates the advantages of the low-level, direct language-action mapping approach that can be applied to other problem domains beyond image editing such as audio editing or industrial design.

Emily Dinan, Varvara Logacheva, Valentin Malykh et al.

We describe the setting and results of the ConvAI2 NeurIPS competition that aims to further the state-of-the-art in open-domain chatbots. Some key takeaways from the competition are: (i) pretrained Transformer variants are currently the best performing models on this task, (ii) but to improve performance on multi-turn conversations with humans, future systems must go beyond single word metrics like perplexity to measure the performance across sequences of utterances (conversations) -- in terms of repetition, consistency and balance of dialogue acts (e.g. how many questions asked vs. answered).

Huiting Liu, Tao Lin, Hanfei Sun et al.

RubyStar is a dialog system designed to create "human-like" conversation by combining different response generation strategies. RubyStar conducts a non-task-oriented conversation on general topics by using an ensemble of rule-based, retrieval-based and generative methods. Topic detection, engagement monitoring, and context tracking are used for managing interaction. Predictable elements of conversation, such as the bot's backstory and simple question answering are handled by separate modules. We describe a rating scheme we developed for evaluating response generation. We find that character-level RNN is an effective generation model for general responses, with proper parameter settings; however other kinds of conversation topics might benefit from using other models.

Arjun Bhardwaj, Alexander Rudnicky

In this report, we provide a comparative analysis of different techniques for user intent classification towards the task of app recommendation. We analyse the performance of different models and architectures for multi-label classification over a dataset with a relative large number of classes and only a handful examples of each class. We focus, in particular, on memory network architectures, and compare how well the different versions perform under the task constraints. Since the classifier is meant to serve as a module in a practical dialog system, it needs to be able to work with limited training data and incorporate new data on the fly. We devise a 1-shot learning task to test the models under the above constraint. We conclude that relatively simple versions of memory networks perform better than other approaches. Although, for tasks with very limited data, simple non-parametric methods perform comparably, without needing the extra training data.