A delay plug-in reports 512 samples, a clip edit is 2,205 samples long, or a manual gives latency in samples while your DAW shows milliseconds. Knowing how to convert samples to milliseconds at 44.1 and 48 kHz lets you tell whether that number is musically meaningful, technically normal, or simply being read at the wrong sample rate.
The calculation is simple, but the sample-rate part matters. The same number of samples occupies less time at 48 kHz than it does at 44.1 kHz because 48,000 samples pass every second instead of 44,100.
The samples-to-milliseconds formula
Use this formula:
`milliseconds = (samples / sample rate) × 1,000`
For a 44.1 kHz project, use 44,100 as the sample rate. For a 48 kHz project, use 48,000.
So, 512 samples at 44.1 kHz are:
`(512 / 44,100) × 1,000 = 11.61 ms`
At 48 kHz, the same 512 samples are:
`(512 / 48,000) × 1,000 = 10.67 ms`
That difference may look small, but it is enough to matter when you are checking recording latency, lining up layered drum hits, setting a short delay, or comparing plug-in latency reports.
The reverse calculation is useful too when a plug-in gives you a time value and you need a sample offset:
`samples = (milliseconds / 1,000) × sample rate`
For example, 50 ms at 44.1 kHz equals 2,205 samples. At 48 kHz, 50 ms equals 2,400 samples.
Why 44.1 kHz and 48 kHz give different answers
Sample rate means how many individual sample frames are captured or played back each second. At 44.1 kHz, one sample lasts:
`1 / 44,100 seconds = 0.02268 ms`
At 48 kHz, one sample lasts:
`1 / 48,000 seconds = 0.02083 ms`
A single sample is extremely short in either case. The distinction becomes easier to hear and manage once hundreds or thousands of samples are involved.
Imagine you move a vocal consonant 1,000 samples later on the timeline. At 44.1 kHz, that move is 22.68 ms. At 48 kHz, it is 20.83 ms. If you copied a sample-count offset from one session to another without converting it, the timing would change.
This is why samples are precise for editing inside one project, while milliseconds are often clearer for communicating a time setting across different sessions, devices, and sample rates.
Quick reference values
| Samples | 44.1 kHz | 48 kHz |
|---|---|---|
| 1 | 0.02268 ms | 0.02083 ms |
| 64 | 1.45 ms | 1.33 ms |
| 128 | 2.90 ms | 2.67 ms |
| 256 | 5.80 ms | 5.33 ms |
| 512 | 11.61 ms | 10.67 ms |
| 1,024 | 23.22 ms | 21.33 ms |
| 2,048 | 46.44 ms | 42.67 ms |
| 4,096 | 92.88 ms | 85.33 ms |
These values are especially handy when reading audio-interface buffer settings. A 256-sample buffer is not automatically “5 ms” or “6 ms” in every setup. It depends on whether the session runs at 48 kHz or 44.1 kHz, and the real round-trip latency also includes input and output conversion plus any driver safety buffers.
Worked example: matching a short delay by time
Suppose you want a 30 ms slap-style delay on a percussion sound. Your delay device accepts sample values rather than milliseconds.
At 44.1 kHz:
`(30 / 1,000) × 44,100 = 1,323 samples`
At 48 kHz:
`(30 / 1,000) × 48,000 = 1,440 samples`
Both settings create a 30 ms delay. The sample number changes because the project processes more sample frames per second at 48 kHz.
This is also a useful way to check a manual delay compensation value. If a hardware insert is measured at 15 ms, enter 661.5 samples at 44.1 kHz or 720 samples at 48 kHz. Since most DAWs require whole samples, use 662 samples at 44.1 kHz and listen for the best alignment. A half-sample cannot be represented as a simple timeline shift, although some processing can use fractional-delay techniques internally.
Samples, milliseconds, and tempo are different measurements
Samples and milliseconds describe absolute time. BPM describes musical time. They connect only after you choose a tempo.
At 120 BPM, one quarter note lasts 500 ms. That is 22,050 samples at 44.1 kHz or 24,000 samples at 48 kHz. The musical length is identical, but the sample count differs.
At 128 BPM, one quarter note is 468.75 ms. It equals 20,671.875 samples at 44.1 kHz and 22,500 samples at 48 kHz. In practice, a DAW can place events at sub-sample musical positions in its internal timing system, but rendered audio still resolves to actual sample frames.
For rhythmic effects, start with a BPM-based time value such as an eighth note, dotted eighth, or quarter-note triplet. Convert that value to samples only when a device or workflow requires it. For sound design, milliseconds are often the more useful creative unit. A 12 ms Haas-style offset, a 25 ms slap, or a 60 ms pre-delay remains the same physical time regardless of tempo.
Common mistake: using the wrong sample rate
The most frequent error is dividing by 44,100 because it is familiar, even though the project is running at 48 kHz. That makes every conversion slightly too long.
For instance, 2,400 samples are exactly 50 ms at 48 kHz. If you calculate them as though they were 44.1 kHz samples, you get 54.42 ms. That 4.42 ms difference can be noticeable when aligning close-miked drum layers or checking a latency offset.
Check the project sample rate, not just the sample rate of an imported audio file. A 44.1 kHz loop imported into a 48 kHz session is normally resampled by the DAW. Once it is in the 48 kHz project timeline, its current sample positions should be interpreted at 48 kHz.
Do not confuse this with playing 44.1 kHz audio as if it were 48 kHz without proper resampling. That is a playback-rate error that changes both duration and pitch. It is not merely a different way of displaying milliseconds.
When rounding is safe, and when it is not
For a one-off edit, rounding to the nearest whole sample is usually correct. Audio files are made of discrete sample frames, after all.
For timing notes, presets, and documentation, keep one or two decimal places in milliseconds. Saying a 512-sample buffer at 44.1 kHz is 11.61 ms is more useful than calling it 12 ms, especially when you are comparing several stages of latency.
Avoid repeatedly converting and rounding. For example, converting 23.22 ms to 1,024 samples is fine if you keep enough precision. But if you first round that value to 23 ms, then convert back at 44.1 kHz, you get 1,014 samples. Ten samples may be insignificant for a reverb tail, but not necessarily for phase-sensitive parallel processing.
A practical way to check your numbers
When a device gives a sample count, divide by the active session rate and multiply by 1,000. When it gives milliseconds, divide by 1,000 and multiply by the active session rate. Keep the original unit visible until you finish the calculation.
For repeated conversions, a samples-to-milliseconds calculator such as the one at BeatTiming can reduce arithmetic slips, particularly when you are moving between 44.1 and 48 kHz projects. Still, the useful habit is knowing what the result means: samples are fixed frames, milliseconds are elapsed time, and sample rate is the bridge between them.
If an edit sounds late, do not assume the sample count is wrong. Confirm the project rate first, then check whether you are hearing a real timing offset, plug-in latency, or a creative choice that simply needs to be adjusted by ear.
