Scicom-intl/Emilia-YODAS-Voice-Conversion
收藏Hugging Face2026-02-09 更新2026-03-29 收录
下载链接：
https://hf-mirror.com/datasets/Scicom-intl/Emilia-YODAS-Voice-Conversion
下载链接
链接失效反馈官方服务：
资源简介：
---
language:
- de
- en
- fr
- ja
- ko
- zh
- ms
configs:
- config_name: audio_length_ratio_text
data_files:
- split: train
path: audio_length_ratio_text/train-*
- config_name: audio_text
data_files:
- split: train
path: audio_text/train-*
- config_name: default
data_files:
- split: train
path: data/train-*
- config_name: original
data_files:
- split: train
path: original/train-*
dataset_info:
- config_name: audio_length_ratio_text
features:
- name: audio_filename
dtype: string
- name: audio_filename_trim
dtype: string
- name: audio_length
dtype: float64
- name: text
dtype: string
- name: audio_length_ratio_text
dtype: float64
- name: audio_length_ratio_text_accept
dtype: bool
splits:
- name: train
num_bytes: 3249280650
num_examples: 11365350
download_size: 1360996742
dataset_size: 3249280650
- config_name: audio_text
features:
- name: audio_filename
dtype: string
- name: text
dtype: string
splits:
- name: train
num_bytes: 2486382629
num_examples: 11365354
download_size: 1192189223
dataset_size: 2486382629
- config_name: default
features:
- name: reference_audio
dtype: string
- name: reference_text
dtype: string
- name: target_audio
dtype: string
- name: target_text
dtype: string
splits:
- name: train
num_bytes: 15168651049
num_examples: 32845483
download_size: 3454397610
dataset_size: 15168651049
- config_name: original
features:
- name: text
dtype: string
- name: duration
dtype: float64
- name: speaker
dtype: string
- name: language
dtype: string
- name: dnsmos
dtype: float64
- name: phone_count
dtype: int64
- name: _id
dtype: string
splits:
- name: train
num_bytes: 2940653776
num_examples: 11365354
download_size: 1629097675
dataset_size: 2940653776
---
# Emilia-YODAS-Voice-Conversion
We sample https://huggingface.co/datasets/amphion/Emilia-Dataset YODAS set for voice conversion.
1. Filter transcriptions based on character repetitiveness and word ngrams.
2. Filter speaker similarity using https://huggingface.co/nvidia/speakerverification_en_titanet_large during speaker permutation.
3. Convert audio to speech tokens using https://huggingface.co/neuphonic/neucodec
We also upload the full permutation as zip files.
## Speech Tokenizer
Convert audio to speech tokens using https://huggingface.co/neuphonic/neucodec 50Hz, **with total 5.7B speech tokens**.
## Statistics
1. DE, 5558.53 hours.
2. EN, 13493.57 hours.
3. FR, 6954.43 hours.
4. JA, 1120.36 hours.
5. KO, 6991.33 hours.
6. ZH, 326.01 hours.
提供机构：
Scicom-intl
