On records, hiss is caused by basic roughness of the sound carrier material and friction of the cutting-head in the sound groove. This hiss is recorded on the printing sample and it is copied onto every gramophone disc. Sound engineers tried to reduce this noise on the mother gramophone disc by using better basic materials then zinc or shellac. This better material was wax that was carried on a metal disc. Wax is very smooth and soft, so the cutting-head can work with small friction. Today nitro-lacquer on aluminum plate is used for this aim and the cutting-head is electrically heated. The cutting-head slides with very small friction in the melted lacquer, so no plus noise is added to the copies.
Studio tape devices are another noise sources. They are used for mixing the recordings. These devices were not used at old sound-recordings but I think it is an important subject and without it this part would not be complete. The non-continuous granulation of magnetic carrier materials on magnetic tapes and the non-continuous magnetization of tapes cause a fairly high hiss-level on the records. The problem of studio-tapes was solved only in 1970 by digital tape-recorders that have no hiss due to the digital technique.
The biggest problem with old gramophone records is the hiss produced during storing. Both the shellac and the plastic can dry out and the surface of the record becomes to dust. This process causes a strong hiss-type noise.
Hiss has a wide amplitude-spectrum. The spectrum is different on every record. Hiss can be modeled as an additive noise.
Cracklings and clicks
Crackle sound contains randomly distributed impulsive disturbances in the recording. The picture of a crackly sound part can be seen in figure 4. This degradation sounds like a "chip-fryer" noise. Small scratches, dust or fungi can cause this noise.
Figure 4. Picture of a 44.1 kHz sampled crackly sound part.
There is a fungus that eats the vegetable matter contained in old shellac records. It leaves millions of pock-marks on the surface of the disc, and these create impulsive disturbances in the replayed waveform, thus giving a crackly surface noise .
Another problem is the dust that sticks into the sound-grooves. Dust contains microscopic fibers (from textiles) and grains of sand . When the needle of the record-player reaches a grain, it produces a small impulse. The same will happen in case of a small scratch.
The shape of the sand-grains are not the same, therefore the shapes of the crackling disturbances are not the same either. These differences can be seen in figure 5. In this figure 40 crackling shapes can be seen. The crackling amplitudes are normalized to the fourth sample and the base-lines are removed. As we can see, only the starting impulse is similar. If we want to remove the crackles, we will have to take this fact into consideration.
Figure 5. The shape of 40 normalized cracklings. These cracklings were digitized
from a noisy record at 44.1 kHz sampling frequency.
Clicks are nearly the same than cracklings. The only difference is that
cracklings can be modeled as additive noise, but clicks are not. The amplitude
of clicks is so high that they will cause non-linearity and we cannot use
the superposition lemma.
Back to Table of contents - Back to previous part - Next part