arXiv subjects

Yohei Kawaguchi

Publications and source records attributed to Yohei Kawaguchi.

At least 19 recordsLinked to original sources

arXiv2026

Description and Discussion on DCASE 2026 Challenge Task 2: Noise-aware Unsupervised Anomalous Sound Detection for Machine Condition Monitoring

This paper presents an overview of DCASE 2026 Challenge Task 2, titled "Noise-aware unsupervised anomalous sound detection (UASD) for machine condition monitoring." The task aims to advance noise-robust anomalous sound detection for machine condition monitoring under the unsupervised setting, where only normal machine sounds are available for training. Reliable detection under noisy conditions is crucial for practical deployment, but previous DCASE Task 2 settings provided limited information about environmental noise, potentially limiting UASD performance in highly noisy situations. To address this limitation, DCASE 2026 allows participants to exploit two-channel audio samples simultaneously captured at locations near and far from the target machine. Since the distant microphone is expected to contain relatively stronger environmental noise and weaker direct machine sounds, it may help distinguish environmental noise components from the target machine sounds. After the challenge submission deadline, challenge results and an analysis of the submitted systems will be added.

eess.AS↗

arXiv2026

LaSTR: Language-Driven Time-Series Segment Retrieval

Effectively searching time-series data is essential for system analysis, but existing methods often require expert-designed similarity criteria or rely on global, series-level descriptions. We study language-driven segment retrieval: given a natural language query, the goal is to retrieve relevant local segments from large time-series repositories. We build large-scale segment--caption training data by applying TV2-based segmentation to LOTSA windows and generating segment descriptions with GPT-5.2, and then train a Conformer-based contrastive retriever in a shared text--time-series embedding space. On a held-out test split, we evaluate single-positive retrieval together with caption-side consistency (SBERT and VLM-as-a-judge) under multiple candidate pool sizes. Across all settings, LaSTR outperforms random and CLIP baselines, yielding improved ranking quality and stronger semantic agreement between retrieved segments and query intent.

cs.CL↗

arXiv2025

Retaining Mixture Representations for Domain Generalized Anomalous Sound Detection

Anomalous sound detection (ASD) in the wild requires robustness to distribution shifts such as unseen low-SNR input mixtures of machine and noise types. State-of-the-art systems extract embeddings from an adapted audio encoder and detect anomalies via nearest-neighbor search, but fine tuning on noisy machine sounds often acts like a denoising objective, suppressing noise and reducing generalization under mismatched mixtures or inconsistent labeling. Training-free systems with frozen self-supervised learning (SSL) encoders avoid this issue and show strong first-shot generalization, yet their performance drops when mixture embeddings deviate from clean-source embeddings. We propose to improve SSL backbones with a retain-not-denoise strategy that better preserves information from mixed sound sources. The approach combines a multi-label audio tagging loss with a mixture alignment loss that aligns student mixture embeddings to convex teacher embeddings of clean and noise inputs. Controlled experiments on stationary, non-stationary, and mismatched noise subsets demonstrate improved robustness under distribution shifts, narrowing the gap toward oracle mixture representations.

eess.AS↗

arXiv2025

Can VLM Pseudo-Labels Train a Time-Series QA Model That Outperforms the VLM?

Time-series question answering (TSQA) tasks face significant challenges due to the lack of labeled data. Alternatively, with recent advancements in large-scale models, vision-language models (VLMs) have demonstrated the potential to analyze time-series signals in a zero-shot manner. In this paper, we propose a training approach that uses pseudo labels generated by a VLM. Although VLMs can produce incorrect labels, TSQA models can still be effectively trained based on the property that deep neural networks are inherently robust to such noisy labels. Our experimental results demonstrate that TSQA models are not only successfully trained with pseudo labels, but also surpass the performance of the VLM itself by leveraging a large amount of unlabeled data.

cs.LG↗

arXiv2025

Description and Discussion on DCASE 2025 Challenge Task 2: First-shot Unsupervised Anomalous Sound Detection for Machine Condition Monitoring

This paper introduces the task description for the Detection and Classification of Acoustic Scenes and Events (DCASE) 2025 Challenge Task 2, titled "First-shot unsupervised anomalous sound detection (ASD) for machine condition monitoring". Building on the DCASE 2024 Challenge Task 2, this task is structured as a first-shot problem within a domain generalization framework. The primary objective of the first-shot approach is to facilitate the rapid deployment of ASD systems for new machine types without requiring machine-specific hyperparameter tunings. For DCASE 2025 Challenge Task 2, sounds from previously unseen machine types have been collected and provided as the evaluation dataset. We received 119 submissions from 35 teams, and an analysis of these submissions has been made in this paper. Analysis showed that various approaches can all be competitive, such as fine-tuning pre-trained models, using frozen pre-trained models, and training small models from scratch, when combined with appropriate cost functions, anomaly score normalization, and use of clean machine and noise sounds.

cs.SD↗

arXiv2025

DiffNator: Generating Structured Explanations of Time-Series Differences

In many IoT applications, the central interest lies not in individual sensor signals but in their differences, yet interpreting such differences requires expert knowledge. We propose DiffNator, a framework for structured explanations of differences between two time series. We first design a JSON schema that captures the essential properties of such differences. Using the Time-series Observations of Real-world IoT (TORI) dataset, we generate paired sequences and train a model that combine a time-series encoder with a frozen LLM to output JSON-formatted explanations. Experimental results show that DiffNator generates accurate difference explanations and substantially outperforms both a visual question answering (VQA) baseline and a retrieval method using a pre-trained time-series encoder.

cs.CL↗

arXiv2025

CLaSP: Learning Concepts for Time-Series Signals from Natural Language Supervision

This paper presents CLaSP, a novel model for retrieving time-series signals using natural language queries that describe signal characteristics. The ability to search time-series signals based on descriptive queries is essential in domains such as industrial diagnostics, where data scientists often need to find signals with specific characteristics. However, existing methods rely on sketch-based inputs, predefined synonym dictionaries, or domain-specific manual designs, limiting their scalability and adaptability. CLaSP addresses these challenges by employing contrastive learning to map time-series signals to natural language descriptions. Unlike prior approaches, it eliminates the need for predefined synonym dictionaries and leverages the rich contextual knowledge of large language models (LLMs). Using the TRUCE and SUSHI datasets, which pair time-series signals with natural language descriptions, we demonstrate that CLaSP achieves high accuracy in retrieving a variety of time series patterns based on natural language queries.

cs.CL↗

arXiv2025

Domain-Independent Automatic Generation of Descriptive Texts for Time-Series Data

Due to scarcity of time-series data annotated with descriptive texts, training a model to generate descriptive texts for time-series data is challenging. In this study, we propose a method to systematically generate domain-independent descriptive texts from time-series data. We identify two distinct approaches for creating pairs of time-series data and descriptive texts: the forward approach and the backward approach. By implementing the novel backward approach, we create the Temporal Automated Captions for Observations (TACO) dataset. Experimental results demonstrate that a contrastive learning based model trained using the TACO dataset is capable of generating descriptive texts for time-series data in novel domains.

cs.CL↗

arXiv2025

Model-free Speculative Decoding for Transformer-based ASR with Token Map Drafting

End-to-end automatic speech recognition (ASR) systems based on transformer architectures, such as Whisper, offer high transcription accuracy and robustness. However, their autoregressive decoding is computationally expensive, hence limiting deployment on CPU-based and resource-constrained devices. Speculative decoding (SD) mitigates this issue by using a smaller draft model to propose candidate tokens, which are then verified by the main model. However, this approach is impractical for devices lacking hardware accelerators like GPUs. To address this, we propose \emph{Token Map Drafting}, a model-free SD technique that eliminates the need for a separate draft model. Instead, we leverage a precomputed n-gram token map derived from domain-specific training data, enabling efficient speculative decoding with minimal overhead. Our method significantly accelerates ASR inference in structured, low-perplexity domains without sacrificing transcription accuracy. Experimental results demonstrate decoding speed-ups of $1.27\times$ on the CI-AVSR dataset and $1.37\times$ on our internal dataset without degrading recognition accuracy. Additionally, our approach achieves a $10\%$ absolute improvement in decoding speed over the Distill-spec baseline running on CPU, highlighting its effectiveness for on-device ASR applications.

cs.CL↗

arXiv2025

MIMII-Agent: Leveraging LLMs with Function Calling for Relative Evaluation of Anomalous Sound Detection

This paper proposes a method for generating machine-type-specific anomalies to evaluate the relative performance of unsupervised anomalous sound detection (UASD) systems across different machine types, even in the absence of real anomaly sound data. Conventional keyword-based data augmentation methods often produce unrealistic sounds due to their reliance on manually defined labels, limiting scalability as machine types and anomaly patterns diversify. Advanced audio generative models, such as MIMII-Gen, show promise but typically depend on anomalous training data, making them less effective when diverse anomalous examples are unavailable. To address these limitations, we propose a novel synthesis approach leveraging large language models (LLMs) to interpret textual descriptions of faults and automatically select audio transformation functions, converting normal machine sounds into diverse and plausible anomalous sounds. We validate this approach by evaluating a UASD system trained only on normal sounds from five machine types, using both real and synthetic anomaly data. Experimental results reveal consistent trends in relative detection difficulty across machine types between synthetic and real anomalies. This finding supports our hypothesis and highlights the effectiveness of the proposed LLM-based synthesis approach for relative evaluation of UASD systems.

eess.AS↗

arXiv2025

LLM-based Generative Error Correction for Rare Words with Synthetic Data and Phonetic Context

Generative error correction (GER) with large language models (LLMs) has emerged as an effective post-processing approach to improve automatic speech recognition (ASR) performance. However, it often struggles with rare or domain-specific words due to limited training data. Furthermore, existing LLM-based GER approaches primarily rely on textual information, neglecting phonetic cues, which leads to over-correction. To address these issues, we propose a novel LLM-based GER approach that targets rare words and incorporates phonetic information. First, we generate synthetic data to contain rare words for fine-tuning the GER model. Second, we integrate ASR's N-best hypotheses along with phonetic context to mitigate over-correction. Experimental results show that our method not only improves the correction of rare words but also reduces the WER and CER across both English and Japanese datasets.

cs.SD↗

arXiv2025

Retrieving Time-Series Differences Using Natural Language Queries

Effectively searching time-series data is essential for system analysis; however, traditional methods often require domain expertise to define search criteria. Recent advancements have enabled natural language-based search, but these methods struggle to handle differences between time-series data. To address this limitation, we propose a natural language query-based approach for retrieving pairs of time-series data based on differences specified in the query. Specifically, we define six key characteristics of differences, construct a corresponding dataset, and develop a contrastive learning-based model to align differences between time-series data with query texts. Experimental results demonstrate that our model achieves an overall mAP score of 0.994 in retrieving time-series pairs.

cs.CL↗

arXiv2024

Timbre Difference Capturing in Anomalous Sound Detection

This paper proposes a framework of explaining anomalous machine sounds in the context of anomalous sound detection~(ASD). While ASD has been extensively explored, identifying how anomalous sounds differ from normal sounds is also beneficial for machine condition monitoring. However, existing sound difference captioning methods require anomalous sounds for training, which is impractical in typical machine condition monitoring settings where such sounds are unavailable. To solve this issue, we propose a new strategy for explaining anomalous differences that does not require anomalous sounds for training. Specifically, we introduce a framework that explains differences in predefined timbre attributes instead of using free-form text captions. Objective metrics of timbre attributes can be computed using timbral models developed through psycho-acoustical research, enabling the estimation of how and what timbre attributes have changed from normal sounds without training machine learning models. Additionally, to accurately determine timbre differences regardless of variations in normal training data, we developed a method that jointly conducts anomalous sound detection and timbre difference estimation based on a k-nearest neighbors method in an audio embedding space. Evaluation using the MIMII DG dataset demonstrated the effectiveness of the proposed method.

eess.AS↗

arXiv2024

Retrieval-Augmented Approach for Unsupervised Anomalous Sound Detection and Captioning without Model Training

This paper proposes a method for unsupervised anomalous sound detection (UASD) and captioning the reason for detection. While there is a method that captions the difference between given normal and anomalous sound pairs, it is assumed to be trained and used separately from the UASD model. Therefore, the obtained caption can be irrelevant to the differences that the UASD model captured. In addition, it requires many caption labels representing differences between anomalous and normal sounds for model training. The proposed method employs a retrieval-augmented approach for captioning of anomalous sounds. Difference captioning in the embedding space output by the pre-trained CLAP (contrastive language-audio pre-training) model makes the anomalous sound detection results consistent with the captions and does not require training. Experiments based on subjective evaluation and a sample-wise analysis of the output captions demonstrate the effectiveness of the proposed method.

eess.AS↗

arXiv2024

End-to-End Integration of Speech Emotion Recognition with Voice Activity Detection using Self-Supervised Learning Features

Speech Emotion Recognition (SER) often operates on speech segments detected by a Voice Activity Detection (VAD) model. However, VAD models may output flawed speech segments, especially in noisy environments, resulting in degraded performance of subsequent SER models. To address this issue, we propose an end-to-end (E2E) method that integrates VAD and SER using Self-Supervised Learning (SSL) features. The VAD module first receives the SSL features as input, and the segmented SSL features are then fed into the SER module. Both the VAD and SER modules are jointly trained to optimize SER performance. Experimental results on the IEMOCAP dataset demonstrate that our proposed method improves SER performance. Furthermore, to investigate the effect of our proposed method on the VAD and SSL modules, we present an analysis of the VAD outputs and the weights of each layer of the SSL encoder.

cs.SD↗

arXiv2024

Can We Estimate Purchase Intention Based on Zero-shot Speech Emotion Recognition?

This paper proposes a zero-shot speech emotion recognition (SER) method that estimates emotions not previously defined in the SER model training. Conventional methods are limited to recognizing emotions defined by a single word. Moreover, we have the motivation to recognize unknown bipolar emotions such as ``I want to buy - I do not want to buy.'' In order to allow the model to define classes using sentences freely and to estimate unknown bipolar emotions, our proposed method expands upon the contrastive language-audio pre-training (CLAP) framework by introducing multi-class and multi-task settings. We also focus on purchase intention as a bipolar emotion and investigate the model's performance to zero-shot estimate it. This study is the first attempt to estimate purchase intention from speech directly. Experiments confirm that the results of zero-shot estimation by the proposed method are at the same level as those of the model trained by supervised learning.

eess.AS↗

arXiv2024

MIMII-Gen: Generative Modeling Approach for Simulated Evaluation of Anomalous Sound Detection System

Insufficient recordings and the scarcity of anomalies present significant challenges in developing and validating robust anomaly detection systems for machine sounds. To address these limitations, we propose a novel approach for generating diverse anomalies in machine sound using a latent diffusion-based model that integrates an encoder-decoder framework. Our method utilizes the Flan-T5 model to encode captions derived from audio file metadata, enabling conditional generation through a carefully designed U-Net architecture. This approach aids our model in generating audio signals within the EnCodec latent space, ensuring high contextual relevance and quality. We objectively evaluated the quality of our generated sounds using the Fréchet Audio Distance (FAD) score and other metrics, demonstrating that our approach surpasses existing models in generating reliable machine audio that closely resembles actual abnormal conditions. The evaluation of the anomaly detection system using our generated data revealed a strong correlation, with the area under the curve (AUC) score differing by 4.8\% from the original, validating the effectiveness of our generated data. These results demonstrate the potential of our approach to enhance the evaluation and robustness of anomaly detection systems across varied and previously unseen conditions. Audio samples can be found at \url{https://hpworkhub.github.io/MIMII-Gen.github.io/}.

eess.AS↗

arXiv2024

Stream-based Active Learning for Anomalous Sound Detection in Machine Condition Monitoring

This paper introduces an active learning (AL) framework for anomalous sound detection (ASD) in machine condition monitoring system. Typically, ASD models are trained solely on normal samples due to the scarcity of anomalous data, leading to decreased accuracy for unseen samples during inference. AL is a promising solution to solve this problem by enabling the model to learn new concepts more effectively with fewer labeled examples, thus reducing manual annotation efforts. However, its effectiveness in ASD remains unexplored. To minimize update costs and time, our proposed method focuses on updating the scoring backend of ASD system without retraining the neural network model. Experimental results on the DCASE 2023 Challenge Task 2 dataset confirm that our AL framework significantly improves ASD performance even with low labeling budgets. Moreover, our proposed sampling strategy outperforms other baselines in terms of the partial area under the receiver operating characteristic score.

cs.SD↗