Verbatim, clean verbatim and timestamps: a working guide for transcribers
· a.pannella87
Last updated: 27 October 2026 · OpenJob editorial team
Verbatim transcription keeps everything the speaker says, including fillers, false starts and repetitions. Clean verbatim removes them and keeps the completed phrase. Speech-data projects for AI often want something close to full verbatim; projects producing readable text lean towards clean verbatim. The project’s own style guide always overrides general rules.
Key points
- Most points in transcription tests are lost on style, not hearing.
- Fillers and repetitions: kept in verbatim, removed in clean verbatim.
- Timestamp rules vary by project: per speaker change, fixed interval or unclear audio only.
- Set a timestamp where the sound starts, not where you started typing.
Ask anyone who reviews transcription tests where candidates lose points and you will rarely hear “they misheard a word”. The usual problem is that the transcript follows the wrong style. Before you type anything, find out which one the project wants.
What each style keeps
| Feature | Verbatim | Clean verbatim |
|---|---|---|
| Fillers (um, uh, you know) | Kept | Removed |
| False starts (“I went — I drove”) | Kept, marked | Removed, keep the completed phrase |
| Repetitions (“the the”) | Kept | Removed |
| Grammar errors by the speaker | Kept | Usually kept; check the guide |
| Non-speech sounds | Tagged as specified | Often tagged only if meaningful |
Speech-data projects for AI often want something close to full verbatim, because the model needs to learn how people actually talk. Projects producing readable text for humans lean towards clean verbatim. Some projects define their own canonical style; when they do, their guide overrides anything you learned elsewhere.
A short exercise
Record yourself explaining your morning routine for thirty seconds without a script. Transcribe it twice, once in each style. The differences will show you exactly where your instincts lean, which is useful to know before a timed test.
Timestamps without guesswork
Timestamp rules vary more than style rules. Some projects want a mark at every speaker change, some at fixed intervals, some only around unclear audio. Three habits help whatever the rule:
- Set the stamp where the sound starts, not where you started typing.
- Replay the boundary at reduced speed instead of estimating.
- When two people overlap, follow the project’s overlap rule literally, even if it feels clumsy.
Where this applies right now
Several open roles on OpenJob are transcription-first, including the Dutch Transcription Expert and Swiss German Transcription Expert listings. Each page shows the published rate and the language requirement; the style guide itself is given by the platform after you are accepted.
Open roles that do not ask for previous experience
Refreshed automatically from our verified catalogue. Check eligible countries and requirements on each listing.
Corporate Attorney (BigLaw Firms)
Remote contract work for corporate attorneys from large law firms on an AI training project: analyze and draft complex agreements and legal instruments in corporate law, M&A, private equity and cap…
Finance Specialist
Remote contract for finance professionals on an executive-presentation project: rebuild and improve PowerPoint slides that present complex financial data and insights for investors and executives…
Bilingual Japanese Psychologist (PhD)
Remote contract for bilingual psychologists: assess psychological case studies and documentation and develop culturally sensitive scenarios and materials in Japanese and English for AI training.
Simulation Expert — Real-to-Sim 3D Environments & Assets
Remote contract work for simulation experts on an AI training project: build 3D assets and environments in Isaac Sim, Gazebo, MuJoCo or similar, and tune physical parameters so they match real-worl…
Japanese Language Expert
What if your mastery of Japanese could directly influence how AI understands, speaks, and connects with millions of people?
Transcription roles · Audio and speech roles
Frequently asked questions
Is clean verbatim the same as editing?
No. Clean verbatim removes fillers, false starts and repetitions, but keeps the speaker’s words and usually their grammar. Editing would rephrase or correct meaning, which transcription projects do not ask for. When unsure, follow the examples in the project guide.
Which style do AI transcription projects use?
Speech-data projects often want close to full verbatim, because models need to learn how people actually talk, including hesitations. Some define their own house style. The style is stated in the project guide you receive after acceptance, and it overrides anything learned elsewhere.
How do I practise for a transcription test?
Record yourself speaking for thirty seconds without a script and transcribe it twice, once verbatim and once clean verbatim. Then use public-domain audio with two speakers. Compare your file against a style guide line by line, since style errors cost more points than mishearing.