Sorry, you need to enable JavaScript to visit this website.

This work proposes a frame-wise online/streaming end-to-end neural diarization (FS-EEND) method in a frame-in-frame-out fashion. To frame-wisely detect a flexible number of speakers and extract/update their corresponding attractors, we propose to leverage a causal speaker embedding encoder and an online non-autoregressive self-attention-based attractor decoder. A look-ahead mechanism is adopted to allow leveraging some future frames for effectively detecting new speakers in real time and adaptively updating speaker attractors.

In this paper, we explain the model that was developed by the NLP\_POSTECH team for the LIMMITS 2024 Grand Challenge. Among the three tracks, we focus on Track 1, which necessitates the creation of a few-shot text-to-speech (TTS) system that generates natural speech across diverse languages. Towards this end, to realize multi-lingual capability, we incorporate a learnable language embedding. In addition, for precise imitation of target speaker voices, we leverage an inductive speaker bias conditioning methodology.

This paper studies the external division operator, an external division (an affine combination with positive and negative weights) of two proximity operators. We show that the external division operator is cocoercive under some condition, and it can be expressed as the proximity operator of a certain weakly convex function. We then consider using the external division operator as an alternative to the proximity operator in the proximal gradient algorithm, which we show converges to a minimizer of the cost function penalized by the weakly convex function under some conditions.

If our noise-canceling headphones can understand our audio environments, they can then inform us of important sound events, tune equalization based on the types of content we listen to, and dynamically adjust noise cancellation parameters based on audio scenes to further reduce distraction. However, running multiple audio understanding models on headphones with a limited energy budget and on-chip memory remains a challenging task.

In this paper, we present the decomposed triplane-hash neural radiance fields (DT-NeRF), a framework that significantly improves the photorealistic rendering of talking faces and achieves state-of-the-art results on key evaluation datasets. Our architecture decomposes the facial region into two specialized triplanes: one specialized for representing the mouth, and the other for the broader facial features. We introduce audio features as residual terms and integrate them as query vectors into our model through an audio-mouthface transformer.

In this paper, we present the decomposed triplane-hash neural radiance fields (DT-NeRF), a framework that significantly improves the photorealistic rendering of talking faces and achieves state-of-the-art results on key evaluation datasets. Our architecture decomposes the facial region into two specialized triplanes: one specialized for representing the mouth, and the other for the broader facial features. We introduce audio features as residual terms and integrate them as query vectors into our model through an audio-mouthface transformer.

Previous works have shown that perturbing internal-layer features can significantly enhance the transferability of black-box attacks in classifiers. However, these methods have not achieved satisfactory performance when applied to detectors due to the inherent differences in features between detectors and classifiers. In this paper, we introduce a concise and practical untargeted adversarial attack in a label-free manner, which leverages only the feature extracted from the backbone model.

With the rapid development of mobile Internet and video applications, robust video watermarking technology has become a focal area of research for protecting and tracking intellectual property rights in digital media. An important characteristic of video is that it has both spatial and temporal properties. Previous studies in video watermarking have primarily focused on either spatial or temporal distortions and did not uniformly consider all types of video distortion, which restricts the robustness of video watermarking.

Unsupervised anomalous sound detection (ASD) aims to identify anomalous sounds by learning the features of normal operational sounds and sensing their deviations. Recent approaches have focused on the self-supervised task utilizing the classification of normal data, and advanced models have shown that securing representation space for anomalous data is important through representation learning yielding compact intra-class and well-separated intra-class distributions.

The mainstream paradigm of speech emotion recognition (SER) is identifying the single emotion label of the entire utterance. This line of works neglect the emotion dynamics at fine temporal granularity and mostly fail to leverage linguistic information of speech signal explicitly. In this paper, we propose Emotion Neural Transducer for fine-grained speech emotion recognition with automatic speech recognition (ASR) joint training. We first extend typical neural transducer with emotion joint network to construct emotion lattice for fine-grained SER.