Can I transcribe a whole folder at once?
The upload takes one file at a time, and one file may be up to 30 minutes. A folder of short notes usually fits into a few such files.
Transcribe Khmer audio to text: our engine handles 99+ languages, AssemblyAI lists Khmer among 99 languages. How to deal with a folder of voice messages.
Our engine has 99+ languages, and Khmer is not one of them. AssemblyAI lists Khmer among the 99 languages it publishes (assemblyai.com/languages, read 22 October 2026). The Khmer voice notes go to them; the 99+ languages ones can come to us.
A folder of voice messages is a stack of small files: thirty seconds each, a name instead of a number, and no way to find the one with the address in it. Transcribing them means the engine has to know the language of the speaker. Ours knows 99+ languages. The rest of the folder needs a service that publishes Khmer among the languages it supports.
| What | Value |
|---|---|
| Khmer speech | on our list — our engine has 99+ languages |
| A service that lists Khmer | AssemblyAI lists 99 languages, Khmer among them |
| Length of one file | up to 30 minutes on the free plan, no minute limit on Pro |
Number them by the day they were sent. A stack of thirty files with the same name is a stack of thirty files you will never find again.
The Khmer ones go one way, the 99+ languages ones the other. One file holds up to 30 minutes, so a whole folder usually fits.
AssemblyAI publishes 99 languages, Khmer among them, on its own language page, read on 22 October 2026.
Drop the file on the widget above, leave the detection on, and every line comes back with the time it starts and ends.
Name the files before you upload, not after.
Keep the original voice notes; the text is derived from them.
Check the dates and the numbers in the first lines.
Export CSV when the notes have to go into a sheet.
The upload takes one file at a time, and one file may be up to 30 minutes. A folder of short notes usually fits into a few such files.
Detection works between 99+ languages; there is no Khmer in the list to detect.Athe recording language is detected automatically.
The free plan is $0, three files a day, up to 30 minutes per file, and no card. Pro is $10 a month when you pay yearly.
A Russian or mixed file stays with our own engine outside the United States. A pure English file may be sent to a cloud speech provider in the United States when that fallback is enabled.
Yes. The editor in the app takes the correction, and the export files are made from the corrected text, so a name can be fixed once and stays fixed.
Which languages our engine takes, with the code of each and whether the language is detected automatically.
Transcribe Lao audio to textThe same question for another language.
Transcribe English audio to textThe page about the language we do support.
Transcribe a phone call to textThe same idea for a single recording.
Speakers by name, a time on every line, export to TXT, SRT, VTT, CSV and Markdown.