Beijing TTS recording center consists of one master control room and 2 professional recording rooms, each equipped with independent control system. Our recording center has passed the quality check of Tsinghua University Building Environment, and comply with professional NR15 acoustic standard. The reverberation time is less than 0.1s, and the background noise is lower than 30dB(A). It can support professional voice cast, normal people TTS data and the pre-end model production.
Hefei Data Base is located in the Big Data Town of Shushan Economic Development Zone, covering an area of 1500 square meters, and can accommodate 500 professional annotators to work simultaneously. Hefei Data Base has continuously cultivated and hatched artificial intelligence industrial chain enterprises since its establishment. It is able to provide multiple services such as image data annotation, speech data annotation, foreign image and speech collection & annotation service.
Baoding Data Base covers an area of 1200 square meters and has 200 full-time annotators, 60% of whom are senior annotators with more than 3 years of annotation experience. It has the ability to perform speech recognition, human face recognition, ORC recognition, smart driving, etc.
All of our staff have more than 5 years of work experience thus they are familiar with different kinds of data requirements and able to deeply understand clients’ application scenario.
Our annotators have more than 3 years of experience in data annotation, who are skilled in 3D point cloud annotation, segmentation annotation and TTS annotation. For new annotators, we provide a 90-days complete training system.
The data base is equipped with double entrance guard, 24 hours of network monitoring, and double network backups to ensure data security.
Professional QA team. More than 7 years of experience in project management and quality control. The data accuracy rate can reach to 96%-99% after rounds of QA. We make timely and dynamic quality control in the whole process of annotation to ensure to deliver data on time.
Human: living body, key points (human face, human body and gesture)、attributes
Scenario: 3D point cloud, LiDar data annotation
OCR: Q&A, games, multiple languages
Mandarin: natural dialogue, reading, interactive
Dialect: natural dialogue, reading
Foreign language: natural conversation, reading
NLP：multiple interactive annotation, entity annotation, text pronunciation annotation (polyphone, character, number)
TTS：fine annotation, coarse annotation (Pinyin, mixed Chinese and English), rhythm annotation (audio rhythm, text rhythm)