Conyac DATA : A Service for Creating Machine Learning Data
Conyac DATA is a service for creating machine learning data with the aim of improving the quality of programs such as speech recognition software and chatbots.
Contact us now
Utilizing the bilingual crowdsourcing platform service Conyac, we promptly deliver the creation of high volume data sets previously considered difficult to do.
Conyac will create and sell data for AI use according to your needs!
Speech data creation
We record various voice data by bilingual speakers around the world and create on-demand, multilingual speech data regardless of nationality, age, gender or language.
We will create data from scratch for building chat systems designed to take inquiries and provide customer support.
Multilingual corpus data
Monolingual data creation is also possible in addition to parallel data. We will create a corpus that spans many fields.
Strengths of the bilingual platform Conyac
Creating high volume data is possible
Requests are handled manually by 93,000 bilingual speakers with a deep knowledge of various overseas cultures and practices, making high volume, yet useful data creation possible.
Creating data in a short period of time is possible
Even with large requests, we can assign many bilingual speakers, ensuring a quick delivery time.
Low cost data creation is possible
Outsourcing work to the 93,000 bilingual speakers around the world (the crowd) has made low costs a reality at Conyac.
We will propose preparation and deployment strategies that suit your budget.
At Conyac, apart from creating data for AI we also offer the following services:
The learning of poor-quality big data will lead to system degradation. We assess and refine large volumes of data in short periods of time.
AI + crowdsourcing
The power of people in a crowdsourcing group can be utilized to complement AI. Requests to subcontract work out to bilingual speakers around the world or to make revisions to output are possible.
We can expand existing data through crowdsourcing. Please contact us for details.
Examples of using Conyac to create data for AI
Speech data creation
Native English speakers are provided 500 sentences and asked to read each sentence once, recording their voice using a designated smartphone app.
Volume: 500 sentences (English),
500 recordings of speakers
Turnaround: 4 weeks
Chatbot data creation
Creating 3 answer patterns in Japanese for use when a native Japanese speaker asks a situational question to a device capable of understanding speech.
Volume: 20,000 sentences
Turnaround: 3 weeks
Parallel corpus creation
Creating 5 similar sentences in English for every 1 sentence written in Japanese.
Volume: 500,000 sentences
Turnaround: 1.5 months