Tianrui Wang, Ziyang Ma, Yizhou Peng, Haoyu Wang, Zhikang Niu, Zikang Huang, Yihao Wu, Yi-Wen Chao, Yu Jiang, Yuheng Lu, Guanrou Yang, Xuanchen Li, Hexin Liu, Chunyu Qiang, Cheng Gong, Yifan Yang, Tianchi Liu, Junyu Wang, Nana Hou, Meng Ge, Fuming You, Yang Wei, Zhongqian Sun, Hu Haifeng, Xiaobao Wang, Eng Siong Chng, Xie Chen, Longbiao Wang, Jianwu Dang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Current large speech language models (SpeechLLMs) often exhibit limitations in empathetic reasoning, primarily due to the absence of training datasets that integrate both contextual content and paralinguistic cues. In this work, we propose two approaches to incorporate contextual paralinguistic information into model training: (1) an explicit method that provides paralinguistic metadata (e.g., emotion annotations) directly to the LLM, and (2) an implicit method that automatically generates novel training question-answer (QA) pairs using both categorical and dimensional emotion annotations alongside speech transcriptions. Our implicit method boosts performance (LLM-judged) by 38.41% on a human-annotated QA benchmark, reaching 46.02% when combined with the explicit approach, showing effectiveness in contextual paralinguistic understanding. We also validate the LLM judge by demonstrating its correlation with classification metrics, providing support for its reliability.

In speaker verification, we use computational method to verify if an utterance matches the identity of an enrolled speaker. This task is similar to the manual task of forensic voice comparison, where linguistic analysis is combined with auditory measurements to compare and evaluate voice samples. Despite much success, we have yet to develop a speaker verification system that offers explainable results comparable to those from manual forensic voice comparison. A novel approach, Explainable Phonetic Trait-Oriented (ExPO) network, is proposed in this letter to introduce the speaker'sphonetic traitwhich describes the speaker's characteristics at the phonetic level, resembling what forensic comparison does. ExPO not only generates utterance-level speaker embeddings but also allows for fine-grained analysis and visualization of phonetic traits, offering an explainable speaker verification process. Furthermore, we investigate phonetic traits from within-speaker and between-speaker variation perspectives to determine which trait is most effective for speaker verification, marking an important step towards explainable speaker verification.

Speech foundation models have significantly advanced various speech-related tasks by providing exceptional representation capabilities. However, their high-dimensional output features often create a mismatch with downstream task models, which typically require lower-dimensional inputs. A common solution is to apply a dimensionality reduction (DR) layer, but this approach increases parameter overhead, computational costs, and risks losing valuable information. To address these issues, we propose Nested Res2Net (Nes2Net), a lightweight back-end architecture designed to directly process high-dimensional features without DR layers. The nested structure enhances multi-scale feature extraction, improves feature interaction, and preserves high-dimensional information. We first validate Nes2Net on CtrSVDD, a singing voice deepfake detection dataset, and report a 22% performance improvement and an 87% back-end computational cost reduction over the state-of-the-art baseline. Additionally, extensive testing across four diverse datasets: ASVspoof 2021, ASVspoof 5, PartialSpoof, and In-the-Wild, covering fully spoofed speech, adversarial attacks, partial spoofing, and real-world scenarios, consistently highlights Nes2Net’s superior robustness and generalization capabilities. The code package and pre-trained models are available at https://github.com/Liu-Tianchi/Nes2Net.

This work details our approach to achieving a leading system with a 1.79% pooled equal error rate (EER) on the evaluation set of the Controlled Singing Voice Deepfake Detection (CtrSVDD). The rapid advancement of generative AI models presents significant challenges for detecting AI-generated deepfake singing voices, attracting increased research attention. The Singing Voice Deepfake Detection (SVDD) Challenge 2024 aims to address this complex task. In this work, we explore the ensemble methods, utilizing speech foundation models to develop robust singing voice anti-spoofing systems. We also introduce a novel Squeeze-and-Excitation Aggregation (SEA) method, which efficiently and effectively integrates representation features from the speech foundation models, surpassing the performance of our other individual systems. Evaluation results confirm the efficacy of our approach in detecting deepfake singing voices. The codes can be accessed at https://github.com/Anmol2059/SVDD2024.

The effects of language mismatch impact speech anti-spoofing systems, while investigations and quantification of these effects remain limited. Existing anti-spoofing datasets are mainly in English, and the high cost of acquiring multilingual datasets hinders training language-independent models. We initiate this work by evaluating top-performing speech anti-spoofing systems that are trained on English data but tested on other languages, observing notable performance declines. We propose an innovative approach - Accent-based data expansion via TTS (ACCENT), which introduces diverse linguistic knowledge to monolingual-trained models, improving their cross-lingual capabilities. We conduct experiments on a large-scale dataset consisting of over 3 million samples, including 1.8 million training samples and nearly 1.2 million testing samples across 12 languages. The language mismatch effects are preliminarily quantified and remarkably reduced over 15% by applying the proposed ACCENT. This easily implementable method shows promise for multilingual and low-resource language scenarios.

The residual neural networks (ResNet) demonstrate the impressive performance in automatic speaker verification (ASV). They treat the time and frequency dimensions equally, following the default stride configuration designed for image recognition, where the horizontal and vertical axes exhibit similarities. This approach ignores the fact that time and frequency are asymmetric in speech representation. We address this issue and postulateGolden-Gemini Hypothesis,which posits the prioritization of temporal resolution over frequency resolution for ASV. The hypothesis is verified by conducting a systematic study on the impact of temporal and frequency resolutions on the performance, using a trellis diagram to represent the stride space. We further identify two optimal points, namelyGolden Gemini, which serves as a guiding principle for designing 2D ResNet-based ASV models. By following the principle, a state-of-the-art ResNet baseline model gains a significant performance improvement on VoxCeleb, SITW, and CNCeleb datasets with 7.70%/11.76% average EER/minDCF reductions, respectively, across different network depths (ResNet18, 34, 50, and 101), while reducing the number of parameters by 16.5% and FLOPs by 4.1%. We refer to it asGeminiResNet. Further investigation reveals the efficacy of the proposedGolden Geminioperating points across various training conditions and architectures. Furthermore, we present a new benchmark, namely theGeminiDF-ResNet, using a cutting-edge model.Codes and pre-trained models are available athttps://github.com/Tianchi-Liu9/Golden-Gemini-for-Speaker-Verification.

Speech utterances recorded under differing conditions exhibit varying degrees of confidence in their embedding estimates, i.e., uncertainty, even if they are extracted using the same neural network. This paper aims to incorporate the uncertainty estimate produced in the xi-vector network front-end with a probabilistic linear discriminant analysis (PLDA) back-end scoring for speaker verification. To achieve this we derive a posterior covariance matrix, which measures the uncertainty, from the frame-wise precisions to the embedding space. We propose a log-likelihood ratio function for the PLDA scoring with the uncertainty propagation. We also propose to replace the length normalization pre-processing technique with a length scaling technique for the application of uncertainty propagation in the back-end. Experimental results on the VoxCeleb-1, SITW test sets as well as a domain-mismatched CNCeleb1-E set show the effectiveness of the proposed techniques with 14.5%–41.3% EER reductions and 4.6%–25.3% minDCF reductions.

For speaker recognition, it is difficult to extract an accurate speaker representation from speech because of its mixture of speaker traits and content. This paper proposes a disentanglement framework that simultaneously models speaker traits and content variability in speech. It is realized with the use of three Gaussian inference layers, each consisting of a learnable transition model that extracts distinct speech components. Notably, a strengthened transition model is specifically designed to model complex speech dynamics. We also propose a self-supervision method to dynamically disentangle content without the use of labels other than speaker identities. The efficacy of the proposed framework is validated via experiments conducted on the VoxCeleb and SITW datasets with 9.56\% and 8.24\% average reductions in EER and minDCF, respectively. Since neither additional model training nor data is specifically needed, it is easily applicable in practical use.

The time delay neural network (TDNN) represents one of the state-of-the-art of neural solutions to text-independent speaker verification. However, they require a large number of filters to capture the speaker characteristics at any local frequency region. In addition, the performance of such systems may degrade under short utterance scenarios. To address these issues, we propose a multi-scale frequency-channel attention (MFA), where we characterize speakers at different scales through a novel dual-path design which consists of a convolutional neural network and TDNN. We evaluate the proposed MFA on the VoxCeleb database and observe that the proposed framework with MFA can achieve state-of-the-art performance while reducing parameters and computation complexity. Further, the MFA mechanism is found to be effective for speaker verification with short test utterances.

The emergence of large-margin softmax cross-entropy losses in training deep speaker embedding neural networks has triggered a gradual shift from parametric back-ends to a simpler cosine similarity measure for speaker verification. Popular parametric back-ends include the probabilistic linear discriminant analysis (PLDA) and its variants. This paper investigates the properties of margin-based cross-entropy losses leading to such a shift and aims to find scoring back-ends best suited for speaker verification. In addition, we revisit the pre-processing techniques which have been widely used in the past and assess their effectiveness on large-margin embeddings. Experiments on the state-of-the-art ECAPA-TDNN networks trained with various large-margin softmax cross-entropy losses show a substantial increment in intra-speaker compactness making the conventional PLDA superfluous. In this regard, we found that constraining the within-speaker covariance matrix could improve the performance of the PLDA. It is demonstrated through a series of experiments on the VoxCeleb-1 and SITW core-core test sets with 40.8% equal error rate (EER) reduction and 35.1% minimum detection cost (minDCF) reduction. It also outperforms cosine scoring consistently with reductions in EER and minDCF by 10.9% and 4.9%, respectively.

Traditional acoustic-phonetic approach makes use of both spectral and phonetic information when comparing the voice of speakers. While phonetic units are not equally informative, the phonetic context of speech plays an important role in speaker verification (SV). In this paper, we propose a neural acoustic-phonetic approach that learns to dynamically assign differentiated weights to spectral features for SV. Such differentiated weights form a phonetic attention mask (PAM). The neural acoustic-phonetic framework consists of two training pipelines, one for SV and another for speech recognition. Through the PAM, we leverage the phonetic information for SV. We evaluate the proposed neural acoustic-phonetic framework on the RSR2015 database Part III corpus, that consists of random digit strings. We show that the proposed framework with PAM consistently outperforms baseline with an equal error rate reduction of 13.45% and 10.20% for female and male data, respectively.

In this paper, we study a novel technique that exploits the interaction between speaker traits and linguistic content to improve both speaker verification and utterance verification performance. We implement an idea of speaker-utterance dual attention (SUDA) in a unified neural network. The dual attention refers to an attention mechanism for the two tasks of speaker and utterance verification. The proposed SUDA features an attention mask mechanism to learn the interaction between the speaker and utterance information streams. This helps to focus only on the required information for respective task by masking the irrelevant counterparts. The studies conducted on RSR2015 corpus confirm that the proposed SUDA outperforms the framework without attention mask as well as several competitive systems for both speaker and utterance verification.