The MP3 file size formula
Bitrate tells you how many kilobits are used each second. Multiply that by duration to get total kilobits, divide by eight to convert bits to bytes, then divide by 1,000 for decimal megabytes:
For a three-minute song at 256 kbps: 256 × 180 ÷ 8 ÷ 1,000 = 5.76 MB. Operating systems sometimes display binary MiB instead, using 1,024 in the final division, so the number on disk may look a little smaller.
Common music sizes
| Bitrate | 3 minutes | 4 minutes | 5 minutes |
|---|---|---|---|
| 128 kbps | 2.88 MB | 3.84 MB | 4.80 MB |
| 192 kbps | 4.32 MB | 5.76 MB | 7.20 MB |
| 256 kbps | 5.76 MB | 7.68 MB | 9.60 MB |
| 320 kbps | 7.20 MB | 9.60 MB | 12.00 MB |
One-hour audio sizes
| Bitrate | Approximate size | Typical role |
|---|---|---|
| 64 kbps | 28.8 MB | Mono speech |
| 96 kbps | 43.2 MB | High-quality mono speech |
| 128 kbps | 57.6 MB | Stereo speech or compact music |
| 192 kbps | 86.4 MB | General music |
| 256 kbps | 115.2 MB | High-quality music |
| 320 kbps | 144 MB | Maximum standard MP3 bitrate |
Why the final file may differ
The formula is highly accurate for constant bitrate audio, but the complete file also includes headers, seek information and metadata. Embedded artwork can add anything from a few kilobytes to several megabytes.
Variable-bitrate files do not spend the same number of bits every second. Use the final average bitrate for a close estimate. The encoder’s target may be an approximate quality level rather than a guaranteed rate, so complex music can produce a larger file than simple speech.
Planning storage for a library
Multiply the average size per track by the track count, then leave room for artwork and growth. At three minutes per track, 1,000 songs are approximately 5.76 GB at 256 kbps or 7.2 GB at 320 kbps. Real albums are often longer, so using the actual total duration is more accurate than using a song count.
Fitting an upload limit
Reverse the formula to estimate the maximum bitrate:
Leave a margin of a few percent for metadata and artwork. If the result is too low for acceptable quality, trim the recording, use mono for genuine mono speech, or ask whether the platform accepts a more efficient codec.