AI Training Roles
Português Get alertsAlerts
Menu
Menu
Jobs1687 Platforms Fields Languages Guides Remote by countryInterview prepPay reportToolsGlossary Get alerts
Português

Transcription AI Training Interview Questions

Transcription roles ask you to turn speech into accurate text, often with timestamps, speaker labels and strict style rules, and sometimes to review or adjudicate other people's transcripts. The work rewards careful listening and following the guideline to the letter. These practice questions cover the decisions that come up most.

27 open roles in this groupUpdated 9 October 2026

Practice questions

These are practice questions written for this site to help you prepare. No platform has said it asks these exact questions.

  1. Transcribe a 30-second clip with two speakers who interrupt each other.

    A good answer covers: Correct speaker labels, overlapping speech marked the way the guideline asks, no guessed words, and consistent tags for unclear audio.

  2. What is the difference between verbatim and clean transcription?

    A good answer covers: Verbatim keeps fillers, false starts and repetitions; clean removes them while keeping meaning. Say you would follow whichever the project specifies.

  3. How do you mark a word you cannot hear clearly?

    A good answer covers: Use the project's unclear or inaudible tag instead of guessing, and add a timestamp if required.

  4. A speaker uses a brand name or a technical term you do not know. What do you do?

    A good answer covers: Look it up to get the spelling right; if you cannot confirm it, mark it as unclear rather than inventing a spelling.

  5. How do you handle numbers, dates and abbreviations?

    A good answer covers: Follow the style guide exactly (words or digits) and keep it consistent across the file.

  6. Two transcribers disagree on a segment. How would you decide as an adjudicator?

    A good answer covers: Listen again, check the guideline, pick the version that matches the audio and the rule, and record the reason.

  7. What equipment and setup help accuracy?

    A good answer covers: Good closed headphones, a quiet room, playback speed control and keyboard shortcuts.

  8. How do you transcribe a regional accent or dialect words?

    A good answer covers: Write what was said using standard spelling unless the guideline asks for dialect spelling; do not 'correct' the speaker's grammar in verbatim work.

  9. Why do timestamps matter for AI speech data?

    A good answer covers: They align text with audio so models learn which sound matches which word; wrong timestamps teach the wrong alignment.

What the platforms say about the assessment

In the 2026-10-09 data, roles in this group came from micro1, Appen, RWS TrainAI and Welo Data (Welocalize). Each platform runs its own process, and steps can differ by role. Below is what each platform's public pages or postings say, followed by lines quoted from current postings in this group.

micro1

Source: www.micro1.ai

Appen

Source: api.lever.co

RWS TrainAI

Source: api.lever.co

Welo Data (Welocalize)

Source: welodata.ai

Steps were read from public pages and postings, not tested first-hand. Passing a screening does not guarantee a project, hours or pay.

Terms to know

Transcription
Writing down exactly what is said in an audio or video recording, following the project's rules for spelling, fillers, noise and speaker changes.
Timestamping
Marking the start and end times of words, sentences or speakers in a recording.
Speech data
Recordings of people speaking, collected to train or test speech recognition and voice models. Projects usually set rules for setup, noise and script reading.
Guidelines
The project's written instructions that define how to do a task and how to judge answers. On most projects they take precedence over personal preference.
Quality check
Any review of your work by a reviewer, an automated test or a golden set. Many projects use quality scores to decide who keeps getting tasks.

Open roles in this group (27)

Sources

Other role groups