Audio interface buffer settings are usually best kept low for recording and high for mixing. For recording (especially when you monitor through the DAW), use the smallest buffer your computer can handle without pops/clicks; for mixing, raise the buffer to give the CPU more time for plug-ins and large sessions.
What the buffer actually changes (in plain terms)
Your computer and interface don’t process audio one continuous sample at a time. They process audio in chunks. The buffer size is the chunk length, usually measured in samples (32, 64, 128, 256, 512, 1024, etc.). Bigger chunks mean the computer has more breathing room to finish calculations before the next chunk is due. Smaller chunks mean tighter deadlines.
That deadline is why buffer size affects two things that matter to you:
- Monitoring/playing feel (latency): how delayed the sound feels while performing.
- Stability: whether you get crackles, dropouts, or “CPU overload” errors.
A useful mental model:
- Small buffer = low delay, high stress (more likely to glitch if the session is heavy).
- Large buffer = higher delay, lower stress (more stable for dense mixes).
Why recording and mixing want opposite settings
Recording: you’re a human in the loop
When you record vocals, guitar, MIDI keys, or anything you perform, you react to what you hear in real time. If the delay is noticeable, timing and pitch confidence can suffer.
That’s why recording typically wants:
- Lower buffer sizes (commonly 32–128, sometimes 256 depending on system)
- A goal of “feels immediate” monitoring
The key detail: you don’t need maximum plug-in capacity while tracking. You need responsiveness.
Mixing: you’re not performing into the system
When mixing, you’re not trying to play in time with a live monitoring loop. You can tolerate extra delay because you’re mostly starting/stopping playback and adjusting parameters.
That’s why mixing typically wants:
- Higher buffer sizes (commonly 512–1024, sometimes higher if available)
- A goal of “never glitches during playback” while using lots of plug-ins
The simplest “good defaults”
These aren’t rules; they’re practical starting points that work for many setups.
Recording / tracking presets
- 32–64 samples: best feel, but most demanding; often only realistic on strong systems with light sessions
- 128 samples: a common sweet spot for tracking with a few plug-ins
- 256 samples: still workable for many performers, often much more stable (especially with virtual instruments)
Mixing presets
- 512 samples: good stability without feeling sluggish in basic edits
- 1024 samples (or max): best for heavy plug-in chains, big sample libraries, and dense routing
How buffer size translates to “feel”
Buffer size is measured in samples, but what you feel is milliseconds. A quick approximation:
One-way buffer time (ms) ≈ (buffer samples ÷ sample rate) × 1000
Example at 48 kHz:
- 64 samples ≈ (64 ÷ 48000) × 1000 ≈ 1.33 ms
- 256 samples ≈ 5.33 ms
- 1024 samples ≈ 21.33 ms
Real-world “round trip” monitoring latency (input → DAW → output) is higher because it includes input/output conversion and driver overhead, not just the buffer. Still, the buffer is the big lever you control. Many official explanations frame it the same way: larger buffers increase latency but give the system more processing time. (Steinberg Súgóközpont)
Recording use-cases: choose the buffer based on monitoring method
1) Direct monitoring (interface mixer) — buffer matters less
If you monitor straight from the audio interface (often called direct monitoring), your live input doesn’t have to travel through the DAW and back. That means you can often record comfortably even at a moderate buffer—because you’re not listening to the DAW-processed signal.
Use this approach when:
- You’re recording vocals/instruments and don’t need to hear heavy DAW effects in real time
- Your session is already large and you want stability
Practical setting:
- 256–512 can be fine if you’re not software-monitoring
2) Software monitoring (through the DAW) — buffer is critical
If you need to hear amp sims, pitch correction, guitar FX, or virtual instrument processing live, you’re monitoring through the DAW path. This is where low buffer sizes matter most.
Practical setting:
- Start at 128
- Drop to 64 if it still feels delayed
- If the system crackles, go back up (or reduce real-time processing)
3) Virtual instruments and MIDI — often needs the lowest buffer you can run
MIDI performance is extremely sensitive to delay because your fingers and ears expect immediate response. Even if audio recording is “okay” at 256, playing a fast piano part might feel wrong.
Practical setting:
- Try 64–128
- Use track freezing/bouncing if the session is heavy
Mixing use-cases: stability beats latency
1) Heavy plug-in sessions
Oversampling, linear-phase EQ, convolution reverbs, AI noise reduction, look-ahead limiters—these can be expensive. High buffers give the CPU time to finish processing each block without missing deadlines.
Practical setting:
- 512 for medium mixes
- 1024 (or max) for large mixes
2) Editing while mixing
Sometimes a huge buffer makes the DAW feel sluggish when you press play/stop or scrub. If that bothers you, treat buffer like a “gear”:
- Edit at 256–512
- Mix/print at 1024
A decision tree you can actually use
If you hear clicks/pops/dropouts while recording:
- Raise buffer one step (64 → 128 → 256)
- Disable or bypass the heaviest real-time plug-ins (look-ahead limiters, linear-phase, oversampling)
- Freeze/bounce virtual instruments or heavy tracks
- If available, switch to direct monitoring for the live input
If monitoring feels delayed while recording:
- Lower buffer one step (256 → 128 → 64)
- Reduce real-time plug-ins on the monitoring path
- Prefer low-latency plug-ins while tracking (simple EQ/comp instead of heavy processors)
- If you must keep heavy effects, consider printing a temporary sound (commit) and monitor that
If mixing playback stutters:
- Increase buffer (256 → 512 → 1024)
- Raise audio engine safety/processing settings if your DAW provides them
- Freeze tracks or disable oversampling until final print
“But I want low buffer AND lots of plug-ins”
In practice you trade one for the other, unless your computer is significantly overpowered for the session. The best way to get both is workflow, not a magic number:
- Track with a low buffer and a lightweight monitoring chain
- Switch to a high buffer for mixing and sound design
- Commit (freeze/print) CPU-heavy tracks once you like the sound
Think of the buffer as a mode switch: “performance mode” vs “processing mode.”
Where to change it (and why it sometimes looks locked)
Many DAWs expose the buffer control, but on some systems the actual setting is controlled in the interface driver/control panel. Ableton’s documentation, for example, describes changing buffer size via the audio interface’s control panel on Windows ASIO. (help.ableton.com)
If the buffer dropdown is greyed out, it usually means:
- The driver requires changes in its own utility, not inside the DAW
- Another application is currently holding the interface at a fixed setting
Two practical “profiles” to save (if your driver allows presets)
If your interface software supports profiles, save:
- Tracking: 64 or 128 samples
- Mixing: 1024 samples (or maximum stable)
If it doesn’t, just write the numbers down. The point is to stop guessing mid-session.
Why does this matter
Buffer size is one of the few settings that directly controls whether recording feels natural and whether mixing stays stable; choosing the right one prevents timing frustration during takes and prevents crashes or glitches during heavy mixes.
Sources
- Steinberg Help Center — Audio card latency (Steinberg Súgóközpont)
- Focusrite Support — What is latency in audio? (Focusrite Támogatás)
- Ableton Help — Changing the buffer size/sample rate of an ASIO audio interface (Windows) (help.ableton.com)
