metadata
dataset_info:
features:
- name: audio
dtype: audio
- name: timbre_annotation
struct:
- name: casting_caption
dtype: string
- name: context_tags
struct:
- name: contemporary
sequence: string
- name: fantasy
sequence: string
- name: science_fiction
sequence: string
- name: guiding_letters
struct:
- name: casting_caption
dtype: string
- name: trait_caption
dtype: string
- name: listening_pleasantness
dtype: string
- name: tag_caption
dtype: string
- name: trait_caption
dtype: string
- name: trait_tags
struct:
- name: accent_region
dtype: string
- name: articulation_clarity
dtype: string
- name: baseline_speech_rate
dtype: string
- name: baseline_tension
dtype: string
- name: body_size_impression
dtype: string
- name: breathiness
dtype: string
- name: gender_expression
dtype: string
- name: language_register
dtype: string
- name: perceived_age
dtype: string
- name: phonation_type
dtype: string
- name: pitch_level
dtype: string
- name: resonance_placement
dtype: string
- name: timbre_brightness
dtype: string
- name: timbre_clarity
dtype: string
- name: timbre_nasality
dtype: string
- name: timbre_softness
dtype: string
- name: timbre_warmth
dtype: string
- name: vocal_health
dtype: string
- name: voice_commonness
dtype: string
splits:
- name: train
num_bytes: 10063249446.196
num_examples: 32974
download_size: 9253087530
dataset_size: 10063249446.196
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
A dataset for voice design.
~30K samples of voice design data (LAION)