Amberscript
Tool · this deep dive is built from the official pages of the platform itself and of the tools that support it.
Overview
Amberscript turns audio and video into text and makes subtitles. The pricing page splits the service in two, and that split runs through the whole review: the machine mode (machine-made) and the mode with human proofreading (human-made). Translated subtitles are named as a separate service. An API and enterprise solutions are declared.
The input languages are not published as a single list. The pricing page gives round numbers only: "90+ languages available" for the machine mode and "18+ languages available" for the mode with a human — the sets of languages differ by service, and the difference is stated directly. A page listing all ninety languages could not be found, so the fact is recorded as incomplete. The official sitemap does carry a separate page for each of the six languages of our catalogue.
Which platforms it works with
You take the result away as a file. The pricing page lists the formats: transcript — Word, JSON, TXT, CSV; subtitles — "SRT, VTT, EBU-STL and more". EBU-STL is a broadcast subtitle format. There is an API for embedding transcription into your own applications. Direct export to social networks or video platforms is not declared on the pages read: a "publish" button the service does not describe.
Pricing
Prices in euros. The machine mode and the human mode are counted differently, and the two must not be mixed.
Machine mode, monthly. Starter — 19 EUR for 5 hours (3.8 EUR per hour), Pro — 29 EUR for 10 hours (2.9 EUR per hour), Power — 49 EUR for 25 hours (1.96 EUR per hour).
Machine mode, annual payment. Starter — 190 EUR for 60 hours, Pro — 290 EUR for 120 hours, Power — 490 EUR for 300 hours; that works out at 3.16, 2.41 and 1.63 EUR per hour. A saving of 17% is declared.
Without a subscription. One-off credits — 10 EUR per hour, with volume discounts: from 5 hours minus 10%, from 20 hours minus 25%, from 50 hours minus 40%.
The human mode is counted per minute: transcription from 1.85 EUR, subtitles from 4.40 EUR per minute, translated subtitles on request.
The cheapest paid step is 10 EUR for an hour as a one-off credit; among subscriptions the minimum is 19 EUR a month.
What is free
On the pricing page there is no permanent free plan: only subscriptions and one-off credits. But the service's own machine-readable reference file llms.txt describes that same page with the words "starting with 10 free minutes" — that is, it declares ten free minutes at the start. On the pricing page itself no such line was found. One domain says two different things about itself; recorded as is, without picking the "correct" version.
Restrictions
The unit of account is hours, not minutes. Monthly subscriptions give 5, 10 or 25 hours a month, annual ones 60, 120 or 300 hours a year. Hours do not roll over from month to month. Purchased one-off credits expire after 60 days if they have not been activated. For the human mode turnaround times are declared: 5 working days in the normal mode and 1 working day in the rush mode.
Limits on the length or size of a single file are named neither on the pricing page nor in the terms — the service does not write about such a ceiling, which is not the same as promising there is none.
Rights to the output
There is no answer. The terms and conditions were read in full, and the question of rights to the uploaded file and to the resulting transcript or subtitles is not addressed in them. About commercial use of the result nothing is said either. An absent clause is neither permission nor prohibition.
Do they train on your data
No, and it is stated directly. The privacy policy says verbatim: "The audio and transcripts that are being generated, will never be used for training purposes of the language models that are used". The denial covers both the audio and the resulting transcripts.
There is no opt-out mechanism in the document, and none is required: training is denied by default for everyone — there is nothing to opt out of.
There is one reservation that matters for other people's speech. When human proofreading is ordered, the file and the draft transcript are visible to staff transcribers: "your audio or video file and the automatically generated transcript will be visible to our transcribers and to our operations team in order to perform corrections and quality checks". The access is described as "view only", downloading is closed to them. The machines do not learn from the recording, but people hear it.
How you earn with it
On commercial use of the result there is nothing to say: the terms and conditions do not address the question at all. Neither permission nor prohibition on selling transcripts and subtitles is in the document read. That is silence, not consent. What is known exactly is the cost of an hour of work: 10 EUR as a one-off credit, 1.96 EUR on the Power subscription.
What it will not do
Publish for you it will not: direct export to social networks and video platforms the service does not declare.
Let you upload someone else's recording without written permission it will not. The terms directly prohibit "Files that contain another person's personal information without that person's written consent" — files with another person's personal data without that person's written consent. Not verbal, but specifically written.
Take someone else's material it will not allow either: prohibited are "Files that the user uploads without the right to use those files (e.g. copyrighted material)", "Files that contain information that encourages conduct that may be deemed a criminal offense" and "Files that are technically harmful". A thematic list of forbidden subjects is not in what was read: the bans run through third-party rights, lawfulness and safety.
Require you to mark the transcript as machine-made it does not: neither watermarks, nor metadata, nor any duty to label the result is mentioned in the terms, in the policy or on the pricing page.
Answer whose the result is and whether you may sell it, it also cannot. Silence does not prove the capability is absent.
Verified data
The checked data this deep dive rests on.
About the platform
transcription of audio and video into text and creation of subtitles in two modes — machine-made and with human proofreading (human-made); translated subtitles as a separate service; there is an API and enterprise offerings
the division into machine-made and human-made is the main split of the pricing page, not a marketing heading
source, checked 2026-08-07
THERE IS NO EXACT LIST IN ONE PLACE — the pricing page gives only round numbers: «90+ languages available» for the machine mode and «18+ languages available» for the human mode, which means THE SET OF LANGUAGES CLEARLY DIFFERS BY SERVICE. The service does, however, have a separate page for each language, and all six languages of the catalogue appear in the sitemap: automatic-subtitles-in-russian, -in-english, -in-spanish, -in-chinese-mandarin, -in-hindi, -in-arabic
the figures are taken from the pricing page; the presence of the six languages is confirmed by the official sitemap at https://www.amberscript.com/en/page-sitemap1.xml and onwards — these are the service's own pages, not a review. No single page listing 90 languages could be found
source, checked 2026-08-07
Platforms
export formats from the pricing page: transcription — Word, JSON, TXT, CSV; subtitles — «SRT, VTT, EBU-STL and more» (verbatim). There is an API for embedding transcription into your own applications. No direct export to social networks or video platforms is claimed on the pages read
EBU-STL is a broadcast subtitle format, rare among online services
source, checked 2026-08-07
Pricing
Machine mode, monthly subscriptions: Starter — 19 EUR a month for 5 hours (3.8 EUR an hour), Pro — 29 EUR a month for 10 hours (2.9 EUR an hour), Power — 49 EUR a month for 25 hours (1.96 EUR an hour). Yearly: Starter — 190 EUR a year for 60 hours (3.16 EUR an hour), Pro — 290 EUR a year for 120 hours (2.41 EUR an hour), Power — 490 EUR a year for 300 hours (1.63 EUR an hour). One-off credits — 10 EUR an hour with volume discounts: from 5 hours minus 10%, from 20 hours minus 25%, from 50 hours minus 40%. Human mode: transcription from 1.85 EUR a minute, subtitles from 4.40 EUR a minute, translated subtitles — price on request
prices in euros; a saving of 17% is claimed for yearly billing
source, checked 2026-08-07
10 EUR for one hour as a one-off credit — the cheapest paid step without a subscription; among subscriptions the minimum is 19 EUR a month (Starter) or 190 EUR a year
source, checked 2026-08-07
Sources diverge: there is no permanent free plan on the pricing page. The service's own reference file for AI (llms.txt) describes the pricing page with the words «starting with 10 free minutes» — that is, 10 free minutes at the start are claimed; that line is not found on the pricing page itself
the sources disagree: the pricing page is silent about free minutes while llms.txt on the same domain names them. Recorded as it stands, without choosing the «right» version
source, checked 2026-08-07
Limits and restrictions
hours, not minutes: 5, 10 or 25 hours a month on monthly subscriptions and 60, 120 or 300 hours a year on yearly ones. Hours do not carry over from month to month. Purchased one-off credits expire after 60 days if not activated. For the human mode, turnaround times are stated: 5 working days in the standard mode and 1 working day in the rush mode. No limits on the length or size of a single file are named in the terms or on the pricing page
SILENCE OF THE SOURCE as to the length of a single file
source, checked 2026-08-07
Restrictions
it is expressly forbidden to upload: «Files that contain another person's personal information without that person's written consent», «Files that the user uploads without the right to use those files (e.g. copyrighted material)», «Files that contain information that encourages conduct that may be deemed a criminal offense», «Files that are technically harmful» (all verbatim)
the requirement of WRITTEN consent from the person whose personal data ended up in the recording is the strictest rule among those checked
source, checked 2026-08-07
Legal
NO, expressly denied in the privacy policy. Verbatim: «The audio and transcripts that are being generated, will never be used for training purposes of the language models that are used». It is separately stipulated that when human proofreading is ordered, the file and the draft transcript are visible to in-house transcribers: verbatim «your audio or video file and the automatically generated transcript will be visible to our transcribers and to our operations team in order to perform corrections and quality checks», with «view only» access and downloading closed to them
the most direct refusal to train among the services checked; the caveat about access by human transcribers matters no less when someone else's speech is involved
source, checked 2026-08-07
no separate opt-out mechanism is described and none is needed: training on uploaded recordings is denied by default for all users
formally there is no opt-out button in the document because there is no use for training either
source, checked 2026-08-07
Catalogue section: all similar See also: catalogue index · scheduler comparison · find by situation · platform restrictions