WAV bit depth and sample rate, explained
Two numbers decide what a WAV file is and how big it gets. Neither of them is a quality dial you should turn up because bigger sounds better.
Short answer for MP3 sources. 16-bit, at whatever sample rate the MP3 already uses. Choose 24-bit only when someone asked for it, 32-bit float only if you are about to process the audio hard, and 48 kHz only when the project is video.
Sample rate: how often the audio is measured
Digital audio is a series of measurements. Sample rate is how many are taken per second, so 44,100 Hz means 44,100 snapshots of the waveform every second.
The number that matters is half of it. A given sample rate can represent frequencies up to half its value, which is why 44.1 kHz reaches 22.05 kHz and comfortably covers human hearing, generally reckoned to top out near 20 kHz and lower than that for most adults. Doubling the rate does not make the audible part more detailed. It extends a range you cannot hear.
| Rate | Where it is used | Worth choosing when |
|---|---|---|
| 44.1 kHz | CD, most music, streaming | Burning a CD, or matching a music project |
| 48 kHz | Video, broadcast, film | The audio will sit on a video timeline |
| 32 kHz | Some broadcast and older formats | Something specifically requires it |
| 22.05 kHz | Speech, older games, embedded devices | Size matters more than fidelity |
| 96 and 192 kHz | Recording and mastering sessions | You are tracking and processing, not delivering |
Changing the sample rate means resampling, which is arithmetic that reconstructs the waveform at new measurement points. Good resamplers are very good, but the operation can only ever be approximate, so avoid it when nothing requires it. That is why the converter here defaults to keeping the source rate.
Bit depth: how precisely each measurement is stored
Bit depth is the number of possible values each sample can take. 16-bit gives 65,536 levels. 24-bit gives about 16.7 million.
What that buys you is dynamic range, the gap between the loudest sound and the noise floor underneath it. 16-bit provides roughly 96 dB, which is more than a CD ever uses in practice. 24-bit provides around 144 dB, well past what any room or microphone can deliver. The extra bits are not more detail in the music. They are more room underneath it.
| Depth | Dynamic range | Size vs 16-bit | Use it for |
|---|---|---|---|
| 16-bit | About 96 dB | Baseline | Delivery, CDs, anything from an MP3 source |
| 24-bit | About 144 dB | 50 percent larger | Recording, mixing, client and studio requests |
| 32-bit float | Effectively unlimited | Twice as large | Working files where peaks may exceed full scale |
Why 32-bit float exists
In 16 and 24-bit files, a sample louder than full scale has nowhere to go and gets clipped flat, which is permanent and sounds harsh. Floating point stores values above and below that limit instead, so a take recorded far too hot can be turned down afterwards and come back intact.
That is a genuinely useful property while you are working. It is not useful for delivery, since most players and every CD expect integer PCM. The converter on this site deliberately leaves float output unclamped, because preserving those peaks is the entire reason to choose it. The 16 and 24-bit paths clamp instead, so a hot master cannot wrap around into distortion.
How the two numbers set file size
WAV size is arithmetic, with no dependence on the music itself:
bytes per second = sample rate × (bit depth ÷ 8) × channels
So 44,100 × 2 × 2 comes to 176,400 bytes per second, or about 10.1 MB per minute for the standard CD format. Some worked examples:
| Format | Per minute | Per hour |
|---|---|---|
| 44.1 kHz, 16-bit, stereo | 10.1 MB | 606 MB |
| 44.1 kHz, 24-bit, stereo | 15.1 MB | 908 MB |
| 44.1 kHz, 32-bit float, stereo | 20.2 MB | 1.21 GB |
| 48 kHz, 24-bit, stereo | 16.5 MB | 989 MB |
| 44.1 kHz, 16-bit, mono | 5.0 MB | 303 MB |
| 96 kHz, 24-bit, stereo | 33.0 MB | 1.98 GB |
Try your own numbers in the WAV file size calculator.
What this means for an MP3 source
An MP3 has already had audio removed, and what remains sits comfortably inside 16 bits. Writing it into a 24-bit file stores the same information with more zeros after it. The file grows by half and the sound does not change.
There is one situation where 24 or 32-bit is the right call: you are importing into a session that runs at that depth and you plan to apply heavy processing. Keeping the extra headroom through the chain avoids introducing new limits partway. That is about the processing, not about the MP3.
A note on dithering
Reducing bit depth throws away the least significant bits, which introduces a small, correlated error that can sound like faint distortion on quiet passages. Dither adds a tiny amount of random noise to make that error behave like ordinary hiss instead, which the ear tolerates much better.
You need it when exporting a 24-bit mix to a 16-bit master. You do not need it when converting an MP3, because the compression artefacts already sit far above the level where 16-bit quantisation would matter. This is why the converter here does not apply dither and does not pretend to.
Common questions
Should I convert MP3 to a 24-bit WAV?
Only if something specifically asked for it. An MP3 does not contain enough detail to fill 16 bits properly, let alone 24, so the extra 50 percent of file size buys you nothing on its own. It becomes worthwhile when you are about to process the audio heavily inside a 24-bit or floating point session.
What does 44.1 kHz 16-bit mean?
The audio is measured 44,100 times a second, and each measurement is stored as one of 65,536 possible levels. That combination is the Red Book CD standard, it covers the full range of human hearing, and it is what most software expects unless told otherwise.
Is 48 kHz better than 44.1 kHz?
Not better, just different. 48 kHz is the standard for video and broadcast, 44.1 kHz for CD and most music. Match whatever your project uses. Converting between them means resampling, which is a small quality cost for no benefit if nothing required it.
What is 32-bit float for?
Headroom. In a 32-bit float file, samples above 0 dBFS are stored rather than clipped, so a recording that peaked too hot can be pulled back down afterwards with no damage. It is a working format. Deliver 16 or 24-bit.
Do I need dithering?
Only when reducing bit depth on audio that started with more, for example exporting a 24-bit mix to a 16-bit master. Converting an MP3 to 16-bit WAV does not need it, because the MP3 noise floor already sits well above the point where 16-bit quantisation matters.
What sample rates does MP3 support?
32, 44.1, and 48 kHz under MPEG-1; 16, 22.05, and 24 kHz under MPEG-2; and 8, 11.025, and 12 kHz under MPEG-2.5. Anything outside that list has to be resampled before it can be encoded, which is why a 96 kHz session file becomes 48 kHz on the way to MP3.