Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2312.04846

Sound Source Localization for a Source inside a Structure using Ac-CycleGAN

Shunsuke Kita·Choong Sik Park·Yoshinobu Kajikawa
View original source ↗Read source document ↗Export metadata ↓

Abstract

We propose a method for sound source localization (SSL) for a source inside a structure using Ac-CycleGAN under unpaired data conditions. The proposed method utilizes a large amount of simulated data and a small amount of actual experimental data to locate a sound source inside a structure in a real environment. An Ac-CycleGAN generator contributes to the transformation of simulated data into real data, or vice versa, using unpaired data from both domains. The discriminator of an Ac-CycleGAN model is designed to differentiate between the transformed data generated by the generator and real data, while also predicting the location of the sound source. Vectors representing the frequency spectrum of the accelerometers (FSAs) measured at three points outside the structure are used as input data and the source areas inside the structure are used as labels. The input data vectors are concatenated vertically to form an image. Labels are defined by dividing the interior of the structure into eight areas with one-hot encoding for each area. Thus, the SSL problem is redefined as an image-classification problem to stochastically estimate the location of the sound source. We show that it is possible to estimate the sound source location using the Ac-CycleGAN discriminator for unpaired data across domains. Furthermore, we analyze the discriminative factors for distinguishing the data. The proposed model exhibited an accuracy exceeding 90\% when trained on 80\% of actual data (12.5\% of simulated data). Despite potential imperfections in the domain transformation process carried out by the Ac-CycleGAN generator, the discriminator can effectively distinguish between transferred and real data by selectively utilizing only those features that generate a relatively small transformation error.

Explore related subjects

cs.SDeess.AS

Keep this discovery

BibTeXRIS

Shunsuke Kita, Choong Sik Park, Yoshinobu Kajikawa. 2023-12-08. Sound Source Localization for a Source inside a Structure using Ac-CycleGAN. https://arxiv.org/abs/2312.04846

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Xiaomi-CocktailASR-1 Technical Report

Recently, large language model (LLM) based ASR models have achieved significant progress, yet they generally lack support for multi-speaker scenarios, where the cocktail party problem remains a critical bottleneck for further advancing ASR. Existing TS-ASR methods, including end-to-end architectures with speaker embeddings and latest LLM-based explorations suffer from degraded single-speaker performance and the inability to reject when the target speaker is absent. In this paper, we propose Xiaomi-CocktailASR-1, an LLM-based end-to-end TS-ASR architecture. By utilizing reference speech as voiceprint prompts, it directly transcribes the target speaker's speech without requiring speech separation. Xiaomi-CocktailASR-1 maintains competitive performance in single-speaker scenarios, comparable to mainstream ASR models. It also features a negative sample rejection capability, outputting empty text when the target speaker is absent from the mixed speech. Additionally, Xiaomi-CocktailASR-1 supports a Chain-of-Thought (CoT) reasoning mode to provide explicit reasoning steps. Extensive experiments on various synthetic and real-world multispeaker benchmarks demonstrate that Xiaomi-CocktailASR-1 achieves state-of-the-art performance, effectively addressing the cocktail party problem through a unified architecture that balances multispeaker and single-speaker recognition accuracy, along with rejection capability.

cs.SD↗
arXiv2026

EConv-TasNet: Efficient Conv-TasNet for Effective Speech Separation

Conv-TasNet has served as a strong baseline for time-domain speech separation, and many studies have extended it with advanced architectures such as dual-path networks, U-Nets, and attention mechanisms. However, these methods often introduce high computational cost and complexity, limiting their deployment in resource-constrained scenarios. To address this issue, we propose eConv-TasNet, an efficient variant of Conv-TasNet that improves both effectiveness and efficiency without relying on resource-intensive modules. The proposed model consists of a group-wise early-splitting (GES) module and a multi-group feature aggregation (MGFA) module. GES generates discriminative speaker embeddings at intermediate stages, while MGFA progressively aggregates these group-level representations for refined mask estimation. Experimental results show that eConv-TasNet reduces model size by 22.4%, accelerates inference by 18.9%, and improves SI-SNRi by 14.0%-28.0% across three public benchmarks. Moreover, it achieves competitive performance compared with state-of-the-art methods while requiring significantly fewer parameters and lower inference cost. These results demonstrate a favorable efficiency-effectiveness trade-off for edge deployment.

cs.SD↗