The recording goes in
Audio or video, Russian or English, one file per job. The engine listens to the whole recording before it writes a line.
Russian and English · our own engine
Put the recording in.
Get the text out.
Transcribe Russian and English recordings with a timestamp on every segment and speaker labels on request.
Drop an audio or video file
or — mp3, wav, m4a, mp4, up to 64 MB and two hours
Transcription runs under your account: sign in or create a free account.
How it works
Three steps, and the text is yours.
- 1 · Upload the file
- Drop it on this page or into the app. Audio or video, up to 64 MB and two hours per file. The same endpoint takes it from your own code, so a pipeline does not need a browser.
- 2 · Wait for the segments
- The engine returns timed segments and, with diarization on, names the speakers. On a short file our own measurement was 0.84 s for the transcription and 0.39 s for language detection; long recordings take longer.
- 3 · Read, fix, export
- Correct a name in the editor, save it, and export to TXT, Markdown, SRT, VTT, CSV or DOCX. The API reads the same transcript back, line by line, with the timecodes intact.
00:04Мы перенесли релиз на четверг, потому что QA ещё закрывает два бага.
00:11Speaker 1: Okay, but the API rate limit is still 60 requests per minute.
00:18Понял. Я подниму лимит до 300 и напишу в Slack.
The shape of the result — your file comes back in the same form.
Guides
Answers to the questions people ask us.
Each guide is a full answer — with the numbers, the limits and a file you can try on the spot. The Russian versions are marked as such.
What comes back, what it costs, and the limits of one file.
Transcribe Russian audio to text
Our own engine, mixed Russian-English speech, and where your file goes.
A checklist, speakers, timecodes and the numbers behind the plan.
The 64 MB and 120-minute limits, and how to split around them.
We take files, not links: here is the three-step way.
Short recordings, what they cost and how to get them out of your phone.
Тот же гайд на русском.
Чек-лист на русском.
An honest comparison, including where we are worse.
A meeting assistant against a transcription service.
Compliance, hardware and prices.
Inside the engine
The words come out with their timecodes
Every segment knows where it starts and where it ends. Subtitles line up with the audio without a manual pass.
The speakers are separated
Switch diarization on and the people in the room are told apart, so a two-hour interview stays readable.
The text leaves in your format
TXT, Markdown, SRT, VTT, CSV or DOCX. A download in the browser, or the same bytes through the API.
What to hold us to
Facts, not adjectives.
- Engine
- Our own engine, with a model profile tuned for Russian speech — not a general model with a language switch.
- Timestamps
- Every segment carries its start and end time. Speaker labels are a switch, not an add-on.
- Export formats
- TXT, Markdown, SRT, VTT and CSV on every plan; DOCX on Pro. All of them come out of the same transcript.
- Free plan
- No card and no trial clock. Three files a day, up to 30 minutes each.
About privacy. Your file is used to produce the transcript and to run the service. It is not sold, and it is not used to train models for other customers. Delete a transcript in the app and the audio goes with it. Your audio is not sent to third parties — with one exception: for pure English, when a cloud fallback is configured on the server, that file goes to the cloud provider instead of our engine.
About accuracy. We do not publish a single accuracy number here. A score without the test set behind it means nothing. Ask for the evaluation report and the sample set, and we will send both.
Pricing
Free to start, cheap to keep.
| Plan | Price | What it gives |
|---|---|---|
| Free | $0 | 3 files a day, up to 30 minutes each. No card, no trial clock. |
| Pro | $10/month | Billed yearly at $120, or $20 a month if you pay monthly. Fair use of 1,500 minutes of audio a month, DOCX export, API keys and MCP access. |
| Team | On request | Shared workspaces, one glossary, SSO and a written volume allowance. |
| API | By volume | Credits, no subscription. Keys live in the app. |
The full plans, fair-use limits and what happens at the limit: pricing.
Comparison
Against a typical online transcription service.
The right-hand column describes the usual behaviour of general-purpose speech tools, not a review of any single product.
| What matters | Transcrib Si | Typical online service |
|---|---|---|
| Russian speech | Yes — own engine, model profile tuned for Russian | General model with a language switch |
| Russian mixed with English | Yes — both languages stay as they were spoken | Often transliterated or dropped |
| Timestamps | Yes — every segment | Usually available |
| Speaker labels | Yes — included, on request | Often a paid add-on |
| Export formats | TXT, MD, SRT, VTT, CSV · DOCX on Pro | Usually TXT and SRT |
| Price for one hour of audio | Inside the Pro fair-use allowance | Billed per minute or per hour, on top of a subscription |
| API and MCP access | Included on Pro and Team | Metered, billed separately |
FAQ
Questions people ask first.
- Which languages does it handle?
- Russian and English, including speech that switches between them inside one sentence. The upload screen can detect the language itself; on our own measurement the detection took 0.39 s on a short file.
- How much does it cost?
- Free is $0 with 3 files a day, up to 30 minutes each. Pro is $10 a month when you pay yearly ($120 a year) or $20 a month when you pay monthly. Team is priced on request, and API usage is billed by volume with no subscription.
- Is there a free plan?
- Yes. Three files a day, up to 30 minutes each, with timestamps, speaker labels and exports. No card, no trial clock, and the limit is per day rather than per month.
- How accurate is it?
- We do not publish a single accuracy number on this page. Our engine runs a model profile tuned for Russian speech, and we will send the evaluation report and the sample set on request — a score without its test set means nothing.
- What happens to my audio?
- It is used to produce the transcript and to run the service. We do not sell it and we do not train models on it for other customers. Delete a transcript and the audio goes with it. Your file is not passed to third parties, with one exception: for pure English, when a cloud fallback is configured on the server, the file goes to that cloud provider instead of our engine.
- Can I transcribe a YouTube video from a link?
- No. The service takes files, not links. Download the audio yourself and upload the file — it accepts audio and video containers up to 64 MB and two hours.
- Is there an API?
- Yes. Create a key in the app and call the REST API with it: upload, list transcripts, read a transcript with its timecodes and speakers, export in any of the six formats. Pro and Team accounts can also connect the MCP server to editors and assistants.
- How long can a file be?
- Up to 64 MB and two hours per file. Longer recordings are split into parts and sent one after another.
Try it on a recording from this week.
Drop a file on the widget at the top of this page, or open the app for your whole history, API keys and billing.