The work

You put on headphones and listen to German text-to-speech audiobooks the way an audiobook fan would — except you stop, rewind, and write down what went wrong. The error types are concrete: a word skipped or hallucinated, `1984` read as eins neun acht vier instead of neunzehnhundertvierundachtzig, `z. B.` expanded as the letters rather than zum Beispiel, a proper noun stressed on the wrong syllable, a question that lands with falling intonation, a click or spectral artifact between sentences. Each issue gets a category and a time span marked in the platform's review tool. You also step back and rate the listen overall: where the narration held the illusion and where it broke.

The volume matters as much as the precision. This is many hours of continuous audio per week, and the value you add is being just as sharp in hour four as in hour one. Reviewers who catch the obvious mispronunciations but drift past subtler phrasing problems late in a session are the ones whose data quality reports come back weak.

What the screen looks for

  • Native-level German (Germany variety) — regional pronunciation judgments, compound stress, and Fugen-s handling all come up, and the screen probes them with specifics rather than taking fluency on faith.
  • Genuine audiobook listening habits — expect to be asked what you've listened to recently and what a narrator did well or badly. Vague answers here read as a non-listener applying for a listening job.
  • Prior detail work — annotation, transcription, proofreading, subtitling, linguistics, editing. You don't need AI-project experience, though it helps.
  • Working English — enough to write clear issue notes and read guidelines. Fluency is not required.

Logistics

Remote and asynchronous, roughly 20 hours per week spread as about four hours per day. You can live anywhere, but candidates in German-speaking regions are preferred. You'll need quiet listening conditions and decent headphones — laptop speakers make artifact detection impossible. Pay is observed at $15–20/hr and is not guaranteed; rates on platform projects vary with task type and batch.