The work
You receive batches of short Thai audio clips and score them on nativeness and fluency: tone accuracy across the five tones, vowel length distinctions, final consonant realization, prosody and phrase-level intonation, and whether the speech sounds like a Thai speaker or like a model approximating one. Ratings alone aren't the deliverable — each one needs a concise English justification that names what you heard, so downstream engineers can trace a low score to a specific defect rather than a general impression.
Expect a rubric, calibration examples, and periodic gold-standard checks. Where clips are ambiguous — regional accent versus non-nativeness, a Isan or Southern speaker versus a flawed synthesis, code-mixed English loanwords — you flag and escalate to coordinators instead of guessing. Consistency across a long batch matters more than being clever on any single item.
What the screen looks for
micro1's screening is AI-led and conversational. It probes native-level Thai command, whether you can describe phonetic phenomena in English precisely enough to be useful (B2 or better in writing and speech), and whether you have any grounding in phonetics, voice work, linguistics, transcription, audio engineering, or language teaching. It also tests judgment: given a borderline clip, can you explain your reasoning and hold it up under follow-up questions rather than restating the conclusion?
Logistics
- Fully remote, contractor engagement, asynchronous within batch deadlines
- Project-based volume; hours fluctuate with client demand and are not guaranteed
- Observed pay band for this listing: $30–65/hr, typically set by background depth and calibration performance
- Requires quiet listening conditions and decent headphones — laptop speakers won't resolve the distinctions being scored