DeFilter

Ultimate Edition Only

This feature is available in the Ultimate Edition only.

About DeFilter

DeFilter is a powerful tool that can be used to automatically address problems in a mix or to remove unwanted resonances, comb filter effects or other tonal imbalances.

In order to accomplish this, DeFilter needs to analyze the audio recording before the actual correction phase. You can do this by activating the learn mode using the corresponding button in the user interface and play a part of the recording for analysis. 15 to 30 seconds of audio is usually enough to obtain good results. We recommend choosing a 'busy' part of the recording with sharp attacks and quick signal level changes since these are helpful for the analysis.

When the analysis is done, you should disable the learn mode and leave it disabled for the correction phase. DeFilter can match the frequency spectrum to predefined profiles such as music and speech or to a measured profile (matched equalization). DeFilter can also be set up to remove local peaks and dips in the spectrum while maintaining the overall timbre, which can be very effective when removing sharp resonances or undesired comb filters.

The correction process can be monitored visually using the spectrum analyzer showing and the visualization of the correction filter gains.

Parameter settings

Filter estimation

The Filter estimation group contains controls that affect the estimation of the input signal's long term frequency spectrum. The input signal is only estimated when the learn button is activated and the estimation will normally be more accurate when the learning time increases. We recommend to use the learn mode for at least 15 seconds on a busy part of the recording.

The learn mode should always be disabled when applying the correction filter.

• Learn button

Enables or disables the estimation of the input signal's frequency spectrum.

• Save button

You can save the estimated long term frequency spectrum to a file. This is essential if you want to match the frequency spectrum of a recording to that of a reference recording. To do so, analyze the reference recording, click the save button and choose a name and folder for the file. Now you can use the Load button in the Target EQ profile group to load the reference file, explained below. Next, you can open the file you want to apply match equalization to and use the learn button to analyze. When done, deactivate the learn mode and apply the saved correction filter.

• Signal level threshold (dB)

The signal level threshold sets the minimum input signal level required for the analysis. Real recordings frequently contain noise sources such as noise from electronic components that have a very different frequency spectrum than the wanted signal.

To prevent these from corrupting the estimation, the threshold level should be set considerably higher than the noise level of the recording.

• Filter length (ms)

The filter length parameter sets the length of the correction filter. Longer filters will be able to capture finer details in the frequency spectrum, but require a longer learning time and increase the CPU consumption during analysis.

Target EQ profile

The target EQ profile group lets you control the properties of the desired target frequency spectrum. You can choose among predefined profiles or load measurements from reference recordings.

The drop-down list contains predefined options:

• Automatic

Automatic mode extracts the smooth frequency spectrum curve from the source so that the overall timbre is preserved. However, narrow peaks and dips will be effectively removed.

• Music

The music profile matches the frequency spectrum to suit music. Some post processing with your favorite equalizer might be required, but the music profile should be a great starting point.

• Music (dark)

This is a slightly darker sounding profile for music.

• Music (less harsh)

Profile intended for music with a slight decrease in levels where the ears are most sensitive, hence reducing the perceived harshness.

• Speech

The speech profile matches the frequency spectrum to a frequency distribution suitable for voice recordings.

• Custom

The custom mode uses measurements of reference signal and is only available when you have loaded a measurement using the load button.

Load button

You can use this button to load a measurement of a reference file. See the Save button section in Filter estimation above for more information.

Global Parameters

• Enable emphasis curve

The emphasis curve allows frequency dependent control over the level of correction ranging from 0% (no correction) to 100% (full correction). The emphasis curve allows you focus the correction on problematic regions in the frequency spectrum. Click the Enable emphasis curve toggle button to enable the emphasis curve.

You can add or remove curve points by double clicking the left mouse button. Click and hold down the left mouse button to move curve points.

• Enable spectrum analyzer

The spectrum analyzer enables the visualization of the short term input and output spectra and helps monitoring the effect of the correction filter.

• Master Gain (dB)

The Master Gain knob is located at the bottom right corner of the Defilter interface. If the correction filters alters the perceived loudness of the signal, you can adjust the output gain in Decibel using this knob.

Using DeFilter for Analysis

Even in situation where you want to use a conventional equalizer for the actual processing, DeFilter can be a very helpful companion for analysis purposes. The correction curve helps identify peaks or dips that can be cumbersome to find manually.

The center frequency and gain values are easier to read out when using the crosshair cursor mode. Activate the crosshair cursor by pressing and holding down the Alt (Windows) or Option (Mac) key on your keyboard while hovering over the curve display. DeFilter displays the exact frequency and gain values as shown in the screenshot below: