NoNoise Tools
Audio artifacts come in many types, and every restoration job is unique. NoNOISE consists of a group of linked tools for analysis and processing. In using NoNOISE, you will apply your listening skills and knowledge to determine how the individual tools may be applied to achieve the desired results.
FFT Analysis
The Frequency Analysis (Fourier transform, or FFT) is a primary tool of analysis for NoNOISE projects. The Fourier transform converts time domain information into the frequency domain where it can be used to pinpoint the center frequency of a hum component or to examine noise and signal bandwidth.
Manual Declicking
In Manual declicking, you identify audio clicks and glitches by listening, then locating them precisely using the Sonic System's graphic waveform display.
Once a click or glitch is identified and marked, you can use any one of several interpolation algorithms to resynthesize and replace damaged audio with a seamless reconstruction of the original program.
Manual declicking is most useful when source material contains a relatively small number of prominent clicks or glitches.
Complex Filtering
You can attack discretely pitched and steady-state noises such as hum and buzz using filtering techniques. Complex filtering applies up to 512 high-density digital filters to surgically remove the fundamental and all harmonics of a contaminating noise.
Complex filtering can be used to clean up production dialog, live concert recordings, telephone conversations and other situations with complex pitched or band-localized noise components.
The system performs the designated processing out of real-time, creating a new sound file as output. Processing is performed in the background, so that you can process and edit simultaneously.
Production Declicking
Analog disc records contain far too many scratches and clicks to identify and correct by hand. Production Declicking automates the process, correcting hundreds or even thousands of clicks in a single pass.
Declicking is a two-stage process. First, use the Click Detect tool to identify the locations of clicks and scratches, with variable parameters that determine how the system will distinguish clicks and scratches from program audio. This process produces a Click List that is used for actual click removal. After click detection, the source file waveform is displayed with colored bars to indicate click sites. You may add, change, or remove these sites to ensure proper removal of all audio glitches.
Once the Click List is prepared, a declick pass is performed to correct the marked areas. As in the detect pass, you specify the parameters that determine how the system replaces detected clicks with new audio.
Decrackling
In many cases impulsive noises are so dense that they are no longer heard as individual clicks. These high-density impulses are perceived as surface scratch, or crackle. The Decrackling tool removes this high-density impulse noise.
Decrackling runs as a single-stage (no detect pass) background process that produces a new sound file, leaving the original file untouched. As with other NoNOISE processes, the operator controls the degree and quality of decrackling applied.
The Decrackler can also be used to correct many types of peak distortion. By specifying an Amplitude Weighting factor, you can direct the decrackle process to concentrate waveform reconstruction on signal peaks, using undistorted sections between peaks to provide the spectral information required for reconstruction.
Broadband Denoising
Denoising begins with analysis of the background noise to produce a Noise Estimate indicating noise level versus frequency. This estimate is used to set up the denoiser, specifying a threshold for downward expansion in each of over 2,000 individual frequency bins. The Denoiser then analyzes the source audio and reduces gain only at those frequencies where the signal level falls below a designated threshold between signal and noise floor in that band.
Denoising can run in real time or as a background process. The advantage of real time processing is that it provides instant feedback on parameter settings. You may wish to use real-time processing to determine optimal parameter settings, and then use the non-real-time processing to create an output file.