Choose uncompressed PCM or compressed audio. PCM uses sample rate, stored bit depth and channel count. Compressed audio uses the average bitrate of the whole stream; do not multiply it by channel count again.
How the calculation works
PCM bytes = sample rate × stored bits per sample × channels × seconds ÷ 8. Compressed bytes = average kbps × 1,000 × seconds ÷ 8. The allowance adds an assumed percentage for file headers and metadata. Output uses both decimal and binary units.
Worked example
One minute of 48,000 Hz, 24-bit, two-channel PCM is 17.28 MB before header bytes. One minute at a combined compressed bitrate of 192 kbps is 1.44 MB. A container storing 24 meaningful bits in a 32-bit sample slot needs the stored width, 32, for this estimate.
Before using the result
The tool accepts common stored widths of 8, 16, 24, 32 and 64 bits. Floating-point storage size depends on its stored width, not only its precision. Lossless compressed formats do not have a fixed predictable size from PCM settings alone. Use a measured average bitrate for variable-rate compression. This calculation does not evaluate loudness, clipping or sound quality.
Keep the source values with your project notes. A number calculated from an assumed bitrate or representative sample is only as useful as that assumption. Check a real file or measured export when a deadline, upload cap or storage purchase depends on the answer.