Noise Reduction Manual

When using a very large fft size the attack speed

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: EDUCTION 26 Release speed This control determines how quickly the noise reduction algorithm reacts to a noisy signal. To change this setting, you may click and drag the slider left or right, or type a value into the edit box. The default value of 50 is recommended. Higher values can cause the algorithm to trim the ends of long decaying sounds. When using a very large FFT size, the Release speed should be higher. High-shelf start frequency This control determines the frequency above which a gain or cut is applied to compensate or enhance the signal after the noise reduction process. High-shelf gain This control determines the amount of high-frequency boost or cut applied to the signal after noise reduction. A slight high-frequency boost (+3 dB) often compensates for highfrequency components that might get lost during the noise reduction process. Windowing area This area has two controls that allow you to measure frequency analysis: FFT size and Overlap. The FFT size control determines the size (in samples) of the analysis window and the number of discrete frequencies used during noise reduction. The default value of 2,048 is recommended for most cases. Higher numbers yield more frequency resolution at the expense of longer processing time. After selecting a new parameter, you should capture a new noiseprint. NOISE REDUCTION CHP. 2 27 The Overlap control determines the amount of overlap between FFT analysis windows (67% to 90%). A lower setting decreases the number of distinct analysis functions performed, which decreases processing time. High settings allow for more analysis, which results in extremely slow processing. Raising the overlap factor above 75% is often sonically unnoticeable, so we recommended that the default of 75% be used. Capture noiseprint Selecting this check box puts Noise Reduction into Capture mode. In this mode, the plug-in analyzes the data for frequency content rather than applying noise reduction. Use this mode to gather the initial noiseprint. You may capture a noiseprint during playback at any time, but since the audio that you are hearing might not be completely synchronized with what is being analyzed, you should not use this method when the segment of noise-only material is small. It is much safer to either make a small audio selection beforehand and to use the start of the file and the noise source. When not making a small selection beforehand, using the Automatic capture timeout control helps to set the analysis endpoint, otherwise you must uncheck the Capture noiseprint check box yourself to begin processing. The Capture noiseprint check box is automatically unchecked after processing the whole audio selection. Note: The Capture noiseprint is duplicated on the General and Noiseprint pages for convenience. CHP. 2 NOISE REDUCTION 28 Automatic capture timeout (.005 to 15 sec.) Checking this box enables the Automatic capture timeout feature when in Capture noiseprint mode. When enabled, the plug-in will switch out of Capture mode after the specified amount of time. This is useful when you only want to analyze a certain amount of time after starting a preview. To use this control, make sure that there is a segment at the start of the audio selection containing only noise. For example, if you have only 0.75 seconds of noise at the start of the file, set the Timeout to 0.75 seconds. This will prevent the algorithm from using non-noisy material as the noiseprint. Note: This control is duplicated on the General and Noiseprint pages for convenience. Keep residual output When this check box is checked, the detected noise is the output instead of the de-noised signal. This check box allows you to listen to the audio material that is removed during normal processing. This feature can be very helpful in determining how much of the good signal is being affected by your settings. Turn this off when you want to remove the noise from the signal. Noiseprint page The Noiseprint page contains controls for manipulating and navigating the noiseprint. In most cases, these controls will not need to be adjusted. Noiseprint calculation mode This control determines if the noiseprint graph displays the peak value or the average value of the selected noise. NOISE REDUCTION CHP. 2 29 Noiseprint graph This graph displays amplitude versus frequency of the analyzed noise. It contains the noiseprint envelope, which can be manually adjusted for specific applications. Amplitude Frequency Noiseprint envelope Envelope points are used to model the noiseprint and are editable. Envelope points Noiseprint The envelope, not the noiseprint graph, is used in the algorithm when separating the noise signal from the source; any signals below the envelope are treated as noise. Therefore, the envelope contour should be above the noiseprint graph if the desired frequency noise is to be removed. To move all the envelope points up or down, use the Envelope Amplitude control. For more information, see Manually adjusting envelope points on page 19. CHP. 2 NOISE REDUCTION 30 Envelope amplitude Use the...
View Full Document

This note was uploaded on 05/15/2013 for the course EMUS 201 taught by Professor Pardal during the Winter '10 term at Life.

Ask a homework question - tutors are online