|
--- |
|
task_categories: |
|
- text-to-speech |
|
language: |
|
- zh |
|
--- |
|
# Dataset Card for Nexdata/Chinese_Mandarin_Entertainment_anchor_Style_Multi-emotional_Synthesis_Corpus |
|
|
|
## Description |
|
12 Hours - Chinese Mandarin Entertainment anchor Style Multi-emotional Synthesis Corpus. It is recorded by Chinese native speaker. six emotional text+modal particles, phonemes and tones are balanced. Professional phonetician participates in the annotation. It precisely matches with the research and development needs of the speech synthesis. |
|
|
|
For more details, please refer to the link: https://www.nexdata.ai/datasets/1304?source=Huggingface |
|
|
|
# Specifications |
|
## Format |
|
48,000Hz, 24bit, uncompressed wav, mono channel |
|
## Recording environment |
|
professional recording studio |
|
## Recording content |
|
seven emotions (happiness, anger, sadness, surprise, fear, disgust)+sentences with filler word |
|
## Speaker |
|
professional CharacterVoice; Role: An 18-year-old girl who works as an entertainment anchor and enjoys singing and dancing |
|
## Device |
|
microphone |
|
## Language |
|
Mandarin |
|
## Annotation |
|
word and pinyin transcription, prosodic boundary annotation, phoneme boundary annotation |
|
## The amount of data |
|
The amount of neutral data is not less than 1.6 hours; the amount of data with filler word is not less than 0.4 hours; and the remaining six types of emotional data is not less than 1.67 hours each |
|
|
|
# Licensing Information |
|
Commercial License |