How this calculation works
Each sample frame contains one stored sample from every channel. Count complete frames, multiply by bytes per frame and add an explicit container-overhead assumption.
Total bytes = floor(seconds × sample rate) × channels × stored bits / 8 + entered header bytesA worked example
Sixty seconds of 48 kHz, 24-bit stereo has 2,880,000 frames and 17,280,000 PCM bytes. Adding an illustrative 44-byte header gives 17,280,044 bytes, about 16.48 MiB.
Good to know
Uncompressed packed-width storage only, not compressed audio prediction or file generation. Header is entered, not universal; additional chunks, padding, RF64 and format variants differ.
Reference: MDN: digital audio concepts (opens in a new tab) ↗
A sample frame includes every channel
Sample rate describes frames per second for each channel. Stereo does not halve the per-channel rate; it stores two samples in each frame. At 24 bits, each channel contributes three packed bytes.
For 48,000 frames per second and two channels, the raw rate is 288,000 bytes per second. Confusing bits with bytes or forgetting channels produces a large error even when the duration is correct.
The uart throughput and frame time calculator adds model UART start, data, parity and stop overhead, finite message-batch gaps and useful payload-bit throughput without treating baud as byte rate.
Storage width is not perceived quality
The selected bit depth is the number of bits actually stored per channel sample in this model. A recording system can use a larger container width than its meaningful signal precision, so use the stored format when planning storage.
The result does not predict audible quality, noise floor, codec performance or a compressed download size. Lossless and lossy formats need different information beyond a raw PCM sample count.
Decimal and binary units answer the same byte count
MB divides bytes by one million; MiB divides by 1,048,576. Both are shown from the same total rather than treating two labels as interchangeable. Fractional final frames are discarded.
Classic RIFF uses finite-width size fields. The warning checks whether the entered data and total would exceed those fields, but it does not validate a real file or choose an appropriate extended container for your software.
Use the result with the next measurement
The UART throughput and frame time calculator exposes transport framing overhead. PCM file bytes are stored payload; estimating a serial transfer also requires the interface overhead and any message gaps.
Uncompressed packed-width storage only, not compressed audio prediction or file generation. Header is entered, not universal; additional chunks, padding, RF64 and format variants differ.
A few questions, answered
Does this estimate MP3 size?
No, it counts uncompressed stored PCM samples.
Is a WAV header always 44 bytes?
No, extra format chunks and metadata can make it larger.
Why are MB and MiB different?
They divide the same bytes by different unit sizes.