> ## Content Index
> Fetch the complete content index at: https://soundforgepro.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# How to Use the Noise Gate in Sound Forge Pro
- URL: https://soundforgepro.com/how-to-use-noise-gate-sound-forge-pro/
- Published: 2026-04-15T08:50:33.000Z
- Updated: 2026-09-20T20:59:26.000Z
- Description: A seven-step Sound Forge noise-gate workflow for setting Level, Attack and Release without cutting quiet words or natural tails.
- Author: Erick Finn
- Tags: Sound Forge Restoration Guides

A noise gate is useful when the unwanted sound bothers you mainly between words or notes. It closes when the signal falls below a level you set and opens when wanted audio rises above it. What it can’t do is remove fan noise, hiss, or hum while speech is holding the gate open. That boundary decides whether this tool will help before you touch a control.

Part of the [audio restoration guide](https://soundforgepro.com/audio-restoration/): this page covers noise between phrases, not broadband denoising or deleting silence from the timeline.

Quick answer: select a test passage, choose FX Plug-Ins → Apply Plug-In Chain and add coreFX Gate. With Peak detection and a fully wet mix, use a noise-only pause as a threshold reference, then audition the weakest words. Lower Level if a consonant never opens the gate; shorten Attack if it opens too slowly. Noise still audible under speech needs a different treatment, not a higher threshold.

## Check whether a gate fits the recording

Listen once with the wanted sound active. If the noise is acceptable under speech or music but distracting in the pauses, a gate is a sensible test. When it stays objectionable throughout the performance, gating the pauses only makes the changing background more obvious and cleans nothing in the active material. If you need to remove whole silent sections or trim file boundaries rather than react to level, use [Sound Forge Auto Trim/Crop](https://soundforgepro.com/sound-forge-auto-trim-crop/) instead.

The [current Boris FX Noise Gate help](https://cdn.borisfx.com/borisfx/Documentation/soundforge/2026/en/content/noise%5Fgate.htm) defines the processor by this level decision: signal below the specified threshold is removed, which makes it suitable for noise in silent breaks. A gate doesn’t classify a sound as fan, hiss, breath, or voice; it reacts to level.

| What you hear                                     | Likely cause                                                | First move                                                                       |
| ------------------------------------------------- | ----------------------------------------------------------- | -------------------------------------------------------------------------------- |
| Noise remains in pauses                           | Threshold, bypass, dry Mix or detector routing may be wrong | Verify the effect and detector, then remeasure before raising Level              |
| First consonant or transient is missing           | Attack may be slow or Level too high                        | Lower Level if the sound never opens the gate; shorten Attack if opening is slow |
| Word ending, breath, or reverb tail is cut        | Release is too short, or Level is too high                  | Lengthen Release, then lower Level if needed                                     |
| Gate chatters around quiet words                  | Wanted audio is hovering around Level                       | Lower Level and lengthen Release                                                 |
| Noise vanishes in pauses but returns under speech | The gate is working by level, not identifying noise         | Use Noise Reduction or another repair process                                    |
| Pauses sound unnaturally dead                     | Full suppression is too abrupt                              | Use gentler expansion or controlled room tone                                    |

## Which Noise Gate should you open in Sound Forge?

The current documented command is **Effects → Noise Gate**. Sound Forge installations can also expose **coreFX Gate** through the plug-in list. The [current Sound Forge plan comparison](https://www.soundforge.com/sound-forge/pricing/) lists the coreFX Suite, while upgraded or older systems can have a different plug-in inventory. Use the processor that is actually present in your build rather than copying a menu path from another version.

The classic Noise Gate and coreFX Gate are not identical dialogs. The classic page points to its installed **?** help; the [coreFX documentation](https://cdn.borisfx.com/borisfx/Documentation/soundforge/2026/en/content/dynamic%20effects.htm) describes **Level, Attack and Release** plus shared detector and mix controls.

For the selection-based walkthrough here, choose **FX Plug-Ins → Apply Plug-In Chain**, add only **coreFX Gate** and leave it enabled. That route is documented in [Applying a Plug-In Chain](https://cdn.borisfx.com/borisfx/Documentation/soundforge/2026/en/content/apply%5Ffx%5Fchain.htm). Use Peak detection, fully wet Mix and neutral Out Gain for the first test. Keep the detector on internal input, with Listen off and no intentional detector filtering. Review Soft and any limiter in the preset so an unrelated setting does not confuse the test; this is a starting setup, not a factory-default claim. Changing to RMS or filtering the detector calls for a new threshold audition. If coreFX Gate is absent, follow the installed classic processor’s help instead.

## A seven-step Sound Forge noise gate workflow

The reliable setup uses two measurements and two listening checks. You need the loudest representative noise in a pause, the quietest wanted sound that must survive, the transition into that sound, and its ending.

1. **Protect the source file.** Save an untouched copy before adding effects. Select a representative test passage, open FX Plug-Ins → Apply Plug-In Chain and add coreFX Gate with the setup described above.
2. **Measure a noise-only pause.** Use Tools → Statistics on a pause with no wanted sound. Check both minimum and maximum sample excursions; in dB, the larger magnitude is the value nearer 0\. Use this as a reference, not a guaranteed gate setting.
3. **Find the quietest wanted sound.** Listen to the weakest consonant, breath, note or decay you need to keep. A whole word’s peak is not its quietest part.
4. **Audition Level inside the gap.** Start above the representative noise and lower than wanted details. If no setting keeps those details while quieting the pause, choose gentler expansion or another repair.
5. **Tune Attack and Release.** Lower Level when a beginning never triggers the gate; shorten Attack when opening is slow. Lengthen Release for chopped endings, then recheck the pause.
6. **Check chattering and transitions.** Try a lower Level and longer Release on the weakest passage. Check the start and end of the selected range and other sections; stop if useful speech is still being lost.
7. **Compare and commit once.** Preview with and without Bypass at the same listening volume. Turn Bypass off, confirm the time range and channels, then click OK in Apply Plug-In Chain. Undo a rendered test before processing an overlapping range. Save a new output and listen with no extra gate running.

## 1\. Protect the original before gating

Direct processing changes the selected audio; saving over the original can make that loss permanent. Keep an untouched source and test on a copy. In the current [View → Plug-In Chain window](https://cdn.borisfx.com/borisfx/Documentation/soundforge/2026/en/content/proonly/plug-in%5Fchainer.htm), effects instead run during playback and are applied on saving. That window is not the selection-processing dialog used here. The [Plug-In Chainer guide](https://soundforgepro.com/how-to-use-plug-in-chainer-sound-forge-pro/) explains the two routes.

For this test, leave the data window’s other playback effects bypassed so you hear coreFX Gate only once. Quiet syllables lost in a rendered pass cannot be restored merely by lowering the threshold afterwards; undo or return to the source.

## 2\. Measure a representative noise-only pause

Select a pause with the room noise you want to suppress, but no wanted breath, consonant, reverb tail or instrument decay. Choose **Tools → Statistics** and set **Level format** to **Decibels (dB)**. The [Statistics help](https://cdn.borisfx.com/borisfx/Documentation/soundforge/2026/en/content/statistics.htm) recommends the minimum and maximum samples in noisy silence as a gate reference. Compare both extrema: in dB, use the one nearer 0; with signed sample values, compare their absolute magnitudes. A negative waveform excursion can be larger than the positive one.

This is a sample-peak reference for the selected audio, not a prediction of every detector setting. Do not substitute whole-file RMS, integrated LUFS or true peak for the pause measurement. Confirm the result by listening, especially if the detector mode, filtering or processing before the gate changes.

Measure more than one pause when the background changes. A quiet pause near the start can’t stand in for a later section with a louder fan, traffic burst, or microphone movement. If the noise floor changes materially, process sections separately or fix the varying noise before relying on one gate setting.

## 3\. Find the quietest wanted word, note, or decay

Now loop the weakest wanted material in the same section. For speech that means soft consonants, sentence endings, and any deliberately quiet delivery; for instruments, note attacks and natural decays. The gate needs a usable level gap: the wanted sound must rise above the noise often enough to open cleanly.

![Noise gate Level set between a -48 dBFS noise peak and -30 dBFS quiet speech](https://storage.ghost.io/c/af/4b/af4b05d4-2e0b-4edc-8378-1e2a62fecc70/content/images/2026/08/noise-gate-threshold-desktop-v1.webp)

Measure both sides of the decision. In this example, -42 dBFS suppresses the -48 dBFS pause while leaving the -30 dBFS quiet word above the gate.

In the illustrated example, noisy silence peaks at -48 dBFS, the quietest wanted word reaches -30 dBFS, and Level is tested at -42 dBFS. The ordering matters more than the numbers: values closer to 0 are louder, so -30 dBFS passes while -48 dBFS falls below the gate. These figures describe a hypothetical recording, not a tested preset. A word peaking at -30 dBFS can still have consonants or tails below -42 dBFS. Listen through the entire word before accepting the threshold.

## 4\. Put Level inside the measured gap

Start with Level above the noise peak and safely below the quietest wanted sound, raising it only until the pause closes reliably. Then replay the weakest word or note. If that wanted sound disappears, lower Level; forcing the gate harder won't bring it back.

A narrow or unstable gap is a warning. When noise and quiet speech occupy the same level range, no threshold can distinguish them consistently. A higher setting removes more wanted audio; a lower setting lets more noise open the gate. That's a limitation of the source, and no secret number will get around it.

## 5\. Tune Attack and Release by the edge you hear

Attack controls how quickly the gate opens after the signal crosses Level. First check whether the missing consonant crosses that threshold at all: shorten Attack only when the gate is opening too slowly. If it never opens for that sound, lower Level and recheck the pause. A faster Attack cannot rescue audio that never triggers the gate. If opening clicks or snaps, try a slightly slower Attack and listen again.

**Release** controls how the gate closes after the signal falls below Level. Lengthen it if phrase endings, breaths, or room decay are cut off. If background noise hangs on too long after every phrase, shorten it carefully and recheck the quietest ending.

![Noise gate Attack and Release fixes for missing starts and chopped audio tails](https://storage.ghost.io/c/af/4b/af4b05d4-2e0b-4edc-8378-1e2a62fecc70/content/images/2026/08/noise-gate-attack-release-desktop-v1.webp)

Tune the edge you can hear: shorten Attack when starts disappear and lengthen Release when endings are cut off.

There is no attack-and-release pair that is correct for every voice or instrument. A useful first audition for spoken word is a fast opening and a release long enough to preserve a complete phrase ending. The waveform, delivery, and room decay decide the final timing.

## 6\. Stop chattering without sacrificing quiet words

Chattering is the rapid open-close sound produced when wanted audio keeps crossing Level. Lower Level first so quiet syllables stay above it, then lengthen Release so brief dips don't turn into abrupt closures. Test the weakest passage rather than the average one.

If chattering persists, recheck the detector setup and the passage before concluding that the recording cannot be gated. Other plug-ins may provide a separate Hold control, but the current coreFX Gate help does not document one. If no usable setting preserves the wanted sound, try gentler expansion or local edits, or address the underlying noise.

## 7\. Compare Bypass and apply only the verified range

Loop a weak wanted passage and a representative pause, then compare Bypass without changing monitoring volume. Check the first sound after silence, the end of a phrase, a breath, and any reverb or instrument decay. After that, move to at least two other parts of the file.

Before committing, turn the plug-in and dialog Bypass off, keep the gate enabled, and confirm the intended time range and channels. The [processing-dialog controls](https://cdn.borisfx.com/borisfx/Documentation/soundforge/2026/en/content/using%5Fprocessing%5Fdialogs.htm) also expose wet/dry gain and fades: an unintended dry blend can leave noise audible. Click **OK** in **Apply Plug-In Chain** only after the test works.

If you rendered a short test, undo it before applying to an overlapping longer range. For changing backgrounds, use separately checked selections and listen across their boundaries. Use [Save As for a new audio file, or Render As for project output](https://cdn.borisfx.com/borisfx/Documentation/soundforge/2026/en/content/saving%5Fa%5Ffile%5Fas%5Fa%5Fdifferent%5Fname%5For%5Ftype.htm). Open the output with no additional playback gate and listen through it, including quiet passages and fades. Keep the original.

## Why the noise comes back under speech

When speech crosses Level, the gate opens for the whole signal. The voice and the background recorded with it pass together, which is why a fan can disappear between sentences and return as soon as the speaker talks. A [recording-community discussion of vocal gating](https://www.reddit.com/r/recordingmusic/comments/1ilclb3/when%5Fusing%5Fa%5Fnoise%5Fgate%5Fi%5Fget%5Fa%5Flot%5Fof%5Fnoise/) describes the same practical symptom: gating the gaps does not remove noise under the voice. It’s a diagnostic clue rather than a product-specific defect.

![Noise Gate, Noise Reduction, and expander decision guide for background noise](https://storage.ghost.io/c/af/4b/af4b05d4-2e0b-4edc-8378-1e2a62fecc70/content/images/2026/08/noise-gate-tool-choice-desktop-v1.webp)

Choose by when the noise is audible: gate pauses, reduce noise that remains under speech, and use softer attenuation when dead silence sounds abrupt.

A gate fits when the unwanted sound is mainly a pause problem and a clear level gap exists. When a steady noise remains under speech, use the [Sound Forge background-noise workflow](https://soundforgepro.com/how-to-remove-background-noise-in-sound-forge-pro/). When the problem is tonal or frequency-specific, [hiss and hum diagnosis](https://soundforgepro.com/remove-hiss-hum-sound-forge/) is the right page. An expander or controlled room tone is usually more natural when complete silence makes every edit obvious.

If steady noise is distracting during speech, test denoising first, then decide whether the remaining pauses need any gating. Remeasure after the noise reduction: the old threshold may no longer fit. This is a useful starting order, not a requirement to run both tools. Compare each stage separately; if the result pumps, sounds watery or loses words, ease the offending stage rather than adding more processing.

## Frequently Asked Questions

### Where is the Noise Gate in Sound Forge Pro?

Choose Effects → Noise Gate for the documented processor. If coreFX Gate appears in your plug-in list, you can audition that version too. Installed plug-ins can differ between current, upgraded, and older systems, so verify the menu in your own build.

### What Level should I use for the Noise Gate in Sound Forge Pro?

With coreFX Gate using Peak detection, internal input and fully wet Mix, measure a noise-only pause and audition the weakest wanted material. Use the noise peak only as a starting reference for Level: a word’s peak does not prove its consonants and tails survive. If the ranges overlap, a hard gate cannot reliably separate them.

### What Attack and Release settings should I use?

Check Level as well as timing. If a quiet beginning never opens the gate, lower Level; if opening is slow, shorten Attack. Lengthen Release for chopped endings and recheck the pause. There is no universal millisecond preset that preserves every voice or instrument.

### Why does the gate chatter or stutter?

Repeated crossings of Level can make the gate chatter. Verify detector routing and the preset, then lower Level and lengthen Release while looping the weak passage. If no setting preserves it cleanly, try gentler expansion or a different repair process.

### Why does background noise return when I speak?

The gate opens for the complete signal when speech crosses Level, so the voice and its background pass together. Use noise reduction when the noise remains distracting under speech rather than raising the gate Level into quiet words.

### Does coreFX Gate have Hold, Floor, or Range controls?

The current Boris FX 2026 coreFX documentation describes Level, Attack, and Release for coreFX Gate. Other plug-ins may add Hold, Floor, Range, or Hysteresis, so rely on the controls visible in the processor you installed.

### Can I use Noise Gate and Noise Reduction together?

Yes, if they solve different problems. For steady noise under speech, try denoising first, then remeasure and audition a gate only if the pauses still need it. Compare each stage separately; running both is optional, and the old threshold may not suit the cleaned signal.

If the result still depends on an aggressive Level setting, return to diagnosis. The [audio restoration hub](https://soundforgepro.com/audio-restoration/) routes clicks, clipping, hum, hiss, and broadband noise to different workflows, while [Spectrum Analysis](https://soundforgepro.com/how-to-use-spectral-analysis-tool-sound-forge-pro/) helps confirm tonal problems before processing.

Documentation and the linked community symptom report checked September 17, 2026\. This is a document-verified Windows workflow, not a hands-on test of an installed Sound Forge build; available controls still need checking in your own plug-in.