Research teams
Expert calls, in-depth interviews and focus groups, with speaker labels and timestamps set the way your analysis needs them.
AI speed · Human accuracy · Confidential by default
Transcripts you can quote
without checking twice.
Automatic transcripts slip on accents, numbers, names and speakers who switch language. For research teams, law firms and health researchers, a trained language specialist checks every line against your recording. Where you allow it, secure AI makes the first draft. Prefer no AI at all? Choose our human-only service.
- Every line checked against the recording by a trained language specialist
- English as people really speak it, plus Swahili and mixed-language audio with translation into English
- Deadlines in your time zone, with next-day English transcription
- A free sample of up to 10 minutes, back within one working day
Services
Delivered in your template: your style guide, speaker labels, timestamp intervals and file naming, followed as specified.
Interview transcription
Expert calls, in-depth interviews and focus groups with accurate speaker attribution. Clean or strict verbatim, with a human-only option.
Transcript check
Already have an automatic transcript from Zoom, Teams or an app? A language specialist checks it line by line against the audio and returns it corrected, with a short report of what changed.
Subtitles and captions
SRT and VTT files checked for timing, reading speed and line length, ready for your edit, platform or event. Captions for deaf and hard-of-hearing viewers include sound cues and speaker identification.
Translation into English
Swahili and mixed-language audio rendered into English that keeps what the speaker meant.
Who it's for
Teams whose recordings carry people's words, names and decisions, and who need them right the first time.
Law firms
Hearings, interviews and statements, transcribed as spoken, with names and terms confirmed against your glossary.
Health studies and evaluations
Interviews and focus groups from health research and programme evaluations, in English, Swahili or both in one sentence.
Producers and media
Podcasts, documentaries, broadcasts and online video, with subtitles timed to speech and checked for reading speed.
Churches and ministries
Sermons, teaching and radio programmes, made searchable with transcripts and shared more widely with subtitles.
Businesses and events
Meetings, panels and conferences, turned into accurate written records and captioned video.
Subtitles, timed to the frame
Every cue is checked for timing, line length and reading speed, then delivered as SRT or VTT, ready for your edit, platform or event.
interview_EN.srt4 cues
- 42 characters per line
- 2 lines maximum
- 17 characters per second or less
- Timed to speech
- SRT and VTT
Swahili, English and everything in between
Many recordings switch between Swahili and English in the same sentence. We transcribe them exactly as spoken, then translate into English that keeps what the speaker meant.
As spoken[00:14:32] Respondent: Wateja wengi walisema the price is too high, lakini bado walinunua kila wiki.
With translation into English[00:14:32] Respondent: Wateja wengi walisema the price is too high, lakini bado walinunua kila wiki. [Many customers said the price was too high, but they still bought every week.]
- Every word in the language used. Swahili stays Swahili and English stays English. Sheng is kept as spoken, never “corrected”.
- Translation that keeps the meaning, placed after each passage or delivered as a separate English transcript, as your project needs.
- Your glossary first. Names, places and terms are spelled the way you specify, consistently across every file.
- Another language in the recording? Tell us in your quote request. Anything outside the agreed languages is tagged with its timestamp, never guessed.
An illustrative example, not client material.
Choose your transcript style
Both styles keep exactly what the speaker meant. They differ in how much of the speech around it you see.
Default
Clean verbatim
Fillers, stutters and false starts removed; the speaker's words and grammar kept. Best for analysis, reports and quotes.
[00:02:10] Respondent: We tested the new dosing schedule in three clinics last spring.Quoted separately
Strict (full) verbatim
Every word and audible sound, including ums, repeats and false starts. Best for legal records and conversation analysis.
[00:02:10] Respondent: So, um, we-we tested the new, uh, dosing schedule in, like, three clinics [laughs] last spring.
Both include
- Speaker labels
- Timestamps at the interval you set
- [inaudible] and [unclear] tags, each with its own timestamp
- Your template and file naming
- Word (.docx) unless you ask for another format
How it works
Three steps, with the checking built in.
You request a quote, then send the audioOnce we have agreed terms, you upload files through a secure, expiring link, with a glossary of the names, terms and acronyms you want spelled a particular way.
We draft, then verify against the recordingThe draft is fast; the checking is the work. Every uncertain passage is re-heard at its timestamp, and anything truly inaudible is marked, never guessed.
You receive the finished file on the agreed dateIn your format, with a short note on anything ambiguous. Corrections within seven days are included.
Send files any time
Built for every time zone
Send files at any hour, any day of the week, from anywhere. Every deadline is set in your time zone, not ours, so a short file sent at the end of your day can be waiting when you start the next.
- Los Angeles–Uploads open
- New York–Uploads open
- London–Uploads open
- Berlin–Uploads open
- Dubai–Uploads open
Confidential by default
Research recordings contain people's words and identities. They are handled accordingly.
Agreements before files
Mutual NDA on request, and data processing agreements under the EU and UK GDPR, with Standard Contractual Clauses. Clients elsewhere can ask for terms under their own data protection law. Every language specialist works under a signed confidentiality and data protection agreement.
Subprocessors named privately
Our processing providers and editing team are listed in the data processing agreement, not on a public website.
No training without your permission
Your recordings and transcripts are never used to train AI models without your written permission, and we use only providers that do not train on customer files.
Deleted on a schedule
Source media and working files are deleted 30 days after delivery, sooner on request, with written confirmation. Copies in our providers' backups expire within 90 days.
Pricing
Per minute of audio. A firm quote follows a quick listen, since recording quality and the number of speakers change the work more than length does.
| Service | Turnaround | Per minute |
|---|---|---|
| English transcription | 3–5 working days | from $1.25 |
| English transcription | Next working day | from $1.75 |
| Human-only English, no AI at any stage | 3–5 working days | from $2.25 |
| Swahili or mixed-language transcript, in the language spoken | 3–5 working days | from $2.50 |
| English transcript from Swahili or mixed-language audio | 5–7 working days | from $4.50 |
| Transcript check: your automatic transcript, corrected | 3–5 working days | from $1.00 |
| Subtitles and document translation | By project | Quoted |
Four or more speakers, such as focus groups: add $0.50 a minute. Minimum order $40. Strict verbatim and difficult audio are quoted separately. New clients can send up to 10 minutes of audio for a free sample. Regular or high-volume work? Ask us about volume pricing. Invoices in US dollars, or Kenya shillings for clients in Kenya. US clients receive our Form W-8BEN-E at onboarding, so no Form 1099 or US withholding applies to our fees.
Questions
The things clients ask before their first project. Anything else? Write to contact@cuelinemedia.com.
Do you use AI?
Only where you allow it. A first draft may be made with secure automated speech recognition, and a trained language specialist then checks and corrects every line against the recording. Prefer no AI at all? Our human-only service is typed from scratch by our editors, with no AI or automated tools at any stage.
Will our recordings be used to train AI?
No. Your recordings and transcripts are never used to train AI models without your written permission, and we use only providers that do not train on customer files.
How do you make sure the transcript is right?
Every line is heard against the recording, not just read. Names and technical terms are checked against your glossary or an authoritative source, numbers and dates are checked twice, every speaker change is re-heard, and each file has a final review before delivery. Anything truly inaudible is tagged with its timestamp, never guessed.
How do we send recordings?
Once we have agreed terms, we send you a secure upload link that expires. Please don't email recordings or attach them to the quote form.
How long does it take?
English transcription takes 3 to 5 working days, or the next working day if you need it faster. Swahili and mixed-language work takes 3 to 5 working days. Every deadline is set in your time zone, and translation and subtitles are scheduled by project.
What do you deliver?
Transcripts in your template and file naming, in Word (.docx) unless you ask for another format. Subtitles as SRT or VTT files, and captions for deaf and hard-of-hearing viewers with sound cues and speaker identification. Corrections within seven days are included.
Can you sign our NDA or data processing agreement?
Yes. We sign a mutual NDA on request, and data processing agreements under the EU and UK GDPR, with Standard Contractual Clauses. Clients elsewhere can ask for terms under their own data protection law.
Our recordings include sensitive information. Can you help?
Tell us in your quote request if your recordings include health or other sensitive personal information, US patient information or speakers under 18, and about any rules on where your data may be stored. We agree the right terms before any file is sent.
Can you fix a transcript we already have?
Yes. Send the recording with the transcript. A language specialist checks every line against the audio and returns it corrected, with a short report of what changed, from $1.00 a minute. If the transcript is too poor to correct, we tell you and quote for a new one.
Can we try you before committing?
Yes. New clients can send up to 10 minutes of audio for a free sample, returned within one working day. Send your hardest recording: the noisiest, most accented or most mixed-language one you have.
Do you offer volume pricing?
Yes. Regular and high-volume work qualifies for volume pricing. Tell us roughly how many hours of audio a month you expect when you request a quote.
How do invoices work?
Invoices are in US dollars, or Kenya shillings for clients in Kenya. US clients receive our Form W-8BEN-E at onboarding, so no Form 1099 or US withholding applies to our fees.
Start with a quote
Tell us what you need: the type and length of your recordings, the languages and your deadline. We reply by email with a firm quote and, once terms are agreed, a secure upload link. Please don't send recordings before then.
Or write directly to contact@cuelinemedia.com