metadata
license: cc-by-4.0
dataset_info:
features:
- name: file_id
dtype: string
- name: audio
dtype: audio
- name: text_scribe
dtype: string
- name: events_scribe
dtype: string
- name: CE
dtype: float64
- name: CU
dtype: float64
- name: PC
dtype: float64
- name: PQ
dtype: float64
- name: text_emilia
dtype: string
- name: duration
dtype: float64
- name: speaker
dtype: string
- name: language
dtype: string
- name: dnsmos
dtype: float64
- name: phone_count
dtype: int64
splits:
- name: train
num_bytes: 120416127175.768
num_examples: 244562
download_size: 116352684092
dataset_size: 120416127175.768
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
language:
- de
size_categories:
- 100K<n<1M
the dataset is 680h out of the german part from https://huggingface.co/datasets/amphion/Emilia-Dataset ( Emilia Yodas - cc by 4.0)
audio event classified via scribe v1 (elevenlabs stt/asr)
- facebook audio aestetics to be used as prefilter
the dataset is very much at a v1 -
if you want to help - lets talk
https://discord.gg/RUs3uzBdW3 (nsfw is fully opt in only - as sfw)
if you want full transaction timestamps as they come from scribe v1 - they are cc by 4.0 NC and can be found here https://huggingface.co/datasets/MrDragonFox/DE_Emilia_Yodas_680h_raw_timestamps