![\"One](\"/assets/img/articles/stem-storage.svg\")
A three-minute stereo mix and twelve three-minute stereo stems have very different storage requirements even when every export uses the same settings. Count the duration, channels and files separately before deciding how much space the handover needs.
\nThe Audio File Size Calculator estimates uncompressed sample data in bytes, decimal MB and binary MiB. It does not inspect recordings or predict compression. The calculation is useful for planning an export; the delivered files remain the final evidence of their actual size.
\nFor byte-aligned uncompressed audio:
\nbytes = sample frames × channels × stored bits per sample / 8
A sample frame holds one sample for each channel at a time position. At 48,000 Hz, a one-second stereo file has 48,000 frames containing 96,000 individual samples. Packed 24-bit storage uses three bytes per sample, giving 48,000 × 2 × 3 = 288,000 bytes per second.
\nFor a three-minute file, multiply by 180 seconds. The result is 51,840,000 bytes, or 51.84 MB of audio data. Microsoft’s waveform format documentation defines the corresponding relationships for byte rate and block alignment.
\nThe calculator rounds duration × sample rate to a whole number of frames before multiplying. That detail matters for a duration with many decimal places: a file cannot hold half a frame. It also reports the resulting frame-aligned duration so you can see exactly which duration the byte count represents.
\nTwelve stereo stems mean two channels per file and twelve files. They do not mean twenty-four channels per file as well as twelve files. That second interpretation counts stereo twice.
\n| Export at 48 kHz, packed 24-bit, 180 seconds | Files | Channels each | Combined audio payload |
|---|---|---|---|
| One mono track | 1 | 1 | 25.92 MB |
| One stereo mix | 1 | 2 | 51.84 MB |
| Twelve mono tracks | 12 | 1 | 311.04 MB |
| Twelve stereo stems | 12 | 2 | 622.08 MB |
Two mono files with matching duration contain the same sample payload as one interleaved stereo file. Their combined actual size differs slightly because there are two containers instead of one. Whether the recipient prefers split mono or interleaved stereo is a delivery question, not a way to remove half the audio.
\nCount a separate reference mix as an additional file. If some stems are mono and others stereo, calculate the groups separately. If the files have unequal lengths, add their actual durations within each shared format rather than multiplying the longest duration by every file unless that is the export you intend.
\n“24-bit audio” does not always mean three stored bytes per sample. A format can hold 24 valid bits inside a 32-bit container, which occupies four bytes. The calculator’s packed 24-bit option assumes three bytes; use 32 when the stored width is four.
\nMicrosoft’s extensible waveform format documentation distinguishes container size from valid sample bits. Inspect the export settings or resulting file properties when this distinction matters to the estimate.
\nA 32-bit float file and a 32-bit integer file both occupy four bytes per stored sample. That equality of size does not make their numeric representations equivalent. Choose the format for the processing or delivery task, then calculate its storage. The sample-rate and bit-depth guide separates those decisions.
\nOne MB in this calculation is 1,000,000 bytes. One MiB is 1,048,576 bytes. The twelve-stem payload of 622,080,000 bytes is therefore 622.08 MB or about 593.262 MiB.
\nBoth numbers refer to the same amount of data. A different displayed unit does not indicate that samples went missing during export. Save the byte count when comparing operating-system displays, transfer tools and storage budgets that may use different labels.
\nThe same distinction applies at larger scales. A decimal GB is 1,000,000,000 bytes, while a GiB is 1,073,741,824. Avoid rounding a payload down to a convenient-looking whole GB when deciding whether a recording will fit. Containers and the surrounding project require additional space.
\nA WAV or AIFF file contains more than sample data. Format headers, cue points, Broadcast Wave metadata and artwork can add bytes. Some WAV files also use compressed encodings, so the filename extension alone does not establish that this formula applies.
\nA DAW project folder can be much larger than its final stems. Alternate takes, consolidated copies, rendered effects, peak caches and retained edits may remain in the folder. The calculator knows none of those quantities. Estimate the export and inspect the project separately rather than treating the result as a disk-space guarantee.
\nFor a compressed delivery such as MP3 or AAC, use the encoded bit rate and duration as a different starting point. Lossless compression such as FLAC depends on the signal and encoder settings. Silence, noise and music need not compress to the same fraction of their PCM size. There is no reliable fixed “WAV divided by two” rule for every recording.
\nClassic RIFF/WAV uses size fields that put the format’s limit around 4 GiB. The exact room for audio is smaller than that headline boundary because headers and other chunks occupy space. A long multichannel recording can approach it sooner than a stereo mix.
\nAt 48 kHz and packed 24-bit stereo, four hours contain 4,147,200,000 bytes of audio. That is already close enough to the classic boundary to check the complete file format and export implementation carefully. More channels or a 32-bit storage width reaches it sooner.
\nLarge-file variants such as RF64 or W64 can address this problem when both ends of the workflow support them. The EBU’s RF64 specification notice describes a 64-bit format for audio beyond the original size limit. Confirm the recipient’s software support or agree on a split export. The calculator flags payloads near the boundary; it cannot determine a program’s actual format support.
\nRecord sample rate, stored width, channel layout, duration and file count beside the estimate. Name the instruments and intended stereo pairs separately; “two channels” says nothing about which signal belongs on either side. Keep original artist credits, track names and ownership information with the deliverables.
\nCalculate again after changing the export span. Including a count-in or a long decay tail changes every equal-length stem’s payload. If the duration comes from a loop, the bars-and-seconds guide helps establish its musical boundaries before the storage calculation.
", "license": "Content © Yunus Emre Vurgun. Drafted with AI assistance, then reviewed, corrected and published by the developer. Quote with attribution and a link to https://musics.name/articles/wav-file-size-multitrack-stems/; do not republish whole articles without permission.", "provenance": { "drafted": "AI-assisted, against a written house style", "reviewed": "Reviewed, corrected and published by Yunus Emre Vurgun", "publisher": "Yunus Emre Vurgun", "site_developer": "Yunus Emre Vurgun", "disclosure": "https://musics.name/colophon/#how-articles-are-written" }, "source": { "file": "content/articles/wav-file-size-multitrack-stems.md", "syntax": "Markdown with YAML-ish front matter", "published": "2026-10-20", "updated": "2026-10-20", "contributions": "Open an issue or send a correction through https://yunusemrevurgun.com/contact" } }