Industry Solutions
Social Networks
Social Networks
Optimizing AI models with Magic Data AI data total solution.
Magic Data AI data solution in social network scenarios such as sentiment analysis, recommendation systems, virtual host, makes the platform intelligent.
Scenarios
Video
Subtitle Generator/Review&Comment Analysis/Content Classification
Livestreaming
Voice Changer/Realtime Captioning/Review&Comment Analysis
Virtual Host
Anchor/E-commerce livestreaming/Digital singer
Challenge
Imprecise voice recognition of special words and phrases and accented speaking
Large volume and great variety of comments and reviews
Stiff and unnatural response of virtual human
Large category and cross-domain topics of video
Annotator® AI-Assisted Annotation Platform
Audio Annotation
Text Annotation
Image Annotation
- Live Show - Content annotation (ASR)
- Virtual Host - Voice interaction annotation (ASR)
- Virtual Host - Text segmentation, part-of-speech, and phoneme annotation (TTS)
- Video - Video segmentation, content classification (Video)
- Live Show - Sentiment analysis of reviews (NLP)
- Virtual Host - Sentiment analysis of reviews (NLP)
- Video - Video segmentation, content classification (Video)
MD Dataset Portfolio
Speech Recognition
→
Text-to-Speech
→
Natural Language Understanding
→
OCR
→
Contact us for data collection and annotation service
Related Datasets
MDT-RJ009 Spanish Spoken Speech Dataset
This dataset is designed to train AI models that better understand spoken Spanish, enhancing natural interaction in speech recognition. It includes diverse real-life dialogues with high transcription accuracy. Key features like liaison and elision are carefully annotated, and punctuation reflects Spanish rhythm. Complete sentences support learning of complex verb forms, improving recognition robustness.
48kHz Multi-Speaker Speech Dataset for Voice Cloning--English
Play Audio
MDT-AG031 Mandarin Heavy Accent (Jiangsu) Conversational Speech Corpus
Play Audio
MDT-LD009 Cantonese Lexicon
MDT-LG003 Nanchang Dialect Lexicon
MDT-RJ001 Japanese Spoken Speech Dataset
This dataset is designed to train AI models that better understand spoken Japanese, improving natural interaction in speech recognition. It includes diverse real-life conversations with high transcription accuracy. Pitch accents and special syllables like sokuon and nasal sounds are carefully annotated. Punctuation reflects Japanese pause rhythms, helping models learn ellipses and emotional particles for more natural responses.