Step 1 · Choose the audio or video
MP3, WAV, M4A, MP4, MOV…Drag the file here or look for it on your computer
Audio or video · up 100MB
How to transcribe audio to text
- Upload the file audio or video. MP3, WAV, M4A, MP4, MOV and other common formats are accepted.
- Choose the language or leave auto detection if you don't know.
- Wait for the transcript- You will see the progress and can cancel at any time.
- Review and download the text in TXT, or the subtitles in SRT/VTT with time stamps.
TXT, SRT and VTT: what each format is for
| Format | Content | Regular use |
|---|---|---|
| TXT | Plain text without tenses. | Minutes, notes, articles, summaries. |
| SRT | Text with beginning and end of each fragment. | Subtitles for video editors and YouTube. |
| VTT | Text with times in WebVTT format. | Subtitles for web and HTML5 players. |
What precision can I expect?
The quality depends on the recording: clean audio, without background music and with a A single voice is transcribed much better than a conversation with noise, several people speaking at the same time or very marked accents. The proper names, numbers, acronyms and Technical terms are the most frequent errors. Always check the result before to publish it or use it in an official document.
Privacy: your audio is uploaded to the server
Unlike image and PDF tools, transcription requires the file on the server to run the speech model. The audio is saved in a temporary folder, It is processed and erased upon completion or periodic cleaning. Not sent to platforms external nor is it used to train models. If the content is confidential, review the conditions before uploading it.
Limits and formats
Limit depends on server configuration: default 100 MB and 30 minutes per file. Transcription is done on the CPU, so a long audio may take several minutes. The most reliable formats are MP3, WAV, M4A, MP4 and MOV; if your file is not recognize, convert it to MP3 or WAV first.
Frequently asked questions
Can I transcribe a video?
Yes. The audio track is extracted from the video and transcribed the same way. The image nor a new video is generated.
Are multiple languages detected at the same time?
The model chooses a primary language. If there are two languages mixed, the detection automatic may fail; In that case, choose the predominant language or split the audio.
Are the timestamps accurate?
They are approximate and serve to guide the subtitles. In quick sentences or with Long silences can deviate by a few tenths. Adjust the SRT in your editor if you need perfect timing.
Can I cancel the transcription?
Yes. While processing you will see a "Cancel" button. Canceling deletes the file temporary and no results are downloaded.
Is the text saved on the server?
The result lives in memory for the time necessary for you to download it and is deleted with periodic cleaning. There are no accounts, history or permanent storage.
Do you only need the audio of a video? Also try compress video o trim a clip.