Challenge: Existing methods for deepfake detection fail under speech-to-singing domain shift . a speech-retentive multi-domain fine-tuning strategy enables adaptation to singing .
Approach: They propose a unified deepfake detector based on a multi-branch mixture-of-experts architecture that integrates three complementary feature views.
Outcome: The proposed detector achieves 1.82% EER on CtrSVDD, compared to 37–62% for existing detectors . it can generalize to unseen generators and preserve strong speech performance .

Similar Papers

A Data-Centric Approach to Generalizable Speech Deepfake Detection (2026.acl-long)

Copied to clipboard

Challenge: Speech deepfake detection (SDD) is a critical research area as speech synthesis technologies become more sophisticated.
Approach: They propose a data-centric approach to generalize SDD data from two perspectives . they propose naive aggregation strategies for mixing heterogeneous data and diversity-optimized sampling strategy for a single dataset and multiple datasets.
Outcome: The proposed approach outperforms the naive aggregation baseline on a 12k-hour data pool while using only 3% of the total available data.
SpeechFake: A Large-Scale Multilingual Speech Deepfake Dataset Incorporating Cutting-Edge Generation Methods (2025.acl-long)

Copied to clipboard

Challenge: Existing speech deepfake datasets are limited in scale and diversity, making it challenging to train models that can generalize well to unseen deepfakkes.
Approach: They propose a large-scale speech deepfake dataset that includes over 3 million deepfak samples, totaling more than 3,000 hours of audio, generated using 40 different speech synthesis tools.
Outcome: The proposed dataset includes over 3 million deepfake samples, totaling more than 3,000 hours of audio, generated using 40 different speech synthesis tools.
Heterogeneity over Homogeneity: Investigating Multilingual Speech Pre-Trained Models for Detecting Audio Deepfake (2024.findings-naacl)

Copied to clipboard

Challenge: a recent study has focused on audio deepfake detection (ADD) due to its ability to impersonate and share false, often malicious information.
Approach: They propose to use multilingual speech Pre-Trained models for Audio deepfake detection (ADD) they propose to combine models with existing models to achieve better ADD detection .
Outcome: The proposed models gain knowledge about diverse pitches, accents, and tones, during theirpre-training phase and are more robust to variations.
RTCFake: Speech Deepfake Detection in Real-Time Communication (2026.findings-acl)

Copied to clipboard

Challenge: Existing detection studies focus on offline simulations and struggle to cope with complex distortions introduced during RTC transmission.
Approach: They propose a large-scale speech deepfake dataset tailored for RTC scenarios . the dataset is constructed by transmitting speech through multiple social media and conferencing platforms .
Outcome: The proposed dataset is constructed by transmitting speech through multiple mainstream social media and conferencing platforms, enabling precise pairing between offline and online speech.
UniSpeaker: A Unified Approach for Multimodality-driven Speaker Generation (2025.findings-emnlp)

Copied to clipboard

Challenge: a new framework for speaker generation is proposed to enable multimodal speaker generation . multimodal cues such as visual appearance, textual descriptions, and other biometric signals are still in its early stages.
Approach: a new framework is proposed to enable multimodal speaker generation . the framework uses self-distillation to apply speaker disentanglement to speech generation a model is developed .
Outcome: The proposed framework is the first to support unified voice generation from arbitrary modality combinations.
UniMoE-Audio: Unified Speech and Music Generation with Dynamic-Capacity Mixture-of-Experts (2026.acl-long)

Copied to clipboard

Challenge: Recent advances in unified multimodal models indicate a clear trend towards comprehensive content generation.
Approach: They propose a unified speech and music generation model built upon a novel framework . they propose specialized MoE architectures and curated training strategies to tackle data imbalances .
Outcome: The proposed model achieves state-of-the-art performance on major speech and music generation benchmarks.
Double Entendre: Robust Audio-Based AI-Generated Lyrics Detection via Multi-View Fusion (2025.findings-acl)

Copied to clipboard

Challenge: Existing methods for detecting AI-generated music are weak and vulnerable to audio perturbations.
Approach: They propose a multimodal late-fusion pipeline that combines automatically transcribed sung lyrics and speech features capturing lyrics related information within the audio.
Outcome: The proposed method outperforms existing detectors while being more robust to audio perturbations.
Cross-Domain Audio Deepfake Detection: Dataset and Analysis (2024.emnlp-main)

Copied to clipboard

Challenge: Existing audio deepfake detection datasets are outdated and lack generalization capabilities.
Approach: They construct a new cross-domain audio deepfake detection dataset comprising over 300 hours of speech data that is generated by five advanced zero-shot TTS models.
Outcome: The proposed models achieve 4.1% and 6.5% error rates in the cross-domain ADD dataset generated by five advanced zero-shot TTS models.
XMAD-Bench: Cross-Domain Multilingual Audio Deepfake Benchmark (2026.findings-eacl)

Copied to clipboard

Challenge: Recent advances in audio generation led to an increasing number of deepfakes . however, these methods are typically tested in an in-domain setup .
Approach: They propose a large-scale cross-domain audio deepfake benchmark comprising 668.8 hours of real and deepfak speech.
Outcome: The proposed benchmark compares audio deepfake detectors with existing methods in the wild . the results show that the proposed methods perform better in different languages than existing methods .
Revealing the Truth with ConLLM for Detecting Multi-Modal Deepfakes (2026.findings-eacl)

Copied to clipboard

Challenge: Existing methods for deepfake detection suffer from two limitations: modality fragmentation and shallow inter-modal reasoning.
Approach: They propose a framework for multimodal deepfake detection that uses contrastive learning and large language models to mitigate modality fragmentation and refine embeddings to address shallow inter-modal reasoning.
Outcome: ConLLM reduces audio deepfake EER by 50%, improves video accuracy by 8%, and achieves approximately 9% accuracy gains in audio-visual tasks.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations