metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: text
dtype: string
- name: normalized_text
dtype: string
splits:
- name: train
num_bytes: 3400049523.639
num_examples: 17597
download_size: 474483650
dataset_size: 3400049523.639
Dataset Card for "single_speaker_de_small_librivox"
Created for testing, not suggested for production
Dataset Summary
The corpus consists of ~28 hours from a single speaker extracted frrom LibriVox audiobook.
Languages
The audio is in German.
Source Data
Initial Data Collection and Normalization. The voices used in my Datasets are volenteers who have donated their time and voices to open source LibriVox projects. Please respect their privacy.
Licensing Information
MIT
More Information needed More Information needed