Baseline vocoder
Speech synthesis training without a detector-specific adversarial objective.
Detector target: noneResearch audio companion
The same words. The same voice.
Three ways to shape the signal.
Politecnico di Milano
Hear the comparison
Choose a voice and a prompt.
Switch vocoders to compare the details.
Loading the audio collection…
Headphones recommended. These are the recordings from the original audio collection; no audio processing is applied during playback.
Behind the sound
The study explores adversarial fine-tuning of a BigVGAN vocoder: retaining speech synthesis objectives while adding feedback from forensic deepfake detectors.
Compare how the training objective changes across the three variants. AASIST is held out of vocoder fine-tuning to evaluate transfer to another detector.
Research code is being prepared for public release.
Speech synthesis training without a detector-specific adversarial objective.
Detector target: noneAdversarial fine-tuning against RawNet2 in a white-box setting.
Detector target: RawNet2Adversarial fine-tuning against RawNet2 and TSSD jointly.
Detector targets: RawNet2 + TSSDFour training-set speakers and two unseen voices are included. Gorou and Yae Miko are character voices from Genshin Impact by miHoYo, as identified in the original collection. Reference clips demonstrate the speaker’s voice and contain different text from the synthesized prompts.
Prompt excerpts are reconstructed from the source filenames. Listening examples illustrate the paper’s methods; detector performance is evaluated separately in the study.