Dutch AI Narration Reviewer (Remote, Part-Time)

Obsidian; Amsterdam; Remote; EUR 17,000 - 28,000; Part time

Benefits offered by this job

Remote work

Flexible hours

Job summary

Obsidian is recruiting native Dutch speakers to evaluate AI-narrated Dutch audiobooks. You will listen as a avid audiobook listener but with an analytic ear, flagging where narration slips so models can improve.

The role is remote and part-time, about 20 hours per week. Prior experience in annotation, linguistics, or proofreading is beneficial, and you will log precise issues with exact spans to support model refinement.

Qualifications

• Native or near-native fluency in Dutch (Netherlands).
• Working English to report clearly to the team.
• Experience with annotation, labeling, transcription, proofreading, linguistics, or similar detail-oriented review work.
• Sharp, patient ear and discipline to stay accurate during long listening sessions.
• Reliable availability of about 20 hours per week on a part-time basis.

Responsibilities

• Listen to AI-narrated audiobooks in Dutch and assess naturalness, accuracy, and enjoyment.
• Spot and log errors in synthesized speech—skipped, added, mispronounced words, numbers, abbreviations, or intonation.
• Categorize issues and mark exact spans where they occur using review tooling.
• Report on overall listener satisfaction and where narration feels natural versus where it breaks the experience.
• Work consistently across many hours of audio without losing attention to detail.

Skills

Dutch fluency

English proficiency

Annotation experience

Attention to detail

Listening skills

Job description

Obsidian is recruiting native Dutch speakers to evaluate AI-narrated Dutch audiobooks. You will listen as a avid audiobook listener but with an analytic ear, flagging where narration slips so models can improve.

The role is remote and part-time, about 20 hours per week. Prior experience in annotation, linguistics, or proofreading is beneficial, and you will log precise issues with exact spans to support model refinement.