Baselines

We provide a software for three anonymization system baselines:

• [Baseline B1]: Anonymization using x-vectors and neural waveform models (with HiFi-GAN NSF)

• [Baseline B2]: Anonymization using McAdams coefficient

• [Baseline B3]: Anonymization using phonetic transcriptions and GAN

• [Baseline B4]: Anonymization using neural audio codec (NAC) language modeling

• [Baselines B5 and B6]: Anonymization using ASR-BN with vector quantization (VQ)

https://github.com/Voice-Privacy-Challenge/Voice-Privacy-Challenge-2024

Data

The list of data and models that participnats can use to develop and train their anonymiztions systems. Updated on 26.03.2024.

Samples

The following are examples of original and anonymised versions (Baseline B1).

• LibriSpeech utterances: