Is noise a bigger problem than the language?
For us, the language comes first. The recording language is detected automatically.
Mongolian audio
Transcribe Mongolian audio to text
Transcribe Mongolian audio to text: our engine writes down 99+ languages, AssemblyAI lists Mongolian among 99 languages. What to do with a noisy site recording.
3 files a day up to 30 minutes, free99+ languagesSpeaker separation and timestamps
In short
Mongolian is not one of our 99+ languages, and a file in Mongolian will not be transcribed by our engine. AssemblyAI lists Mongolian among 99 languages (assemblyai.com/languages, read 22 October 2026). Give the Mongolian recording to them; the 99+ languages part can come to us.
The case
Wind, engines and 99+ languages
A recording made outside carries its own problems: wind, engines, people who speak over each other. The language question is separate from the noise question. The recording language is detected automatically.
What a site recording can become
| What | Value |
|---|---|
| Mongolian speech | on our list — our engine has 99+ languages |
| A service that lists Mongolian | AssemblyAI lists 99 languages, Mongolian among them |
| Speed | 0.84 s to transcribe a short file, 0.39 s to detect its language |
| Speakers | named separately, on request |
How to work with a recording made outside
- 1
Listen for the language, not the noise
The wind is a problem for any engine. The language is a problem for ours: find where the talk switches to Mongolian and mark that minute.
- 2
Hand the Mongolian recording over
AssemblyAI publishes 99 languages, Mongolian among them, read on 22 October 2026. The link sits in the sources below.
- 3
Upload the 99+ languages part
Drop it on the widget above: up to 30 minutes. Switch on the split by speaker when the voices are clear enough to tell apart.
- 4
Read the noisy part first
The opening minutes of a site recording are the worst, because the person is still walking. Check them against the tape before you quote anyone.
Tips
Record the first minute standing still, not walking.
Mark the minute where the language changes.
Quote the transcript against the tape, not from memory.
Keep the recording next to the export.
Questions this page answers
Can the speakers be named?
Yes, switch on the split by speaker before the upload. Where the voices overlap the names will be wrong too, and the first minute has to be checked by ear.
What does the paid plan give?
$10 a month when you pay yearly, or $20 a month when you pay monthly, with no minute limit of audio a month as fair use, the DOCX export, the REST API and the MCP server.
Is there an API for this?
Yes. A REST API, an MCP server with four tools, and plugins for Claude and ChatGPT. A file goes in with a language of ru, en or auto.
Do you keep the audio?
The transcript is kept in your account, so a word can be fixed and the file exported again. The audio is never used to train models for other customers and never sold.
Related pages
- Languages we can transcribeWhich languages our engine takes, with the code of each and whether the language is detected automatically.
- Transcribe Pashto audio to textThe same question for another language.
- Transcribe English audio to textThe page about the language we do support.
- How to transcribe a long recordingWhat to do with a file of up to 30 minutes on the free plan, no minute limit on Pro.
Sources
- AAssemblyAI — Supported Languages (99 languages, Mongolian among them)assemblyai.com · checked 22 October 2026
Reading: 6 minutesUpdated 
Transcribsi editorial team
Drop a recording — the text arrives in a minute
Speakers by name, a time on every line, export to TXT, SRT, VTT, CSV and Markdown.
3 files a day up to 30 minutes, free99+ languagesSpeaker separation and timestamps