How to transcribe a large audio or video file (2 GB, 3 hours) without splitting it

The short answer

• One file, up to 5 GB and up to 10 hours. Inside those limits, upload the recording as it is. No compressing, no converting, no splitting.
• Video is fine. Upload the MP4 or MOV and the audio is taken from it. The transcript is the same as if you had uploaded audio only.
• Over 5 GB? Make an audio-only copy. It is usually a few percent of the video's size and keeps the same timeline.
• Price depends on minutes, not gigabytes. $4 buys 500 minutes that never expire, so a three-hour recording uses 180 of them.

Step 1: check the size, the length and the format

Before you upload, look at three things in your file manager (on a Mac, select the file and press ⌘I; on Windows, right-click and choose Properties, then the Details tab):

• Size. Under 5 GB goes straight up. Over 5 GB needs the audio copy in step 3.
• Length. Under 10 hours is one upload. A full conference day that runs longer is better split at a natural break, such as lunch.
• Format. MP3, WAV, MP4, M4A, FLAC, OGG, WEBM and MOV all upload directly. If you recorded in OBS and have an .mkv, use OBS's File → Remux Recordings to turn it into an MP4 first; it takes seconds and does not re-encode anything.

Why the same two hours can be 100 MB or 10 GB

The words in the file are identical in every row. What changes is the picture and the audio quality setting, and neither changes how much the transcript costs.

Step 2: decide between the video and an audio copy

If the file is under 5 GB, the only reason to make an audio copy is your upload speed. Uploading is the slow part of a large file; transcription is quick once the file arrives. As a guide, 2 GB on a 20 Mbps upload takes about 13 minutes, while the 120 MB audio from the same recording takes under a minute. On a fast connection, just send the video.

Step 3: make an audio-only copy (only if you need one)

An audio copy keeps every second of the timeline, so a timestamp in the transcript still points to the same moment in your original video. With the open-source ffmpeg tool, this command copies the audio track out of an MP4 or MOV without re-encoding it:

ffmpeg -i seminar.mp4 -vn -c:a copy seminar.m4a

It finishes in seconds even for a multi-gigabyte file, because nothing is being converted. If your editor has an "export audio only" option, that works too; pick M4A or MP3 rather than WAV to keep it small.

Step 4: rejoin camera chapter files

Many cameras stop a file at around 4 GB and start a new one without a break in the recording, so one two-hour take arrives as C0001.MP4, C0002.MP4 and C0003.MP4. Upload them as three files and you get three transcripts, each starting at 0:00. Join them first and you get one. List the parts in a text file, in order:

file 'C0001.MP4'
file 'C0002.MP4'
file 'C0003.MP4'

Then join them and keep only the audio, which keeps the result well under 5 GB:

ffmpeg -f concat -safe 0 -i parts.txt -vn -c:a copy seminar-full.m4a

Step 5: upload and leave it running

1. Open whipscribe.com on your computer and drop the file onto the upload box. The language is detected automatically, and 99+ languages are supported.
2. When asked after the upload, enter your email, so the transcript is tied to you and you get a link when it is ready.
3. Keep the tab open until the upload bar finishes. After that the work happens on our servers, at about 2 minutes per hour of audio, and you can close the tab.

A worked example

What you get back

• One continuous transcript with clickable timestamps from 0:00 to the end. Click any line and the audio plays from there, which matters in a three-hour file. See how to find any moment in a long recording.
• The Intelligence tab with a summary, chapters and key quotes, so you can see the shape of a long recording before you read it.
• AI Chat that answers questions about the recording, with timestamps you can play, and can turn what you ask for into Word, PDF, Excel or SRT files saved under Resources.
• Exports in TXT, SRT, VTT and DOCX.

Recordings are transcribed on our own servers and are never used for training.

Recording tips for the next long session

• Record a separate audio track. A small recorder or a lapel mic near the speaker gives a clearer file than the camera's built-in mic across the room, and it is a few hundred MB instead of several GB.
• Do not pause and restart. One continuous take gives you one timeline. If you must stop, name the files in order so they are easy to join.
• Choose MP3 or M4A on audio recorders for long sessions. WAV is fine for quality but grows by about 690 MB an hour.

What it costs

Packs start at $4 for 500 minutes, and the minutes never expire. Larger packs are $8 for 1,000 minutes, $12 for 2,000 and $24 for 5,000; payment is by card. A 5 GB file and a 50 MB file of the same length cost the same. Details are on the pricing page. For recordings of seven hours and more, see how to transcribe a 7-hour podcast recording.

Frequently asked