Play back a freshly recorded vocal and there's often a faint "hiss" sitting in the background. The room tone under an acoustic guitar take. The background hum buried in a sampled drum break. The PC fan noise creeping into a livestream. The material changes, but the problem is always the same: how do you quiet the background without killing the attack or nuance you worked so hard to capture?

This isn't a matter of bad equipment or bad technique; it's essentially unavoidable whenever you record anything. In this guide, we'll first cover what this noise is and why EQ alone can't remove it, then walk through practical noise reduction settings (Hz, dB, ms) for four specific materials: home-recorded vocals, acoustic guitar, drum samples, and streaming/podcast audio.

Where Background Noise Comes From

The noise that gets recorded alongside your audio primarily comes from the following sources:

• Microphone Self-Noise: Condenser microphones inherently generate a small amount of electrical noise. A typical mic has a self-noise level of around 15dB-A, which becomes more prominent as you turn up the gain.
• Audio Interface Preamp Noise: For quiet sources like vocals, you often push the preamp gain by +40dB to +50dB. This amplification process brings up a white noise-like hiss.
• Environmental Noise: Air conditioners, ventilation fans, PC cooling fans, and refrigerator motors. Common culprits include 50Hz/60Hz electrical hum and the continuous drone of a fan around 500Hz.
• Room Resonance (Room Modes): Low-frequency muddiness (typically between 100Hz and 250Hz) caused by the acoustics of an untreated room, the same range that clashes with an acoustic guitar's body resonance or a streamer's room tone.

In other words, the battle isn't only about avoiding noise during recording, it's about how you clean it up afterward. Once you understand this shared foundation, the adjustments become a matter of adapting to each material's quirks.

Why Low-Cut EQ Isn't Enough

You might think, "Can't I just cut the low frequencies to make it quiet?" However, the "hiss" that plagues most recordings is mostly broadband noise spread evenly across a wide frequency range.

An EQ is a tool for cutting specific frequency bands, making it poorly suited for targeting noise that is spread out thinly and widely. Even if you apply a high-pass filter (low-cut) at 80Hz with an 18dB/oct slope, you will only remove low-frequency room rumble and plosives. The mid-to-high frequency "hiss" spreading across 2kHz to 10kHz will remain entirely untouched. This is exactly why you need noise reduction (NR) that learns a noise profile and subtracts only the difference.

A Practical Guide by Material

Home-Recorded Vocals: Suggested Reduction Amounts by Style

Vocals vary widely in the ideal reduction amount and timing depending on volume and genre.

• Ballads / Acoustic: Use noise reduction sparingly, with a reduction amount of around -3dB to -6dB. Prioritize preserving subtle breaths and the delicate fade-out of the voice over achieving complete, dead silence.
• Rock / Pop: Since the vocals will be heavily compressed (e.g., 4:1 ratio, -6dB or more of gain reduction) to sit in a dense, loud mix, apply noise processing more firmly (around -10dB to -12dB). Be careful not to crush the consonants.
• Narration / Podcast vocal tracks: Prioritize clarity above all else. A noticeable "hiss" during silent moments causes listener fatigue, so apply NR more aggressively (around -15dB) and consider pairing it with a noise gate.
• Rap: The attack (the transients of the words) is crucial for rhythm and flow. Set the attack time very carefully so that the noise processing doesn't dull the sharp transients at the beginning of phrases.

Acoustic Guitar: Dealing with Room Resonance (100-250Hz)

In home recording environments, specific frequency bands often build up depending on the dimensions of the room. The range around 150Hz (with a Q of 1.5 to 2.0), in particular, tends to overlap with the acoustic guitar's natural body resonance, creating a "muddy" or "boxy" sound.

If you constantly cut this range by -3dB with a standard static EQ, the acoustic guitar will permanently lose its "fullness" and sound unnaturally thin.

This is where Dynamic EQ or Multiband Compression becomes highly effective. You can set it up to act intelligently: "Only cut this specific band (e.g., 150Hz) by a maximum of -3dB to -6dB when it exceeds a certain volume threshold and gets muddy." By setting a fast-but-safe attack time of 10ms to 20ms and a release time of 50ms to 100ms, you can control the unwanted boominess dynamically while preserving the rich, natural sustain of the instrument.

Suggested settings by playing style:

• Steel String Strumming (Pop / Singer-Songwriter): Since the strumming is continuous, noise is generally masked and less noticeable. A reduction of -6dB is usually more than enough. Firmly control the 100Hz to 200Hz buildup with a dynamic EQ (cutting up to -4dB), and add a high-pass filter around 80Hz.
• Nylon String Fingerpicking (Bossa Nova / Classical): Prioritize the delicate nuances of the finger touch above all else and keep reduction under -3dB. Nylon strings have a naturally soft high-end, so aggressive NR will kill their character. Keep any high-pass filtering below 60Hz to maintain the warmth of the body resonance.
• Close Mic Recording (Backing for Vocals): The signal-to-noise ratio is usually excellent, resulting in minimal noise, but the proximity effect boosts the low frequencies significantly. A gentle low-shelf cut of -3dB at 100Hz is often necessary.

Drum Samples: Removing Noise Without Killing the Transients

If you dig through sample libraries much, you've run into this: "This sound is amazing, but that background hiss is ruining it." Whether it's a break sampled from an old vinyl record, a free one-shot you downloaded, or a drum loop you recorded at home, they all suffer from the same problem.

If you apply noise reduction (NR), the background gets quiet. But if you do it carelessly, the initial "thump" of the kick or the "crack" of the snare, the transient attack, becomes dull, resulting in a lifeless sound. This is the hardest part of prepping drum samples.

When teaching your plugin the noise profile, the technique changes depending on the material:

• One-Shot Samples (Single Kicks, Snares, Hats): One-shots often contain a tiny bit of silence (just noise) before or after the hit. You should feed only this silent section to the plugin for learning (typically about 100-200ms of audio). It is critical that you do not include the sound of the drum hit itself in the profile, or the algorithm will try to cancel out the fundamental frequencies of your drums.
• Loop Samples (Drum Breaks, Beats): Because there is almost no true silence inside a loop, you must utilize the tiny tail end of a hit or any brief gap available. If that's not enough, finding a separate silent track recorded in the same environment, or relying on gating, often yields better results.

Streaming / Podcast: Integrating Processing into a Live Stream

For streaming and podcasts, poor audio quality is a direct cause of listener drop-off. Unlike a podcast where you edit "after the fact" in a DAW, live streaming requires real-time audio processing with minimal latency, which sets it apart from the other materials here.

• Integrating with OBS Studio: OBS allows you to insert VST plugins directly as filters on your microphone source. Click the gear icon on your mic source in the "Audio Mixer," select "Filters," and add a "VST 2.x Plug-in." This allows you to run your preferred EQ and compressor and broadcast the processed sound directly to Twitch or YouTube.
• Routing with Voicemeeter: By using a virtual audio mixer like Voicemeeter Banana or Potato, you can route your microphone audio through a DAW or a VST host application to apply complex effects chains, and then route that "cleaned-up voice" into OBS, Discord, Zoom, or any other application simultaneously.

For Those Who Want to Focus on the Work, Not the Noise

Whatever the material, creating a quiet background is the crucial first step. Once it's in place, your EQ moves sound cleaner, your compressor controls dynamics better, and your reverb sits perfectly.

However, the process of "sampling a noise profile, tweaking release times, and finding the delicate reduction amount to avoid watery artifacts" for every single piece of material gets tedious fast.

"Instead of fighting noise every time, what if a tool could learn the environment with a single click...?"

The AIDE AUDIO TP-1 Tone Purifier was developed exactly for this purpose, featuring learning-based noise reduction.

TP-1 also has features tuned for specific materials. For vocals, acoustic guitar, and drums, Transient Protection detects instantaneous spikes in energy and automatically relaxes the NR reduction for pick attacks and drum hits, letting them pass through untouched. For acoustic guitar and streaming, the RESONANCE knob automatically detects and suppresses the muddy 100-250Hz room resonance. And for real-time use cases like streaming, TP-1 runs at a low processing delay of roughly 53ms (at a 48 kHz sample rate), so you can drop it straight into OBS as a VST filter without worrying about lip-sync issues.

If you'd rather spend less time worrying about noise and more time on the work that actually matters, try the free version first.

Related Products

• Achieve a Clear Home Studio Sound with TP-1 Tone Purifier