Free tool — audio file size

Audio file size calculator

Estimate a recording's file size from its format, bitrate, and length before you hit record — and see whether it fits cloud transcription limits like Whisper's 25 MB cap.

Recording parameters

60 min

Audio format

128 kbps
Voice (64k)Standard (128k)Music (320k)
Estimated file size
56.3 MB

60 min at 128 kbps (MP3/M4A)

Cloud API upload (25 MB cap)
over Whisper's 25 MB limit — you'd need to compress, split, or go local
Too large
Transcribing this recording manually
typing audio out runs 4–5× realtime, before proofreading
~4h 30m
Dictating with CoScript instead
text appears as you speak — there's no file to transcribe
0 extra hours

Size formula: duration × bitrate ÷ 8. Real files vary slightly with encoder overhead and metadata. Manual transcription estimate uses the industry 4–5× realtime rule for a trained typist.

Stop Recording, Uploading, and Splitting. Speak Once.

Cloud transcription APIs like OpenAI's Whisper enforce strict 25 MB upload limits. As the calculator shows, a standard 1-hour recording at 128 kbps is already over 56 MB — so the usual workflow becomes: record, download, compress with FFmpeg, split into chunks, upload each piece, and stitch the text back together. Then someone still has to proofread it.

The whole pipeline exists because the words were trapped in an audio file in the first place. CoScript removes the file: its speech engine runs locally on your Windows PC and turns your voice into clean, AI-formatted text while you speak, pasting it straight into any app.

  • No audio files to store, compress, or upload — nothing leaves your machine.
  • No 25 MB caps, per-minute API fees, or chunking scripts.
  • No 4–5× realtime transcription backlog — the text is done when you stop talking.

When you do need to estimate audio size

For podcasts, interviews, and archival recordings, the math above holds: file size = duration × bitrate ÷ 8. Voice recordings stay clear down to 64 kbps mono, which is the cheapest way to fit longer recordings under upload limits without re-encoding later.

CoScript for Windows

Stop typing. Start talking.

CoScript turns your voice into clean, AI-formatted text and pastes it straight into any Windows app at up to 200 words per minute — offline engine, 100+ languages, 30+ writing tones. Free to start.