Another common problem is the presence of a continuous background noise on a recording. One example is the background hiss on a recording coming from a magnetic tape. Another example is the 50Hz or 60Hz hum coming from the electric grid. (This is commonly picked up by microphones that are somehow connected to a device that runs on a power supply and not on battery.)
Warning: However good a software package is, it cannot resurrect what is lost. You should therefore always try first to get the best possible quality by choosing appropriate recording equipment and only use software-based restoration functions as a matter of last resort. If a recording is important to you and cannot be repeated, try to make tests in similar conditions first.
The Denoising function in the Effects menu provides a very efficient algorithm for suppressing background noise. The easiest way to use it is the following:
Sometimes, there is no piece of sound containing only noise available. In this case, you may want use the Suppress White Noise function. The easiest way to proceed is to use the Preview function to play the selected sound and then to adjust the Noise RMS power with the slider provided until the result is satisfactory. Note that choosing too large a value of the Noise RMS power will result in a significant loss of quality in the clean signal, whereas choosing a too small value will leave some background hiss remaining.
It may happen sometimes that the background noise one wishes to remove is very well localised in frequency (like the case of the hum discussed above). In this case, it is advisable to use the Suppress Frequency Band function to remove every frequency localised between the Lower Frequency and the Upper Frequency.
Finally, the Suppress DC Offset function allows you to remove a low-frequency drift in the microphone (this happens for example on the integrated microphone of some of the older PowerBook models).
The Denoising function is customizeable with the Settings menu item. The Noise Type allows you to choose the type of noise you want to remove. Choosing a peaked noise is suited for background noise which is very well localised in frequency space. A typical example is the removal of the 50Hz or 60Hz humming produced by the power supply. The smooth noise function is adapted to the removal of noise that is spread over the whole frequency range, like for example a tape hiss. If the Adaptive Filtering option is disabled, the program will compute one global filter and apply this filter to the whole sound. This sometimes yields better results when the signal-to-noise ratio is very low. The Sensitivity Enhancement option tells the program the amount of background noise to remove. Usually, values between 30% and 70% yield the best results. For lower values, some background noise may remain after the application of the filter. For higher values, the signal may be substantially altered.
The Algorithm pop-up menu allows you to choose which algorithm should be used for the denoising. Currently, two algorithms are available: Short FFT and Long FFT. Generally, the Long FFT algorithm produces better results, but the Short FFT algorithm is slightly faster.