HABLA-Open-Test

Overview

This dataset contains the open-set evaluation to be used in conjunction with the HABLA-Augmented dataset, although it can be used to evaluate any audio DeepFake detection for Spanish.

Open-test

To obtain the open-test subset the original set was split in two subsets by speakers and synthesis models. First, 2 out of the 6 available synthesis models were reserved exclusively for testing. Additionally 20 % of the speakers were selected randomly from each accent and sex. The combination of both subsets generated the open-test subset.

| Model       | # Samples | % Set |
|-------------|-----------|-------|
| StarGAN     | 16000     | 46.4  |
| CycleGAN    | 6200      | 18    |
| Diff        | 4160      | 12    |
| bonafide    | 4075      | 11.85 |
| TTS-Diff    | 2378      | 6.9   |
| TTS         | 857       | 2.5   |
| TTS-StarGAN | 808       | 2.35  |

with a proportion of samples (labels)

| Label    | # Samples | % Set |
|----------|-----------|-------|
| bonafide | 30403     | 0.12  |
| spoof    |  4075     | 0.88  |

Metadata

All subset contain metadata to facilitate filtering and allow for experimentation, the fields contained are the following

• speaker_id: speaker unique ID

  ex: 'arf_06136'

• country: speaker's accent

  values: {Argentina, Chile, Colombia, Peru, Venezuela}

• sex: speaker's sex

  values: {Female, Male}

• file_name: actual audio file name

  ex: 'arf_00610_00006739039.wav'

• augmentation_algorithm: Integer encoding each rawboost augmentation algorithm utilized to augment audio data

  values: {0, 4, 5, 6, 7}

  notes: The encoding is described by

  0 -> No augmentation (original), 4 -> Series Convolutive-Impulsive-Stationary noise 5 -> Series Convolutive-Impulsive noise 6 -> Series Convolutive-Stationary noise 7 -> Series Impulsive-Stationary noise

• model: synthesized model utilized to generate audio

  values: {'CycleGAN', 'Diff', 'StarGAN', 'TTS', 'TTS-Diff', 'TTS-StarGAN', '-'} notes: '-' placeholder utilized in bonafide samples

• label: category to which the audio belongs

  values: {spoof, bonafide}