How to Use Compression During Recording: When and How Much

After more than fifteen years working sessions across commercial studios, live broadcast rigs, and home setups, I have seen more recordings damaged by heavy-handed compression than by almost any other single mistake. The irony is that compression during recording is often applied with good intentions: protecting the gain structure, taming a dynamic vocalist, or keeping bass guitar from clipping the interface input. The problem is not the intent, it is the execution.

In this guide I will cover exactly when to apply compression before it hits your recorder, how to choose settings that protect rather than damage the signal, what the key parameters actually do to audio in practical terms, and where the common traps are for each instrument type. I am not going to suggest you buy a specific unit to solve your problems. I am going to give you the knowledge to use whatever you already have more effectively.

Why Recording Compression Is a Different Discipline to Mixing Compression

When you compress during mixing, you are working with a recorded file that you can always revert. When you compress during recording, the effect is printed permanently to the track. That distinction changes everything about how conservative you need to be with your settings. In mixing, a compressor with too fast an attack can be bypassed in seconds. In recording, that same compressor has already eaten the transient and it is gone.

The purpose of recording compression is almost always gain control rather than tonal shaping. You are trying to prevent peaks from clipping your analogue-to-digital converter, bring a highly dynamic source into a range that sits consistently on the track, and give yourself a more manageable starting point in the mix. A unit like the dbx 160A or the Universal Audio 610 channel strip applies optical or VCA gain reduction that can be set conservatively enough to do that job without audibly squashing the performance.

I use a simple rule in my own sessions: if I am reaching for more than 6 dB of gain reduction on a recording compressor, I ask myself whether there is a more fundamental problem with gain staging, microphone placement, or the performance itself. Heavy compression on the way in is usually masking a different issue rather than solving it.

Understanding the Four Controls That Matter Most

Threshold determines at what signal level the compressor begins to act. Set it too low and the compressor works constantly, flattening the dynamic range of the entire performance. Set it too high and peaks still clip the converter. For most vocal recording sessions I start with the threshold set so the compressor only engages on the loudest syllables and chorus lines, leaving the quieter passages completely untouched.

Ratio controls how aggressively the compressor clamps down once the signal crosses the threshold. A ratio of 2:1 means that for every 2 dB the signal rises above threshold, the compressor only allows 1 dB to pass through. That is gentle enough for recording use on most sources. A ratio of 4:1 is more noticeable but still reasonable. Once you start pushing to 8:1 or 10:1 during recording, you are almost certainly over-compressing and you will hear it in the playback as a kind of sucking or pumping artefact.

Attack and release are the controls that most engineers underestimate on a recording compressor. A fast attack setting, anything under about 5 milliseconds, will catch and reduce transients before the converter sees them, which is sometimes exactly what you need for a hard-picked acoustic guitar or a snare drum. However, a fast attack on a vocal will dull the initial consonant sounds and make the performance sound flat and lifeless by the time it reaches the mix. For vocals, a slower attack of 20 to 50 milliseconds lets the initial transient through and only controls the body of the note, which preserves the natural character of the performance. Release should be set fast enough that the compressor recovers before the next phrase begins, but not so fast that it creates audible pumping on sustained notes.

Instrument-Specific Approaches That Actually Work

For lead vocals, which represent the highest-stakes recording compression decision in most sessions, I recommend a ratio of no more than 3:1, a threshold set so gain reduction averages around 3 to 4 dB on the loudest phrases, a slow to medium attack of at least 20 milliseconds, and an auto or medium release. On hardware such as the Universal Audio LA-610 or an API 512c feeding a dedicated compressor, this approach keeps the vocal on the track without printing obvious compression artefacts. On an audio interface with a built-in compressor, such as those found on the Focusrite Clarett Plus range, use the lightest available preset as a starting point and back off from there.

Bass guitar is a source where recording compression earns its keep most convincingly. A DI bass signal can swing 20 dB or more between a softly played verse line and a hard-picked chorus, and that range will create havoc with your gain structure and your mix. A ratio of 4:1 with a fast attack of around 5 to 10 milliseconds and a medium release tames that swing without removing the punch of the attack. The dbx 160A remains a favourite of mine for this task because its over-easy compression curve handles bass naturally, avoiding the hard-knee clamp that cheaper units sometimes introduce.

Acoustic guitars and room-miked drum overheads present a different challenge. For acoustic guitars, I generally avoid compression entirely during recording and instead sort out the gain staging, then compress in the mix where I have full control. For drum overheads, a gentle 2:1 ratio with a slow attack preserves the snap of the cymbals and kick while preventing the very loudest crashes from spiking the converter. On kick drum close mics, a faster attack and moderate ratio can help, but I still rarely print more than 4 to 6 dB of reduction.

Hardware Versus Software Compression During Recording

Most modern audio interfaces route signal through the analogue input stage before the analogue-to-digital converter. If you have an interface with sufficient headroom, such as the Universal Audio Apollo x series or the Audient iD44, you can often record clean without any compression at all, relying on the converter headroom to handle transient peaks. These interfaces typically offer 24 dB or more of converter headroom above 0 dBFS, which means a well-gain-staged signal will not clip even with some dynamic variation.

Hardware compressors placed in the signal chain before the interface remain the gold standard for recording compression because they operate entirely in the analogue domain and can impart a pleasing character alongside their gain control function. Units like the Warm Audio WA-2A or the Shadow Hills Mono GAMA offer optical compression that is gentle and musical at conservative settings. The key advantage is zero latency: the compressor acts on the signal before it reaches the converter, so there is no processing delay and no digital artefacts.

Software plugins running in a low-latency monitoring chain, as offered by the UAD Console application or the Focusrite Control software with Focusrite-compatible plugins, can also work well for recording compression. However, I always check that the plugin latency is compensated correctly before trusting the monitoring mix, and I remain cautious about printing software compression during recording unless I have a specific reason to do so. The option to apply compression later in the mix exists precisely because software processing can be deferred without penalty.

Setting Up Your Gain Structure Before You Touch the Compressor

Recording compression cannot fix a poorly gain-staged signal. If the microphone preamp gain is set so high that the signal is already distorting at the preamp stage, a compressor placed after it will compress distorted audio, which is unusable. The correct approach is to set the preamp gain so that the loudest expected peak lands at around -18 dBFS on the digital meter, leaving a comfortable safety margin below 0 dBFS. From that starting point, a recording compressor with a ratio of 2:1 to 4:1 can handle the remaining dynamic variation cleanly.

I check gain structure by asking the performer to sing or play at their absolute loudest expected level, watching the input meter on my interface or DAW, and adjusting the preamp until those loudest peaks sit around -12 to -18 dBFS. The compressor then supplements that setting by catching the unexpected peaks that exceed what was tested. This two-stage approach, gain staging first and compression second, is more reliable than trying to use compression as a substitute for correct preamp gain setting.

How Much Compression Is Too Much During Recording

The threshold for too much compression on the way in is lower than most engineers expect. More than 6 dB of average gain reduction on a vocal is almost always too much for a recording compressor. On bass guitar, up to 8 dB of peak reduction can work if the ratio is gentle and the release is well-matched to the tempo of the track. On drums, I would not exceed 4 dB of reduction on overheads and would be cautious even with that on a slow-attack setting.

A useful test is to record a short section with and without the compressor in the chain and compare the two signals directly on playback. If the compressed version sounds noticeably different from the clean version in terms of energy and dynamics, the settings are too aggressive for recording use. The compressed version should sound essentially identical to the uncompressed version on casual listening, with the difference only apparent on the loudest transients where the uncompressed signal would have clipped.

Conclusion

Recording compression done well is invisible. The goal is a signal on tape or disk that has consistent level, clean peaks, and full dynamic character intact for the mix engineer to work with. Conservative ratios, appropriate attack times for each instrument, and a solid gain structure underneath the compressor are the three elements that determine whether compression during recording helps or harms the session. Get those three things right and the compressor does its job without announcing itself.