Can LLMs Help Localize Fake Words in Partially Fake Speech?

Large language models (LLMs), trained on large-scale text, have recently attracted significant attention for their strong performance across many tasks. Motivated by this, we investigate whether a text-trained LLM can help localize fake words in partially fake speech, where only specific words within a speech are edited. We build a speech LLM to perform fake word localization via next token prediction. Experiments and analyses on AV-Deepfake1M and PartialEdit indicates that the model frequently leverages editing-style pattern learned from the training data, particularly word-level polarity substitutions for those two databases we discussed, as cues for localizing fake words. Although such particular patterns provide useful information in an in-domain scenario, how to avoid over-reliance on such particular pattern and improve generalization to unseen editing styles remains an open question.

Universal Speech Content Factorization

We propose Universal Speech Content Factorization (USCF), a simple and invertible linear method for extracting a low-rank speech representation in which speaker timbre is suppressed while phonetic content is preserved. USCF extends Speech Content Factorization, a closed-set voice conversion (VC) method, to an open-set setting by learning a universal speech-to-content mapping via least-squares optimization and deriving speaker-specific transformations from only a few seconds of target speech. We show through embedding analysis that USCF effectively removes speaker-dependent variation. As a zero-shot VC system, USCF achieves competitive intelligibility, naturalness, and speaker similarity compared to methods that require substantially more target-speaker data or additional neural training. Finally, we demonstrate that as a training-efficient timbre-disentangled speech feature, USCF features can serve as the acoustic representation for training timbre-prompted text-to-speech models. Speech samples and code are publicly available.

Content Anonymization for Privacy in Long-form Audio

Voice anonymization techniques have been found to successfully obscure a speaker's acoustic identity in short, isolated utterances in benchmarks such as the VoicePrivacy Challenge. In practice, however, utterances seldom occur in isolation: long-form audio is commonplace in domains such as interviews, phone calls, and meetings. In these cases, many utterances from the same speaker are available, which pose a significantly greater privacy risk: given multiple utterances from the same speaker, an attacker could exploit an individual's vocabulary, syntax, and turns of phrase to re-identify them, even when their voice is completely disguised. To address this risk, we propose new content anonymization approaches. Our approach performs a contextual rewriting of the transcripts in an ASR-TTS pipeline to eliminate speaker-specific style while preserving meaning. We present results in a long-form telephone conversation setting demonstrating the effectiveness of a content-based attack on voice-anonymized speech. Then we show how the proposed content-based anonymization methods can mitigate this risk while preserving speech utility. Overall, we find that paraphrasing is an effective defense against content-based attacks and recommend that stakeholders adopt this step to ensure anonymity in long-form audio.

Scalable Controllable Accented TTS

We tackle the challenge of scaling accented TTS systems, expanding their capabilities to include much larger amounts of training data and a wider variety of accent labels, even for accents that are poorly represented or unlabeled in traditional TTS datasets. To achieve this, we employ two strategies: 1. Accent label discovery via a speech geolocation model, which automatically infers accent labels from raw speech data without relying solely on human annotation; 2. Timbre augmentation through kNN voice conversion to increase data diversity and model robustness. These strategies are validated on CommonVoice, where we fine-tune XTTS-v2 for accented TTS with accent labels discovered or enhanced using geolocation. We demonstrate that the resulting accented TTS model not only outperforms XTTS-v2 fine-tuned on self-reported accent labels in CommonVoice, but also existing accented TTS benchmarks.

Less is More for Synthetic Speech Detection in the Wild

Driven by advances in self-supervised learning for speech, state-of-the-art synthetic speech detectors have achieved low error rates on popular benchmarks such as ASVspoof. However, prior benchmarks do not address the wide range of real-world variability in speech. Are reported error rates realistic in real-world conditions? To assess detector failure modes and robustness under controlled distribution shifts, we introduce ShiftySpeech, a benchmark with more than 3000 hours of synthetic speech from 7 domains, 6 TTS systems, 12 vocoders, and 3 languages. We found that all distribution shifts degraded model performance, and contrary to prior findings, training on more vocoders, speakers, or with data augmentation did not guarantee better generalization. In fact, we found that training on less diverse data resulted in better generalization, and that a detector fit using samples from a single carefully selected vocoder and a single speaker achieved state-of-the-art results on the challenging In-the-Wild benchmark.

GenVC: Self-Supervised Zero-Shot Voice Conversion

Zero-shot voice conversion has recently made substantial progress, but many models still depend on external supervised systems to disentangle speaker identity and linguistic content. Furthermore, current methods often use parallel conversion, where the converted speech inherits the source utterance's temporal structure, restricting speaker similarity and privacy. To overcome these limitations, we introduce GenVC, a generative zero-shot voice conversion model. GenVC learns to disentangle linguistic content and speaker style in a self-supervised manner, eliminating the need for external models and enabling efficient training on large, unlabeled datasets. Experimental results show that GenVC achieves state-of-the-art speaker similarity while maintaining naturalness competitive with leading approaches. Its autoregressive generation also allows the converted speech to deviate from the source utterance's temporal structure. This feature makes GenVC highly effective for voice anonymization, as it minimizes the preservation of source prosody and speaker characteristics, enhancing privacy protection.

HLTCOE JHU Submission to the Voice Privacy Challenge 2024

We present a number of systems for the Voice Privacy Challenge, including voice conversion based systems such as the kNN-VC method and the WavLM voice Conversion method, and text-to-speech (TTS) based systems including Whisper-VITS. We found that while voice conversion systems better preserve emotional content, they struggle to conceal speaker identity in semi-white-box attack scenarios; conversely, TTS methods perform better at anonymization and worse at emotion preservation. Finally, we propose a random admixture system which seeks to balance out the strengths and weaknesses of the two category of systems, achieving a strong EER of over 40% while maintaining UAR at a respectable 47%.

Privacy versus Emotion Preservation Trade-offs in Emotion-Preserving Speaker Anonymization

Advances in speech technology now allow unprecedented access to personally identifiable information through speech. To protect such information, the differential privacy field has explored ways to anonymize speech while preserving its utility, including linguistic and paralinguistic aspects. However, anonymizing speech while maintaining emotional state remains challenging. We explore this problem in the context of the VoicePrivacy 2024 challenge. Specifically, we developed various speaker anonymization pipelines and find that approaches either excel at anonymization or preserving emotion state, but not both simultaneously. Achieving both would require an in-domain emotion recognizer. Additionally, we found that it is feasible to train a semi-effective speaker verification system using only emotion representations, demonstrating the challenge of separating these two modalities.

Self-supervised Reflective Learning through Self-distillation and Online Clustering for Speaker Representation Learning

Speaker representation learning is critical for modern voice recognition systems. While supervised learning techniques require extensive labeled data, unsupervised methodologies can leverage vast unlabeled corpora, offering a scalable solution. This paper introduces self-supervised reflective learning (SSRL), a novel paradigm that streamlines existing iterative unsupervised frameworks. SSRL integrates self-supervised knowledge distillation with online clustering to refine pseudo labels and train the model without iterative bottlenecks. Specifically, a teacher model continually refines pseudo labels through online clustering, providing dynamic supervision signals to train the student model. The student model undergoes noisy student training with input and model noise to boost its modeling capacity. The teacher model is updated via an exponential moving average of the student, acting as an ensemble of past iterations. Further, a pseudo label queue retains historical labels for consistency, and noisy label modeling directs learning towards clean samples. Experiments on VoxCeleb show SSRL's superiority over current iterative approaches, surpassing the performance of a 5-round method in just a single training round. Ablation studies validate the contributions of key components like noisy label modeling and pseudo label queues. Moreover, consistent improvements in pseudo labeling and the convergence of cluster counts demonstrate SSRL's effectiveness in deciphering unlabeled data. This work marks an important advancement in efficient and accurate speaker representation learning through the novel reflective learning paradigm.

The DKU-DUKEECE System for the Manipulation Region Location Task of ADD 2023

This paper introduces our system designed for Track 2, which focuses on locating manipulated regions, in the second Audio Deepfake Detection Challenge (ADD 2023). Our approach involves the utilization of multiple detection systems to identify splicing regions and determine their authenticity. Specifically, we train and integrate two frame-level systems: one for boundary detection and the other for deepfake detection. Additionally, we employ a third VAE model trained exclusively on genuine data to determine the authenticity of a given audio clip. Through the fusion of these three systems, our top-performing solution for the ADD challenge achieves an impressive 82.23% sentence accuracy and an F1 score of 60.66%. This results in a final ADD score of 0.6713, securing the first rank in Track 2 of ADD 2023.