Native Japanese prose, conversation, and register.
Web-scale Japanese is dominated by a narrow slice of registers. Curated, rights-cleared text in business, regional and professional registers is where models still fall short.
Describe the data you need. We will respond with feasibility, rights approach and next steps.
Unscripted two- and multi-party conversations between native speakers across everyday topics, recorded with consent and transcribed.
Honorific, humble and polite registers in realistic business messages, with annotators marking appropriateness by relationship and context.
Rights-cleared long-form writing from professionals such as engineers, lawyers and journalists, for domain-grounded pretraining and fine-tuning.
Instruction-following pairs written by native experts rather than machine-translated, covering culturally specific tasks.
Written and transcribed material in regional varieties (Kansai, Tohoku, Kyushu, Okinawan and others) with standard-Japanese glosses.
Professionally translated parallel text in specified domains, built to your terminology and style guide.
Technical manuals, specifications and engineering writing in Japanese, paired with code where rights allow.
Not here? Describe the data you need. We source beyond this list.