Fix OpenAI 'Audio file is too large' Error on Mac

How to resolve the OpenAI Whisper 25MB file limit error. Learn how to compress, split, and transcribe large audio files on macOS.

Published on 2026-05-31

If you are using the OpenAI API to transcribe audio, you have likely encountered this error message:

400 Bad Request: "Audio file is too large. Maximum size is 25600000 bytes (25 MB)."

OpenAI's Whisper API imposes a strict 25MB size limit per request. If your meeting recording or interview is longer than 20 minutes (or uses high-bitrate audio), the file will exceed this limit.

This guide explains how to bypass this limitation on macOS.

Why the Error Happens

The OpenAI Whisper API is optimized for processing smaller audio snippets. When you attempt to upload a large audio file:

  • WAV files are uncompressed and will exceed 25MB in roughly 15 to 25 minutes.
  • MP3 / M4A files are compressed but will still exceed 25MB for meetings longer than an hour.
  • The API returns a 400 Bad Request and refuses to process the file.

How to Fix It Manually

To resolve this issue manually, you must compress or split the file:

1. Compress the Audio (Using Terminal)

You can use ffmpeg to compress the audio bitrate. Open Terminal and run:

ffmpeg -i input.wav -codec:a libmp3lame -qscale:a 5 output.mp3

This reduces the file size by converting it to a compressed MP3.

2. Split the Audio Into Chunks

If the compressed file is still over 25MB, you must slice it. Run this ffmpeg command to split the file into 10-minute segments:

ffmpeg -i input.mp3 -f segment -segment_time 600 -c copy parts_%03d.mp3

You must then send each parts_000.mp3, parts_001.mp3 file to the OpenAI API individually and merge the resulting text files manually.

The Automated Solution: justREC

justREC automates this entire compression and splitting pipeline natively on macOS:

  1. Auto-Chunking: When you import an audio file or finish recording a call, justREC automatically checks the file size. If it exceeds the optimal size, it splits the file into segments behind the scenes.
  2. Sequential Uploads: The app uploads the segments one by one to OpenAI using your own API key.
  3. Autosave Manifest: If the upload fails due to network instability, justREC saves your progress locally. You can resume processing the remaining segments without starting over.
  4. Seamless Reconstruction: Once the transcription is complete, justREC merges the text segments, runs your summary templates, and exports the final file (Markdown, DOCX, or PDF) to your workspace.

How to Get Started

Save time managing terminal commands and manual splits. Download justREC, connect your OpenAI API key, and handle long audio transcriptions automatically in a clean macOS app interface.