Tetsuji Ogawa

Yosuke Higuchi, Brian Yan, Siddhant Arora et al.

This paper presents BERT-CTC, a novel formulation of end-to-end speech recognition that adapts BERT for connectionist temporal classification (CTC). Our formulation relaxes the conditional independence assumptions used in conventional CTC and incorporates linguistic knowledge through the explicit output dependency obtained by BERT contextual embedding. BERT-CTC attends to the full contexts of the input and hypothesized output sequences via the self-attention mechanism. This mechanism encourages a model to learn inner/inter-dependencies between the audio and token representations while maintaining CTC's training efficiency. During inference, BERT-CTC combines a mask-predict algorithm with CTC decoding, which iteratively refines an output sequence. The experimental results reveal that BERT-CTC improves over conventional approaches across variations in speaking styles and languages. Finally, we show that the semantic representations in BERT-CTC are beneficial towards downstream spoken language understanding tasks.

Yosuke Higuchi, Tetsuji Ogawa, Tetsunori Kobayashi et al.

We present BERT-CTC-Transducer (BECTRA), a novel end-to-end automatic speech recognition (E2E-ASR) model formulated by the transducer with a BERT-enhanced encoder. Integrating a large-scale pre-trained language model (LM) into E2E-ASR has been actively studied, aiming to utilize versatile linguistic knowledge for generating accurate text. One crucial factor that makes this integration challenging lies in the vocabulary mismatch; the vocabulary constructed for a pre-trained LM is generally too large for E2E-ASR training and is likely to have a mismatch against a target ASR domain. To overcome such an issue, we propose BECTRA, an extended version of our previous BERT-CTC, that realizes BERT-based E2E-ASR using a vocabulary of interest. BECTRA is a transducer-based model, which adopts BERT-CTC for its encoder and trains an ASR-specific decoder using a vocabulary suitable for a target task. With the combination of the transducer and BERT-CTC, we also propose a novel inference algorithm for taking advantage of both autoregressive and non-autoregressive decoding. Experimental results on several ASR tasks, varying in amounts of data, speaking styles, and languages, demonstrate that BECTRA outperforms BERT-CTC by effectively dealing with the vocabulary mismatch while exploiting BERT knowledge.

Yosuke Higuchi, Tetsuji Ogawa, Tetsunori Kobayashi et al.

This paper presents InterMPL, a semi-supervised learning method of end-to-end automatic speech recognition (ASR) that performs pseudo-labeling (PL) with intermediate supervision. Momentum PL (MPL) trains a connectionist temporal classification (CTC)-based model on unlabeled data by continuously generating pseudo-labels on the fly and improving their quality. In contrast to autoregressive formulations, such as the attention-based encoder-decoder and transducer, CTC is well suited for MPL, or PL-based semi-supervised ASR in general, owing to its simple/fast inference algorithm and robustness against generating collapsed labels. However, CTC generally yields inferior performance than the autoregressive models due to the conditional independence assumption, thereby limiting the performance of MPL. We propose to enhance MPL by introducing intermediate loss, inspired by the recent advances in CTC-based modeling. Specifically, we focus on self-conditional and hierarchical conditional CTC, that apply auxiliary CTC losses to intermediate layers such that the conditional independence assumption is explicitly relaxed. We also explore how pseudo-labels should be generated and used as supervision for intermediate losses. Experimental results in different semi-supervised settings demonstrate that the proposed approach outperforms MPL and improves an ASR model by up to a 12.1% absolute performance gain. In addition, our detailed analysis validates the importance of the intermediate loss.

Yusuke Fujita, Tatsuya Komatsu, Robin Scheibler et al.

End-to-end neural diarization (EEND) with encoder-decoder-based attractors (EDA) is a promising method to handle the whole speaker diarization problem simultaneously with a single neural network. While the EEND model can produce all frame-level speaker labels simultaneously, it disregards output label dependency. In this work, we propose a novel EEND model that introduces the label dependency between frames. The proposed method generates non-autoregressive intermediate attractors to produce speaker labels at the lower layers and conditions the subsequent layers with these labels. While the proposed model works in a non-autoregressive manner, the speaker labels are refined by referring to the whole sequence of intermediate labels. The experiments with the two-speaker CALLHOME dataset show that the intermediate labels with the proposed non-autoregressive intermediate attractors boost the diarization performance. The proposed method with the deeper network benefits more from the intermediate labels, resulting in better performance and training throughput than EEND-EDA.

Yosuke Higuchi, Tetsuji Ogawa, Tetsunori Kobayashi

We propose to utilize an instruction-tuned large language model (LLM) for guiding the text generation process in automatic speech recognition (ASR). Modern large language models (LLMs) are adept at performing various text generation tasks through zero-shot learning, prompted with instructions designed for specific objectives. This paper explores the potential of LLMs to derive linguistic information that can facilitate text generation in end-to-end ASR models. Specifically, we instruct an LLM to correct grammatical errors in an ASR hypothesis and use the LLM-derived representations to refine the output further. The proposed model is built on the joint CTC and attention architecture, with the LLM serving as a front-end feature extractor for the decoder. The ASR hypothesis, subject to correction, is obtained from the encoder via CTC decoding and fed into the LLM along with a specific instruction. The decoder subsequently takes as input the LLM output to perform token predictions, combining acoustic information from the encoder and the powerful linguistic information provided by the LLM. Experimental results show that the proposed LLM-guided model achieves a relative gain of approximately 13\% in word error rates across major benchmarks.

Ryosuke Hyodo, Susumu Saito, Teppei Nakano et al.

Through a user study in the field of livestock farming, we verify the effectiveness of an XAI framework for video surveillance systems. The systems can be made interpretable by incorporating experts' decision-making processes. AI systems are becoming increasingly common in real-world applications, especially in fields related to human decision-making, and its interpretability is necessary. However, there are still relatively few standard methods for assessing and addressing the interpretability of machine learning-based systems in real-world applications. In this study, we examine the framework of a video surveillance AI system that presents the reasoning behind predictions by incorporating experts' decision-making processes with rich domain knowledge of the notification target. While general black-box AI systems can only present final probability values, the proposed framework can present information relevant to experts' decisions, which is expected to be more helpful for their decision-making. In our case study, we designed a system for detecting signs of calving in cattle based on the proposed framework and evaluated the system through a user study (N=6) with people involved in livestock farming. A comparison with the black-box AI system revealed that many participants referred to the presented reasons for the prediction results, and five out of six participants selected the proposed system as the system they would like to use in the future. It became clear that we need to design a user interface that considers the reasons for the prediction results.

Ryosuke Hyodo, Teppei Nakano, Tetsuji Ogawa

We have designed a deep multi-stream network for automatically detecting calving signs from video. Calving sign detection from a camera, which is a non-contact sensor, is expected to enable more efficient livestock management. As large-scale, well-developed data cannot generally be assumed when establishing calving detection systems, the basis for making the prediction needs to be presented to farmers during operation, so black-box modeling (also known as end-to-end modeling) is not appropriate. For practical operation of calving detection systems, the present study aims to incorporate expert knowledge into a deep neural network. To this end, we propose a multi-stream calving sign detection network in which multiple calving-related features are extracted from the corresponding feature extraction networks designed for each attribute with different characteristics, such as a cow's posture, rotation, and movement, known as calving signs, and are then integrated appropriately depending on the cow's situation. Experimental comparisons conducted using videos of 15 cows demonstrated that our multi-stream system yielded a significant improvement over the end-to-end system, and the multi-stream architecture significantly contributed to a reduction in detection errors. In addition, the distinctive mixture weights we observed helped provide interpretability of the system's behavior.

Asahi Sakuma, Hiroaki Sato, Ryuga Sugano et al.

This paper presents a novel framework for multi-talker automatic speech recognition without the need for auxiliary information. Serialized Output Training (SOT), a widely used approach, suffers from recognition errors due to speaker assignment failures. Although incorporating auxiliary information, such as token-level timestamps, can improve recognition accuracy, extracting such information from natural conversational speech remains challenging. To address this limitation, we propose Speaker-Distinguishable CTC (SD-CTC), an extension of CTC that jointly assigns a token and its corresponding speaker label to each frame. We further integrate SD-CTC into the SOT framework, enabling the SOT model to learn speaker distinction using only overlapping speech and transcriptions. Experimental comparisons show that multi-task learning with SD-CTC and SOT reduces the error rate of the SOT model by 26% and achieves performance comparable to state-of-the-art methods relying on auxiliary information.

Huaibo Zhao, Yosuke Higuchi, Tetsuji Ogawa et al.

In the present paper, an attempt is made to combine Mask-CTC and the triggered attention mechanism to construct a streaming end-to-end automatic speech recognition (ASR) system that provides high performance with low latency. The triggered attention mechanism, which performs autoregressive decoding triggered by the CTC spike, has shown to be effective in streaming ASR. However, in order to maintain high accuracy of alignment estimation based on CTC outputs, which is the key to its performance, it is inevitable that decoding should be performed with some future information input (i.e., with higher latency). It should be noted that in streaming ASR, it is desirable to be able to achieve high recognition accuracy while keeping the latency low. Therefore, the present study aims to achieve highly accurate streaming ASR with low latency by introducing Mask-CTC, which is capable of learning feature representations that anticipate future information (i.e., that can consider long-term contexts), to the encoder pre-training. Experimental comparisons conducted using WSJ data demonstrate that the proposed method achieves higher accuracy with lower latency than the conventional triggered attention-based streaming ASR system.

Yosuke Higuchi, Keita Karube, Tetsuji Ogawa et al.

In end-to-end automatic speech recognition (ASR), a model is expected to implicitly learn representations suitable for recognizing a word-level sequence. However, the huge abstraction gap between input acoustic signals and output linguistic tokens makes it challenging for a model to learn the representations. In this work, to promote the word-level representation learning in end-to-end ASR, we propose a hierarchical conditional model that is based on connectionist temporal classification (CTC). Our model is trained by auxiliary CTC losses applied to intermediate layers, where the vocabulary size of each target subword sequence is gradually increased as the layer becomes close to the word-level output. Here, we make each level of sequence prediction explicitly conditioned on the previous sequences predicted at lower levels. With the proposed approach, we expect the proposed model to learn the word-level representations effectively by exploiting a hierarchy of linguistic structures. Experimental results on LibriSpeech-{100h, 960h} and TEDLIUM2 demonstrate that the proposed model improves over a standard CTC-based model and other competitive models from prior work. We further analyze the results to confirm the effectiveness of the intended representation learning with our model.

Masaya Morinaga, Susumu Saito, Teppei Nakano et al.

Qualification tests in crowdsourcing are often used to pre-filter workers by measuring their ability in executing microtasks.While creating qualification tests for each task type is considered as a common and reasonable way, this study investigates into its worker-filtering performance when the same qualification test is used across multiple types of tasks.On Amazon Mechanical Turk, we tested the annotation accuracy in six different cases where tasks consisted of two different difficulty levels, arising from the identical real-world domain: four combinatory cases in which the qualification test and the actual task were the same or different from each other, as well as two other cases where workers with Masters Qualification were asked to perform the actual task only.The experimental results demonstrated the two following findings: i) Workers that were assigned to a difficult qualification test scored better annotation accuracy regardless of the difficulty of the actual task; ii) Workers with Masters Qualification scored better annotation accuracy on the low-difficulty task, but were not as accurate as those who passed a qualification test on the high-difficulty task.

Yosuke Higuchi, Hirofumi Inaguma, Shinji Watanabe et al.

For real-world deployment of automatic speech recognition (ASR), the system is desired to be capable of fast inference while relieving the requirement of computational resources. The recently proposed end-to-end ASR system based on mask-predict with connectionist temporal classification (CTC), Mask-CTC, fulfills this demand by generating tokens in a non-autoregressive fashion. While Mask-CTC achieves remarkably fast inference speed, its recognition performance falls behind that of conventional autoregressive (AR) systems. To boost the performance of Mask-CTC, we first propose to enhance the encoder network architecture by employing a recently proposed architecture called Conformer. Next, we propose new training and decoding methods by introducing auxiliary objective to predict the length of a partial target sequence, which allows the model to delete or insert tokens during inference. Experimental results on different ASR tasks show that the proposed approaches improve Mask-CTC significantly, outperforming a standard CTC model (15.5% $\rightarrow$ 9.1% WER on WSJ). Moreover, Mask-CTC now achieves competitive results to AR models with no degradation of inference speed ($<$ 0.1 RTF using CPU). We also show a potential application of Mask-CTC to end-to-end speech translation.

Yosuke Higuchi, Shinji Watanabe, Nanxin Chen et al.

We present Mask CTC, a novel non-autoregressive end-to-end automatic speech recognition (ASR) framework, which generates a sequence by refining outputs of the connectionist temporal classification (CTC). Neural sequence-to-sequence models are usually \textit{autoregressive}: each output token is generated by conditioning on previously generated tokens, at the cost of requiring as many iterations as the output length. On the other hand, non-autoregressive models can simultaneously generate tokens within a constant number of iterations, which results in significant inference time reduction and better suits end-to-end ASR model for real-world scenarios. In this work, Mask CTC model is trained using a Transformer encoder-decoder with joint training of mask prediction and CTC. During inference, the target sequence is initialized with the greedy CTC outputs and low-confidence tokens are masked based on the CTC probabilities. Based on the conditional dependence between output tokens, these masked low-confidence tokens are then predicted conditioning on the high-confidence tokens. Experimental results on different speech recognition tasks show that Mask CTC outperforms the standard CTC model (e.g., 17.9% -> 12.1% WER on WSJ) and approaches the autoregressive model, requiring much less inference time using CPUs (0.07 RTF in Python implementation). All of our codes will be publicly available.

Koki Madono, Masayuki Tanaka, Masaki Onishi et al.

In this study, a perceptually hidden object-recognition method is investigated to generate secure images recognizable by humans but not machines. Hence, both the perceptual information hiding and the corresponding object recognition methods should be developed. Block-wise image scrambling is introduced to hide perceptual information from a third party. In addition, an adaptation network is proposed to recognize those scrambled images. Experimental comparisons conducted using CIFAR datasets demonstrated that the proposed adaptation network performed well in incorporating simple perceptual information hiding into DNN-based image classification.