What the work actually involves

You receive batches of Portuguese audio and produce transcripts that match a client's style guide exactly. That guide decides everything contentious: whether filled pauses are written out, how false starts and repetitions are handled, whether numerals are spelled or digitised, how overlapping speech and crosstalk are marked, what tag set is used for non-speech events (laughter, background noise, inaudible spans). Beyond raw typing, you will annotate — speaker diarisation, timestamping, tagging unintelligible regions rather than guessing at them — and review other contributors' transcripts for consistency. Audio quality varies: phone-quality recordings, several speakers, background noise, and speakers from different Portuguese-speaking regions all show up in the same batch.

What the screen is looking for

micro1 runs an AI-led interview before any task work. It probes for native-level Portuguese comprehension across accents (European, Brazilian, African varieties), not just fluency in your own dialect. Expect follow-ups on concrete transcription conventions — how you would mark a two-second inaudible stretch, what you do when a speaker code-switches into English mid-sentence, how you decide between an orthographic and a verbatim rendering. Answers that stay abstract get pushed on. English proficiency matters mainly because instructions, guidelines and QA feedback arrive in English, and some audio will be English too.

Logistics and pay

  • Remote, contractor, asynchronous — work is claimed from available batches rather than scheduled shifts.
  • Volume is project-driven and can pause between client engagements; treat it as supplementary rather than a stable full-time load.
  • Observed pay for this listing is $10–18/hr, positioned by language, audio difficulty and review responsibility. Rates are as observed and not guaranteed.
  • You supply your own equipment: reliable internet, and closed-back headphones good enough to resolve speech in noisy recordings. A foot pedal or transcription hotkey setup is not required but is why experienced transcribers hit throughput targets.

Who does well

People who have transcribed professionally and can talk about a style guide they worked under, who are comfortable being audited on accuracy, and who flag ambiguity rather than smoothing it over. Prior AI or machine-learning experience is genuinely not expected — the linguistic judgment is the product.