The Common Methods for Noise Reduction

- Sep 20, 2018-

1. Noise gate:

Set a threshold of a level, the signal level below this threshold is all filtered out, and the signal level above the threshold is all passed (here the signal level refers to the sum of the signal and the noise level) ), this sound material with high signal-to-noise ratio does not need to be denoised.

2. Sampling and denoising method:

This is a method for professional audio processing software to effectively remove the continuous stable background noise. The principle of denoising is to sample the waveform samples of the noise, and then analyze the waveform of the entire material and sample noise samples to automatically remove the noise. The advantage of this noise removal is that the noise can be completely removed. The disadvantage is that it has a destructive effect on the original human voice quality, and the signal-to-noise ratio is more destructive. After this denoising, the sound of the metal is very strong. Therefore, this method is generally not recommended to denoise the singing voice. As far as possible, the previous recording control to a higher signal-to-noise ratio is the key to solving the problem.

3. Other methods of noise removal:

The frequency spectrum in which the noise is located is analyzed by the spectrum, and the signal level of the frequency band in which the noise is located is attenuated by adjusting the EQ. This signal to the original vocal frequency band is also attenuated, forming new problems, but also less.

4. EQ processing:

The spectrum distribution of the human voice source is quite special. In terms of its pronunciation, he has three parts: one is the tone produced by the vocal cord vibration, and the pronunciation of this part is flexible, different pitches and different pronunciation modes. The spectrum changes are also large; the second is the low-frequency arpeggio generated by nasal resonance, and the frequency equalization can roughly separate the three parts of the spectrum. The frequency segment for adjusting the nasal sound is below 500HZ, and the equalized midpoint frequency is generally 4 octaves in the 80-150HZ equalization bandwidth.

For example, you can set 100HZ as the midpoint of frequency equalization. The equilibrium curve should be gentle from 100--400HZ. The adjustment range of equalization gain can be +10dB~-6dB. Here you should be reminded that: when making adjustments for this project Monitor speakers should not use boxes with low frequency pronunciation to avoid the nasal sounds being unintentionally over-emphasized. The spectrum of vocal teeth is distributed above 4KHZ. Since this band contains some music and audio spectrum, it is recommended that the frequency band of the adjustment tooth tone should be 6~16KHZ, the equalization bandwidth is 3 octaves, the equalization midpoint frequency is generally 1/2 octave, and the equalization midpoint frequency is 6800HZ. Equalization processing, with a lower equalization gain, can be adjusted down to -10dB.



MAONO is an innovative designer and manufacturer of Lavalier, Podcasting, Wireless, Shotgun, Recording microphones and accessories for Smartphone, Camera and PC, etc.