Cross-Technology Generalization in Synthesized Speech Detection: Evaluating AST Models with Modern Voice Generators
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908288631701504 |
|---|---|
| author | Ustinov, Andrew Yordanov, Matey Kuchma, Andrei Bychkov, Mikhail |
| author_facet | Ustinov, Andrew Yordanov, Matey Kuchma, Andrei Bychkov, Mikhail |
| contents | This paper evaluates the Audio Spectrogram Transformer (AST) architecture for synthesized speech detection, with focus on generalization across modern voice generation technologies. Using differentiated augmentation strategies, the model achieves 0.91% EER overall when tested against ElevenLabs, NotebookLM, and Minimax AI voice generators. Notably, after training with only 102 samples from a single technology, the model demonstrates strong cross-technology generalization, achieving 3.3% EER on completely unseen voice generators. This work establishes benchmarks for rapid adaptation to emerging synthesis technologies and provides evidence that transformer-based architectures can identify common artifacts across different neural voice synthesis methods, contributing to more robust speech verification systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_22503 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Cross-Technology Generalization in Synthesized Speech Detection: Evaluating AST Models with Modern Voice Generators Ustinov, Andrew Yordanov, Matey Kuchma, Andrei Bychkov, Mikhail Sound Cryptography and Security Audio and Speech Processing This paper evaluates the Audio Spectrogram Transformer (AST) architecture for synthesized speech detection, with focus on generalization across modern voice generation technologies. Using differentiated augmentation strategies, the model achieves 0.91% EER overall when tested against ElevenLabs, NotebookLM, and Minimax AI voice generators. Notably, after training with only 102 samples from a single technology, the model demonstrates strong cross-technology generalization, achieving 3.3% EER on completely unseen voice generators. This work establishes benchmarks for rapid adaptation to emerging synthesis technologies and provides evidence that transformer-based architectures can identify common artifacts across different neural voice synthesis methods, contributing to more robust speech verification systems. |
| title | Cross-Technology Generalization in Synthesized Speech Detection: Evaluating AST Models with Modern Voice Generators |
| topic | Sound Cryptography and Security Audio and Speech Processing |
| url | https://arxiv.org/abs/2503.22503 |