Transcribe audio over 25 MB without chunking it.
If your audio is over 25 MB, you have probably hit the documented 25 MB upload limit of the OpenAI Whisper API. Here a single file can be up to 5 GB and 10 hours, so a two-hour MP3 or an hour of WAV goes up whole. The first 20 minutes are free on files up to 2 GB.
Past 25 MB
Twenty-five megabytes is less than half an hour of MP3.
What 25 MB holds
At 128 kbps an MP3 uses about 58 MB per hour, so 25 MB is about 26 minutes. As CD-quality WAV it is barely two and a half minutes. Any meeting, lecture or interview of normal length is already over it.
No chunking pipeline
Getting a long file through a 25 MB limit usually means splitting it with ffmpeg, sending each chunk, and shifting every timestamp by its offset before joining the text. Here the whole file goes up once and comes back with one clock. The chunking script, its retry logic and its offset maths all stop being your code to maintain.
Big files, two routes
Files up to 500 MB fit the box on this page. Larger ones go up from the upload page, which accepts up to 5 GB. A YouTube, podcast, Drive, Vimeo, Twitch or Kick link needs no upload.
For developers
API access comes with the $24 pack of 5,000 minutes, which works out to $0.29 an hour of audio. Smaller packs are for the website and top up an account that already has API access.
Working around 25 MB
A 25 MB limit pushes the splitting work onto you.
✗ Chunk, send, re-join
A small upload limit means writing and maintaining code whose only job is to cut audio up and stitch text back together.
- ffmpeg loops to cut fixed-length chunks
- Words cut in half at chunk edges
- Offsets added to every timestamp
- Retries for each failed chunk
- Context lost between chunks
- Bitrate lowered to fit more per request
✓ One request per recording
The full file goes in as it is, and the output is a single transcript with timestamps from the start of the recording.
- Up to 5 GB per file
- No chunk boundaries in the text
- Timestamps relative to the whole file
- Two hours back in about four minutes
- A brief and chapters on top
- SRT and VTT ready for captions
Sample output
A 70 MB MP3 of a recorded lecture.
Illustrative lines, not taken from a real lecture. A 70 MB file, almost three times the 25 MB limit, comes back as one transcript without any chunk seams.
Export
One transcript. Five clean formats.
Every pack exports all five — no format is held back for a higher tier.
Plain text
De-ummed paragraphs. Ready to paste.
SRT captions
Timed to the speech. Every video editor reads this.
WebVTT
HTML5 player + YouTube uploads.
Show notes
Formatted with chapters and pull-quotes.
Machine-readable
Timestamped text for your own tools.
Pricing
Honest pricing, no surprises.
A pack from $6 (500 minutes) opens the full transcript. After that, no subscription and no auto-renew — every tier is a one-time minute pack, and credits never expire. Buy once, use the minutes whenever you need them.
Casual pack
$6/500 min
8 hours of audio — $0.72 an hour. One-time — credits never expire.
- 500 minutes of audio
- All export formats
- Unlimited AI chat per transcript
- Executive summary & key quote
Best value
$12/2,000 min
33 hours of audio — $0.36 an hour. Credits never expire.
- 2,000 minutes of audio
- Priority queue
- All export formats
- Unlimited AI chat per transcript
Bulk pack
$24/5,000 min
83 hours of audio — $0.29 an hour. Credits never expire.
- 5,000 minutes of audio
- Priority queue
- API access for batch pipelines
- All export formats
FAQ
Audio over 25 MB, answered.
Why is 25 MB a common limit?
The OpenAI Whisper API documents a 25 MB maximum per uploaded file, and many tools built on it inherit that cap. At typical bitrates that is under half an hour of audio. The usual advice is to split the audio or lower the bitrate, which works but moves the effort onto you.
What is the upload limit here?
One file up to 5 GB and up to 10 hours. Up to 2 GB a file starts with the first 20 minutes free; bigger files start once a pack is on the account. In practice that means a two-hour WAV, a four-hour MP4 or a full day of MP3 can each go up without any preparation.
How much does a 2-hour file cost?
120 minutes, covered by the $6 pack of 500 minutes with 380 left. Credits never expire. The first 20 minutes are free for a first-time visitor, so the pack only has to cover the remaining 100.
How quickly does it come back?
About 2 minutes per hour of audio, so a two-hour recording takes roughly four minutes once uploaded. A 70 MB lecture of just over an hour is usually done in a little over two minutes.
Is there an API for files over 25 MB?
Yes. API access comes with the $24 pack of 5,000 minutes, and you submit the recording whole rather than in chunks. The response gives you the transcript with timestamps measured from the start of the file, so there is no offset arithmetic in your code.
Which languages and export formats are supported, and is it private?
99 languages, detected without a setting, and exports in TXT, DOCX, SRT, VTT and JSON. Recordings are transcribed on our own servers and are never used for training.
Related
Bigger than 25 MB.
Over 25 MB? No chunking required.
Upload my audioOperated by Neugence Technology Pvt. Ltd. · contact@whipscribe.com · Security · Privacy · Terms
Comparing prices?
If cost and turnaround decide it: see our transcription service with prices and turnaround, the cheapest way to transcribe an hour of audio and how the best transcription services compare.