High-resolution physical and psychoacoustic analysis of digital sound
“The crux of the question is:” Why constantly increase the sample rate in modern audio communication systems (spending huge amounts of money) if the thresholds of the auditory system are limited in frequency to the 20 Hz range. 20 kHz? ”
The analysis of the accumulated knowledge on this subject allows us to say that this is not enough. Given the complexity of the audio signal and the properties of the auditory system, it can be argued that only an increase in the resolution of the transmission systems in all areas (temporal, spectral, spatial and dynamic) can help solve this problem. At least now it seems clear that high resolution in the time domain is the most important for sound transparency.
As you know, to convert an analog (continuous) signal into a digital (discrete) signal, you need to perform the following operations: sampling, quantization, and encoding (Figure 1). For its implementation in all digital devices (computers, recorders, players, etc.), an ADC analog-to-digital converter (ADC) is used, the block diagram of which is shown in Figure 2. According to Kotelnikov’s theorem (Nyquist) or the “sampling theorem”, to convert an analog signal with the higher frequency f? (Hz) in digital without loss of information, it is necessary that the sampling frequency, that is, the number of samples (samples per second), is not less than 2 x f? (Hz). The digital word used, the number of binary digits in which it is equal to the number of M (bits) selected, represents the instantaneous value of the input signal,
Therefore, the sampling theorem requires that the sampling frequency be chosen high enough fd> 2fb, while the signal must remain almost constant at the time of sampling. The obligation to use a low pass filter is not specified, which is installed in all ADCs, but to avoid the appearance of excessive frequencies in the spectrum in all digital devices, there is an anti-aliasing filter that cuts the signal in the frequency fd / 2.
The recording of signals in any system begins with a microphone (Figure 3), which is a band-pass filter, which already has certain phase and transient distortions, leading to dispersion and blurring of the signal in the domain of the weather. Data on these distortions are rarely given in microphone catalogs, however, a large body of studies carried out in recent years has made it possible to establish a significant difference in these parameters between dynamic and condenser microphones. For condenser microphones, attack values of several microseconds were obtained, while the decay of transient processes reaches several hundred microseconds. The importance of the phase linearity of the microphones not only inside,
Then the analog signal, which is being converted to digital, is processed by a low-pass filter at the ADC (anti-aliasing filter) input. This filter also causes dispersion of the impulse characteristics of the input signal due to uneven frequency response and phase response in the pass band, the slope of the decay curves in the transition band, and the phase non-linearity.
Such distortions lead to time spreading of the input signal and mean that each instantaneous sample at the output will contain information elements from previous samples (the number of which depends on the characteristics of the filter). Since the musical signal is a rapidly changing current with short, sharp pulses, such scattering and blurring have a certain effect on auditory perception, especially for the experienced and attentive listener with a good musical ear.
Acoustic musical signals have a non-stationary ultra-fast dynamic and temporal structure, which is due to various reasons, in particular, a rapid attack on real musical instruments, the presence of a large number of ultrasonic components in the spectrum of many instruments, the appearance of short reverberation time delays in a room, etc.
Recording an actual reverb process without losing data is also extremely difficult. When a sound source emits a complex non-stationary musical signal, each microphone, installed at different points in the room, “picks up” the complex echo. Furthermore, additional incoming signals, altered in amplitude and phase due to reflections from various surfaces, lead to an exponential increase in the total energy level entering the microphone. When the signal is turned off, there is a drop in the overall level, which is usually characterized by the reverberation time.
