Japanese-Speaking Transcription Specialist

You will listen to short Japanese recordings and create clean, standardized verbatim transcripts. The goal is to capture what was said accurately, not to translate, rewrite, or add creative content.

You will also review audio and use set options to flag unclear or unusable recordings and mark sections that cannot be understood with confidence. Careful judgment and consistent use of the transcription rules matter more than speed.

• Transcribe short Japanese audio clips accurately.
• Use kanji, kana, okurigana, and punctuation according to the required writing conventions.
• Follow detailed transcription rules without improvising.
• Mark unclear or unusable audio with the provided options.
• Complete listening-based verification and evaluation tasks.

What it pays and takes

This is an ongoing contractor role for an intermediate-level contributor. Prior transcription, captioning, subtitling, or speech data labeling and quality assurance experience is preferred but not required.

• Pay: $15 per hour.
• Time: About 1 hour per day, Monday through Friday, or less than 20 hours per week.
• Schedule: Flexible each day, while meeting the required daily cadence.
• Duration: Long-term work for 12 months or more.
• Location: You must currently live in Japan.
• Language: Fluent Japanese, with professional-level written accuracy; bilingual English is required.
• Skills: Strong listening, careful judgment, and consistent attention to detail.
• Equipment: Reliable internet, a computer for web-based tools, quality headphones, and a quiet work environment.
• Before starting: Complete a short Japanese transcription screening sample.

How it works

Apply on OpenTrain. The employer reviews applications there.

About AI training work

AI training work uses examples prepared and checked by people to help artificial intelligence understand language, audio, images, and other information. Transcription and audio review help create accurate examples for speech systems, and careful contributors are needed because consistent human judgment improves the data.