Transformer-based models are widely used in natural language understanding (NLU) tasks, and multimodal transformers have been effective in visual-language tasks. This study explores distilling visual information from pretrained multimodal transformers to pretrained language encoders. Our framework is inspired by cross-modal encoders' success in visual-language tasks while we alter the learning objective to cater to the language-heavy characteristics of NLU. After training with a small number of extra adapting steps and finetuned, the proposed XDBERT (cross-modal distilled BERT) outperforms pretrained-BERT in general language understanding evaluation (GLUE), situations with adversarial generations (SWAG) benchmarks, and readability benchmarks. We analyze the performance of XDBERT on GLUE to show that the improvement is likely visually grounded.
Autoregressive models have achieved outstanding performance in neural speech synthesis tasks. Though they can generate highly natural human speech, the iterative generation inevitably makes the synthesis time proportional to the utterance's length, leading to low efficiency. Many works were dedicated to generating the whole speech time sequence in parallel and then proposed GAN-based, flow-based, and score-based models. This paper proposed a new thought for the autoregressive generation. Instead of iteratively predicting samples in a time sequence, the proposed model performs frequency-wise autoregressive generation (FAR) and bit-wise autoregressive generation (BAR) to synthesize speech. In FAR, a speech utterance is first split into different frequency subbands. The proposed model generates a subband conditioned on the previously generated one. A full band speech can then be reconstructed by using these generated subbands and a synthesis filter bank. Similarly, in BAR, an 8-bit quantized signal is generated iteratively from the first bit. By redesigning the autoregressive method to compute in domains other than the time domain, the number of iterations in the proposed model is no longer proportional to the utterance's length but the number of subbands/bits. The inference efficiency is hence significantly increased. Besides, a post-filter is employed to sample audio signals from output posteriors, and its training objective is designed based on the characteristics of the proposed autoregressive methods. The experimental results show that the proposed model is able to synthesize speech faster than real-time without GPU acceleration. Compared with the baseline autoregressive and non-autoregressive models, the proposed model achieves better MOS and shows its good generalization ability while synthesizing 44 kHz speech or utterances from unseen speakers.
Explanation methods in Interpretable NLP often explain the model's decision by extracting evidence (rationale) from the input texts supporting the decision. Benchmark datasets for rationales have been released to evaluate how good the rationale is. The ground truth rationales in these datasets are often human annotations obtained via crowd-sourced websites. Valuable as these datasets are, the details on how those human annotations are obtained are often not clearly specified. We conduct comprehensive controlled experiments using crowd-sourced websites on two widely used datasets in Interpretable NLP to understand how those unsaid details can affect the annotation results. Specifically, we compare the annotation results obtained from recruiting workers satisfying different levels of qualification. We also provide high-quality workers with different instructions for completing the same underlying tasks. Our results reveal that the annotation quality is highly subject to the workers' qualification, and workers can be guided to provide certain annotations by the instructions. We further show that specific explanation methods perform better when evaluated using the ground truth rationales obtained by particular instructions. Based on these observations, we highlight the importance of providing complete details of the annotation process and call for careful interpretation of any experiment results obtained using those annotations.
In this paper, we study the differences and commonalities between statistically out-of-distribution (OOD) samples and adversarial (Adv) samples, both of which hurting a text classification model's performance. We conduct analyses to compare the two types of anomalies (OOD and Adv samples) with the in-distribution (ID) ones from three aspects: the input features, the hidden representations in each layer of the model, and the output probability distributions of the classifier. We find that OOD samples expose their aberration starting from the first layer, while the abnormalities of Adv samples do not emerge until the deeper layers of the model. We also illustrate that the models' output probabilities for Adv samples tend to be more unconfident. Based on our observations, we propose a simple method to separate ID, OOD, and Adv samples using the hidden representations and output probabilities of the model. On multiple combinations of ID, OOD datasets, and Adv attacks, our proposed method shows exceptional results on distinguishing ID, OOD, and Adv samples.
Mean opinion score (MOS) is a typical subjective evaluation metric for speech synthesis systems. Since collecting MOS is time-consuming, it would be desirable if there are accurate MOS prediction models for automatic evaluation. In this work, we propose DDOS, a novel MOS prediction model. DDOS utilizes domain adaptive pre-training to further pre-train self-supervised learning models on synthetic speech. And a proposed module is added to model the opinion score distribution of each utterance. With the proposed components, DDOS outperforms previous works on BVCC dataset. And the zero shot transfer result on BC2019 dataset is significantly improved. DDOS also wins second place in Interspeech 2022 VoiceMOS challenge in terms of system-level score.
Most recent TTS systems are composed of a synthesizer and a vocoder. However, the existing synthesizers and vocoders can only be matched to acoustic features extracted with a specific configuration. Hence, we can't combine arbitrary synthesizers and vocoders together to form a complete TTS system, not to mention applying to a newly developed model. In this paper, we proposed a universal adaptor, which takes a Mel-spectogram parametrized by the source configuration and converts it into a Mel-spectrogram parametrized by the target configuration, as long as we feed in the source and the target configurations. Experiments show that the quality of speeches synthesized from our output of the universal adaptor is comparable to those synthesized from ground truth Mel-spectrogram. Moreover, our universal adaptor can be applied in the recent TTS systems and in multi-speaker speech synthesis without dropping quality.
Speech representations learned from Self-supervised learning (SSL) models have been found beneficial for various speech processing tasks. However, utilizing SSL representations usually requires fine-tuning the pre-trained models or designing task-specific downstream models and loss functions, causing much memory usage and human labor. On the other hand, prompting in Natural Language Processing (NLP) is an efficient and widely used technique to leverage pre-trained language models (LMs). Nevertheless, such a paradigm is little studied in the speech community. We report in this paper the first exploration of the prompt tuning paradigm for speech processing tasks based on Generative Spoken Language Model (GSLM). Experiment results show that the prompt tuning technique achieves competitive performance in speech classification tasks with fewer trainable parameters than fine-tuning specialized downstream models. We further study the technique in challenging sequence generation tasks. Prompt tuning also demonstrates its potential, while the limitation and possible research directions are discussed in this paper.
Speech distortions are a long-standing problem that degrades the performance of supervisely trained speech processing models. It is high time that we enhance the robustness of speech processing models to obtain good performance when encountering speech distortions while not hurting the original performance on clean speech. In this work, we propose to improve the robustness of speech processing models by domain adversarial training (DAT). We conducted experiments based on the SUPERB framework on five different speech processing tasks. In case we do not always have knowledge of the distortion types for speech data, we analyzed the binary-domain and multi-domain settings, where the former treats all distorted speech as one domain, and the latter views different distortions as different domains. In contrast to supervised training methods, we obtained promising results in target domains where speech data is distorted with different distortions including new unseen distortions introduced during testing.
Recently, many novel techniques have been introduced to deal with spoofing attacks, and achieve promising countermeasure (CM) performances. However, these works only take the stand-alone CM models into account. Nowadays, a spoofing aware speaker verification (SASV) challenge which aims to facilitate the research of integrated CM and ASV models, arguing that jointly optimizing CM and ASV models will lead to better performance, is taking place. In this paper, we propose a novel multi-model and multi-level fusion strategy to tackle the SASV task. Compared with purely scoring fusion and embedding fusion methods, this framework first utilizes embeddings from CM models, propagating CM embeddings into a CM block to obtain a CM score. In the second-level fusion, the CM score and ASV scores directly from ASV systems will be concatenated into a prediction block for the final decision. As a result, the best single fusion system has achieved the SASV-EER of 0.97% on the evaluation set. Then by ensembling the top-5 fusion systems, the final SASV-EER reached 0.89%.