How Audio Buffer Size Affects Latency in a DAW

Every DAW asks for a buffer size in its audio settings, usually in samples, with values such as 64, 128, 256, or 512. The number controls how much audio the computer processes at once, and therefore how long the delay is between playing a note and hearing it. Choosing the setting well means balancing responsiveness against the processing load the computer can handle.

What a buffer is

Audio runs as a continuous stream of samples. The operating system and the audio driver do not process each sample individually, because that would be inefficient. Instead they work in blocks. The buffer is one such block. The computer fills the block, the DAW processes it, and the result goes to the interface. While one block is being processed, the next one is being prepared, which keeps playback smooth.

A larger block gives the processor more time, which helps reliability. A smaller block gives the processor less time, so every block has to finish quickly. If processing takes longer than the block allows, the output glitches with pops, clicks, or dropouts.

How buffer size turns into latency

Latency is the time between an event and the sound it produces. For buffer size, the basic formula is simple. Divide the number of samples by the sample rate to get seconds. At a sample rate of 48,000 samples per second, a 128-sample buffer lasts 2.67 milliseconds. A 512-sample buffer lasts about 10.67 milliseconds.

The table below shows one-way buffer delay at 48 kHz. Each value covers only the time for one buffer. Real round-trip latency is larger because the converters, the driver, and the plugin chain all add their own delay.

Buffer size (samples)	One-way delay at 48 kHz
32	0.67 ms
64	1.33 ms
128	2.67 ms
256	5.33 ms
512	10.67 ms
1024	21.33 ms

At 1024 samples, the delay is large enough that playing a keyboard part feels sluggish. Many players notice delays above about 10 milliseconds when they play in real time. The numbers are not a strict limit, because players vary, but they give a useful reference point.

Why recording and mixing want different settings

Recording and mixing pull in opposite directions. When you record a live instrument or a vocal, you want low latency so the performer hears their own sound without a noticeable lag. A small buffer, such as 64 or 128 samples, gives that response. During tracking, the DAW usually has few plugins running, so the processor can keep up.

Mixing is different. When recording is finished, you are no longer playing live, and a few milliseconds of delay does not matter. You can raise the buffer size to 512 or 1024 samples. The extra time per block lets the processor handle many plugins, large sample libraries, and complicated routing without dropouts. Many DAWs offer a separate setting for recording and playback, or let you change the buffer size on the fly.

The CPU side of the trade-off

A small buffer demands more from the processor because it must finish each block faster and more often. Every plugin in the chain has to run for each block. A heavy synthesizer or a convolution reverb can take a large share of the time available. This is why the same project can play smoothly at 512 samples and crackle at 64.

Two practical habits help. First, freeze or bounce tracks that use heavy plugins once they are finished, which removes their load from the live session. Second, watch the DAW's CPU meter during playback and leave headroom for the tracks still to come.

Sample rate and buffer size work together

Buffer delay depends on the sample rate as well as the buffer size. At 96 kHz, the same 128-sample buffer lasts half as long as it does at 48 kHz, which means a shorter delay, but the processor must handle twice as many samples per second. A higher sample rate therefore reduces delay for a given buffer size while increasing CPU load. The two settings should be chosen together. A common setup is 48 kHz with a 128-sample buffer for tracking, then a larger buffer for mixing.

The driver matters as much as the setting

The buffer size is only a request. The audio driver decides how the hardware actually behaves. On Windows, ASIO drivers from the interface maker usually give the best low-latency results, while generic drivers add more delay. On macOS, Core Audio is generally reliable, and most interfaces work well with it. A good driver can run stably at a smaller buffer than a poor one can. If a small buffer causes trouble, check the driver before blaming the computer.

Checking the driver

Open the interface's own control panel first and set the driver to its recommended mode. Update the firmware and driver when the manufacturer releases a new version, because fixes to buffer handling are common in updates. Close other heavy applications while testing, since background tasks compete for the same processor time.

Setting up a sensible configuration

Open the DAW's audio settings and choose the interface. Set the sample rate to 48 kHz or 44.1 kHz, depending on your project. Then set the buffer to 128 samples for recording. Play a few notes or record a short take and listen for clicks. If the sound is clean, leave it. If not, move up to 256 samples. Before mixing, raise the buffer again to 512 or higher, and watch the CPU meter while playing back the whole song.

Latency you can hear and feel

Two kinds of delay matter when you play. The first is the audible gap between a key press and the sound. The second is the physical feel of the instrument. Drummers often describe a delay as a loss of tightness, because the sound arrives after the stick has already moved on. Guitarists notice a loose, slightly late attack. Singers hear the delay in their own voice through the headphones, which can make pitch and timing harder to hold. These effects are personal, and a setting that feels fine to a producer working with keys may feel unusable to a drummer.

The practical test is simple. Play the part you care about through the headphones at the chosen buffer size, and ask whether the performance feels natural. If it does not, lower the buffer or use direct monitoring on the interface, which sends the input to the headphones before the computer sees it. Direct monitoring gives the performer near-zero delay, but it also means the performer does not hear the DAW's effects during recording, so it works best with a dry signal.

Key takeaways

Buffer size is the number of samples the computer processes at once. It sets a minimum delay between input and output, and a larger buffer adds more delay while reducing the risk of dropouts. The delay follows directly from the sample rate, so you can calculate it.

Use a small buffer for recording and a larger one for mixing. Keep the driver up to date, freeze heavy tracks, and watch the CPU meter. The right setting keeps the performer comfortable and the playback free of glitches.