A five-minute 48 kHz, 24-bit stereo recording contains 86.4 MB of PCM audio. The same duration at a measured average bitrate of 320 kbps comes out to about 12 MB before container tags. Those figures answer one question. The calculator below also works backwards: it'll show recording time from available storage and how many complete files fit on a drive.
Choose PCM when you need the deterministic sample payload, and WAV or AIFF when you want that payload plus an editable wrapper allowance. For MP3, AAC, Opus, or another encoded stream, use the measured average bitrate. FLAC stays a planning range because the signal itself changes the compression ratio.
Audio file size calculator
Calculate size, duration, or drive capacity
The math runs in this page. Your audio never leaves the device.
- Batch total
- 86.40 MB
- Per minute
- 17.28 MB
- Data rate
- 2,304 kbps
- Classic RIFF boundary
- about 4 h 8 min
PCM payload is deterministic. A container and its metadata sit outside this byte count.
300 s × 48,000 samples/s × 24 bits × 2 channels ÷ 8 = 86,400,000 bytes.
The certainty label is part of the answer. PCM payload is exact when duration resolves to complete sample frames. WAV and AIFF estimates carry separate, editable container overhead on top. Average-bitrate results come close to the real payload, and FLAC gives you a band rather than false precision.
The audio file size formula
For uncompressed PCM, use bytes = seconds × sample rate × bit depth × channels ÷ 8. A 60-second, 44.1 kHz, 16-bit stereo payload is 10,584,000 bytes. That is 10.584 MB or 10.094 MiB.
The channel count multiplies complete sample frames. Mono has one sample per frame, stereo has two, and a six-channel recording has six. Microsoft’s current RIFF overview describes WAVE audio as chunks that include format information and the audio data itself. The PCM equation calculates the data; it doesn't account for every possible surrounding chunk.
For encoded audio, use estimated bytes = seconds × measured average bitrate in bits/s ÷ 8. Sample rate and bit depth still matter upstream, but they don't set the size of a finished 128 kbps file on their own. Duration and the encoded average do that.
Real-file checks behind the defaults
I replayed the container check on August 15, 2026 with FFmpeg 8.1.2. The same 60-second, 48 kHz, 24-bit stereo silence contained 17,280,000 PCM bytes. FFmpeg wrote a 17,280,102-byte WAV and a 17,280,054-byte AIFF. The defaults therefore start at 102 and 54 wrapper bytes, but both fields remain editable because another writer can add metadata, padding, markers, artwork, or broadcast chunks.
| Test file | PCM or bitrate formula | Finished file | Observed difference |
|---|---|---|---|
| 60 s WAV, 48 kHz/24-bit stereo | 17,280,000 B | 17,280,102 B | +102 B |
| 60 s AIFF, 48 kHz/24-bit stereo | 17,280,000 B | 17,280,054 B | +54 B |
| 60 s MP3, CBR 128 kbps | 960,000 B | 960,515 B | +0.054% |
| 60 s MP3, CBR 320 kbps | 2,400,000 B | 2,401,220 B | +0.051% |
| 60 s AAC/M4A, target 256 kbps | 1,920,000 B | 1,665,647 B | −13.25% |
| 60 s FLAC pink noise, level 12 | 8,640,000 B PCM | 6,644,628 B | 76.91% of PCM |
The AAC result is the warning that matters. A target is not the finished average. Apple’s AAC rate-control note distinguishes CBR, ABR, constrained VBR, and VBR. ABR aims at a long-term bitrate; VBR controls quality without direct control of the average. If a file already exists, use its measured average bitrate from FFprobe, MediaInfo, or the editor’s properties.
Three calculations, three planning questions
Duration to file size
Use this before recording, rendering stems, or preparing delivery copies. Enter the longest file duration and the number of files. If a project mixes mono and stereo stems, calculate each group separately instead of pretending every file has the same channel count.
Storage to maximum recording time
Use this when the card, recorder, or upload limit is fixed. The calculator divides available bytes by the selected data rate. Its “three-copy working time” is a planning view for one source plus two independent copies, not a claim that three copies fit on one physical device.
Drive capacity to whole files
Use this for repeated interviews, podcast episodes, or delivery batches. The result rounds down to complete files and also shows three-copy sets. A partial final file is not counted.
MB, MiB, GB, and GiB are different units
The BIPM SI prefix table defines mega as 10⁶ and giga as 10⁹. That makes 1 MB equal to 1,000,000 bytes and 1 GB equal to 1,000,000,000 bytes. The NIST binary-prefix reference defines 1 MiB as 2²⁰ bytes and 1 GiB as 2³⁰ bytes.
One byte count can therefore appear as 1.05 MB and 1.00 MiB. The familiar 4.86% and 7.37% differences compare one MiB or GiB with one decimal MB or GB and use the decimal unit as the denominator: (1,048,576 − 1,000,000) ÷ 1,000,000 and (1,073,741,824 − 1,000,000,000) ÷ 1,000,000,000. Reversing the denominator produces different percentages. The calculator keeps the raw bytes underneath and displays both systems.
WAV, AIFF, RF64, and BW64: payload is not the container
WAV and AIFF can carry uncompressed PCM, but they do not share one universal 44-byte wrapper. Each container has its own chunk structure, and real applications can write additional metadata. That is why the calculator has separate WAV and AIFF choices and exposes the wrapper allowance instead of silently adding 44 bytes to every PCM result.
The familiar “4 GB WAV limit” comes from classic RIFF’s 32-bit size fields. In planning terms the boundary is about 4 GiB, or 4.29 decimal GB, with less available to the audio payload after headers. The current ITU-R BS.2088-2 recommendation specifies BW64 and its ds64 mechanism for files beyond 4 Gbytes. The EBU Tech 3306 landing page retains the related RF64 specification. For a long uninterrupted take, confirm that both the recorder and the receiving application support RF64, BW64, or another long-form container.
If the recording approaches that boundary, splitting at a safe edit point is often easier to exchange. The long-recording split guide covers the workflow. Do not lower sample rate or bit depth only to dodge a container limit without checking the delivery requirement. If conversion is required, review the sample-rate workflow and the bit-depth and dither guide before altering a master.
MP3, AAC, and Opus need an average-bitrate estimate
One hour at 128 kbps contains a nominal 57.6 MB of encoded payload. At 320 kbps it is 144 MB. Channel count should not be multiplied again when the selected bitrate already describes the complete stream, as it normally does in an export dialog.
CBR usually stays closest to the formula. ABR can miss the target, and a VBR quality setting may not expose a meaningful bitrate until after encoding. The safest workflow is to encode a representative passage, inspect its average, and use that measured value. The Sound Forge MP3 export guide explains the related export choices; the audio conversion guide covers format changes without overwriting the source.
Why FLAC must remain a range
FLAC reduces storage without discarding PCM information, but its result depends on the signal. Silence, steady tones, and correlated channels can compress strongly. Noise-like audio gives the predictor less redundancy to remove. Compression level changes encoder effort and can change size; it does not turn FLAC into a lossy codec.
The 35–85% default is an editable house planning band derived from a deliberately small local test, not a universal FLAC ratio. The 48 kHz/24-bit mono pink-noise file landed at 80.01% of PCM at level 0 and 76.91% at level 12. Digital silence landed at 0.19%. For a real library, encode a representative minute and replace the default band with the measured ratio.
The current FLAC specification in RFC 9639 supports PCM with one to eight channels and bit depths from 4 to 32 bits. The calculator disables larger FLAC choices, but a valid format ceiling does not guarantee that every player accepts every combination.
Plan the project footprint, not the delivery file
A 128 kbps MP3 can be the smallest object in the job. The source recording, edit master, rendered stems, alternate mixes, cache, and revisions usually dominate storage. Keep the preservation or edit master lossless, calculate delivery versions separately, and place independent backups on separate storage targets. For multichannel sessions, calculate the recorded channel count first; use the channel-conversion guide only when the delivery specification actually calls for a downmix.
The “three-copy sets” result is deliberately literal: one working source and two additional copies. It does not prove geographic separation, media health, or restoreability. Verify backups by opening or checksumming them, and test a restore before deleting cards or source drives.
Leave headroom on every destination. Filesystems, catalog databases, peak files, checksums, and application caches consume space outside the audio calculation. A recorder may also reserve space or stop before the advertised capacity is fully consumed. For field work, calculate with the card size you can actually write, then run a timed test at the exact channel count and recording format.
Five mistakes that break an audio storage estimate
- Calling every uncompressed result “WAV plus 44 bytes.” Calculate PCM first, then use the selected container’s actual or measured overhead.
- Using a target bitrate as if it were measured. ABR and VBR can finish above or below the setting.
- Multiplying total MP3 bitrate by channel count. The export bitrate usually describes the complete encoded stream.
- Treating FLAC as a fixed 50% saving. Measure representative material or retain a wide band.
- Budgeting one copy. Source, work files, delivery versions, revisions, and independent backups all consume space.
For a deadline-critical recording, calculate first and then make a short test with the exact recorder, application, codec, metadata, and destination filesystem. The calculator checks the order of magnitude. The finished test file proves what that workflow writes.
Audio file size calculator FAQ
How do I calculate an audio file size?
For PCM, multiply seconds by sample rate, bit depth, and channels, then divide by eight. For encoded audio, multiply seconds by the measured average bitrate in bits per second and divide by eight.
How many MB is one minute of audio?
At 44.1 kHz/16-bit stereo, one minute contains 10.584 MB of PCM payload. At 48 kHz/24-bit stereo it contains 17.28 MB. A one-minute 128 kbps stream contains about 0.96 MB before tags and container overhead.
How many GB is a three-hour audio file?
It depends on the format. Three hours of 48 kHz/24-bit stereo PCM contain 3.1104 GB. At 128 kbps, the encoded payload is about 172.8 MB; at 320 kbps it is about 432 MB.
How many GB is one hour of MP3?
At a measured average of 128 kbps, one hour is about 57.6 MB or 0.0576 GB before tags and container overhead. At 320 kbps, it is about 144 MB or 0.144 GB.
Why is my WAV larger than the PCM calculation?
WAV wraps the PCM data in RIFF chunks and can include metadata, padding, markers, artwork, or broadcast information. Enter measured wrapper overhead when an exact exported-file prediction matters.
Can I calculate FLAC size exactly before encoding?
No. FLAC size depends on the signal and encoder settings. Use a planning range, then encode a representative minute and measure its ratio to the PCM source.
Why do Finder and another tool show different file sizes?
They may be using decimal MB and GB versus binary MiB and GiB, or they may round differently. Compare the raw byte count before assuming either tool is wrong.
What happens when a WAV reaches the classic 4 GB boundary?
Classic RIFF/WAV has 32-bit size fields. Use a workflow that supports RF64 or BW64, or split the recording before the boundary, and verify compatibility at both ends.