This work presents NEVU, a benchmark for actor-conditioned, event-centric, and direction-aware human value recognition in news-domain texts. NEVU evaluates whether models can infer values from event-structured evidence, attribute them to the correct social actors, and determine their aligned or contradictory direction. Built from 2865 English news articles, NEVU represents news at four semantic levels, from subevents to composite events and full articles. Using a hierarchical taxonomy of 54 fine-grained and 20 coarse-grained values, the benchmark contains 46,589 semantic units, 72,905 annotated unit-actor pairs, and 168,061 directed value instances. NEVU is constructed through a staged LLM-assisted annotation and verification pipeline, with targeted human verification for unresolved cases. Candidate-level acceptance and agreement are further examined through a multi-group assessment. The experiments show that prompting-only models remain limited, whereas LoRA-tuned open-weight models substantially improve performance, with overall Micro-F1 gains of 29.77 and 8.21 percentage points over the strongest prompting-only open-weight and proprietary baselines, respectively. These gains primarily reflect learnability under the NEVU reference-label setting. NEVU provides a structured benchmark for systematic evaluation and supervised adaptation of actor-conditioned human value recognition in English news.
ASVspoof 5 is the fifth edition in a series of challenges which promote the study of speech spoofing and deepfake attacks as well as the design of detection solutions. We introduce the ASVspoof 5 database which is generated in a crowdsourced fashion from data collected in diverse acoustic conditions (cf. studio-quality data for earlier ASVspoof databases) and from similar to 2000 speakers (cf. similar to 100 earlier). The database contains attacks generated with 32 different algorithms, also crowdsourced, and optimised to varying degrees using new surrogate detection models. Among them are attacks generated with a mix of legacy and contemporary text-to-speech synthesis and voice conversion models, in addition to adversarial attacks which are incorporated for the first time. ASVspoof 5 protocols comprise seven speaker-disjoint partitions. They include two distinct partitions for the training of different sets of attack models, two more for the development and evaluation of surrogate detection models, and then three additional partitions which comprise the ASVspoof 5 training, development and evaluation sets. An auxiliary set of data collected from an additional 30k speakers can also be used to train speaker encoders for the implementation of attack algorithms. Also described herein is an experimental validation of the new ASVspoof 5 database using a set of automatic speaker verification and spoof/deepfake baseline detectors. With the exception of protocols and tools for the generation of spoofed/deepfake speech, the resources described in this paper, already used by participants of the ASVspoof 5 challenge in 2024, are now all freely available to the community.
Cryptospace steganography has attracted increasing attention as an effective approach for enhancing data security in cloud environments. This paper proposes a hybrid framework that integrates cycle-consistent generative adversarial networks (CycleGAN) with the difference expansion (DE) technique to provide both image encryption and data hiding services. In the proposed framework, an image encryption network and a data encryption network are designed to encrypt the digital image and secret data, respectively, enabling a key-free architecture. A dropout-driven strategy is further introduced to support secure and isolated access control for multiple user groups on a shared cloud platform. Experimental results show that the proposed method achieves an embedding rate above 0.47 bpp and a secret data extraction accuracy exceeding 91%, demonstrating superior performance compared with state-of-the-art methods.
We present results and analyses from the third VoicePrivacy Challenge held in 2024, which focuses on advancing voice anonymization technologies. The task was to develop a voice anonymization system for speech data that conceals a speaker's voice identity while preserving linguistic content and emotional state. We provide a systematic overview of the challenge framework, including detailed descriptions of the anonymization task and datasets used for both system development and evaluation. We outline the attack model and objective evaluation metrics for assessing privacy protection (concealing speaker voice identity) and utility (content and emotional state preservation). We describe six baseline anonymization systems and summarize the innovative approaches developed by challenge participants. Finally, we provide key insights and observations to guide the design of future VoicePrivacy challenges and identify promising directions for voice anonymization research.
Extractive QA tasks are commonly evaluated using Exact Match (EM) and F1-score, but these metrics often fail to reflect true model performance. Recent studies have proposed using large language models (LLMs) as judges (LLM-as-a-judge), yet they often lack comprehensive evaluation across datasets and overlook key factors such as sensitivity to answer types, prompt variations, and self-preference bias. In this work, we conduct a systematic study of LLM-as-a-judge across four extractive QA datasets and various prompt variations, assessing multiple LLM families in both answering and judging roles. Our results show that LLM-as-a-judge judgments correlate much more strongly with human evaluations than EM (0.22) and F1 (0.40), achieving correlations up to 0.85 with open-source models. Further analysis reveals that LLM-as-a-judge performs particularly well on number-related answers but faces challenges with more complex types, such as job titles. Contrary to findings in other NLP tasks, we observe no self-preference bias, even when the same model serves as both QA model and judge. Finally, we find that prompt phrasing has minimal impact, and zero-shot, context-free judging often yields the best evaluation performance.
