Intelligence enhancing equipment.

Intelligibility enhancement equipment.

Intelligibility enhancement equipment.

1. General Provisions.

During the process of receiving and recording audio information, interference and noise may occur due to various reasons: street noise, rain, noise and interference from various electrical household and industrial devices, music, etc.

The main means for cleaning audio information from the above-mentioned noise and interference are adaptive filtering devices, as well as hardware and software systems, the operation of which is based on the possibility of implementing an adaptive filtering algorithm.

The block diagram of the adaptive filter is shown in Fig. 1.

The adaptive filter operates on the principle of estimating the statistical parameters of the incoming signal and adjusting its own transient response in such a way as to minimize the error e(n).

If we assume that a mixture of a speech signal and interference is received at the input Y(n), and an interference signal is received at the input S(n), then the parameters of the programmable filter will change in accordance with the adaptation algorithm in such a way that at the output Yf(n) there will be a signal approximately equal to the interference signal, and at the output e(n) there will be a purified speech signal.

Devices implementing the block diagram in Fig. 1 are called two-channel adaptive filters or adaptive interference compensators.

If a signal from the setting input is fed to the input of the programmable filter, i.e. the condition S(n)=Y(n) is met, then this implementation of the block diagram is called a single-channel adaptive filter, or simply an adaptive filter.

The adaptation algorithm determines the law according to which the programmable filter is adjusted, where the parameter is the value that determines the rate of change of the parameters of the programmable filter.

Two tuning laws are usually used:

• minimum mean square error (MMSE);
• minimum error modulus (MEM).

The parameter that determines the tuning speed ensures precise adjustment of the adaptive filter to a particular interference, therefore, to effectively combat interference, it is desirable that the devices have the ability to change this parameter within a fairly wide range.

Finite impulse response filters (FIR filters, in the English version) are often used as a programmable filter. The size of such a filter determines the number of adaptation coefficients.

Since all hardware implementations of the adaptive filter without exception are based on digital signal processing methods, the characteristics of the ADC-DAC path (bit depth, nonlinear distortion, dynamic range, operating frequency band) have a great influence on the quality of processing.

2. Comparative analysis of adaptive filtering devices.

Table 1 shows the characteristics of adaptive filtering devices (hereinafter, all devices using an adaptive algorithm for processing will be called adaptive filters for brevity) produced both in our country and abroad.

It can be seen that there is a fairly wide range of devices of this class, allowing to solve various problems of interference and noise control.

Adaptive filtering equipment can be divided into two groups:

• professional equipment, designed for trained users, who are mainly employees of law enforcement agencies.
• equipment for commercial use.

Table 1. Adaptive filters.

Professional equipment is characterized by a wide range of functional capabilities, the main ones of which are listed below:

• the presence of both single-channel and dual-channel processing modes;
• the ability to set various filtering modes;
• the ability to change within wide limits both the number of adaptation coefficients and the working processing band;
• the ability to work with a large number of coefficients (more than 1024);
• availability of various service capabilities.

Commercial equipment is designed for untrained or poorly trained users and has significantly fewer functional capabilities, but is fairly easy to operate, does not require special knowledge, and also has a significantly lower cost (several times) than professional equipment.

Table 1 (continued). Adaptive filters.

n/a — no data;

* — this device contains two cascades of adaptive filters. The first has a maximum number of coefficients of 4096, and the second — 512.

1. Professional equipment.

The following devices can be classified in this group:

• “Cinderella-31”, CRT, St. Petersburg, Russia:
• “MicroDac II,” “MicroDac III,” “DAC32000,” DAC, USA.

The optimal choice for Russian users is the “Cinderella-31” equipment, which is not inferior in technical and functional characteristics to foreign analogues from DAC, but is sold at significantly lower prices.

The characteristics given in Table 1 show that in such essential technical parameters as suppression of sinusoidal signal and white noise, the “Cinderella-31” equipment is almost as good as the “MicroDac II,” “MicroDac III” equipment from DAC, but at the same time significantly surpasses it in terms of service capabilities (pseudo-stereo, ring repeater, PC control capability.

Notes:

• The pseudo-stereo mode is designed to improve comfort and intelligibility when listening to monophonic signals;
• The ring repeater provides the ability to listen to a small section of a soundtrack multiple times (4-10 sec. duration).

It should also be noted that the front panel of the “Cinderella-31” equipment has a liquid crystal display that displays various information (signal level, processing parameter values, etc.), significantly simplifying the operator’s work.

Another device that I would like to draw the attention of professional users to is the DAC32000.

With this device, it is possible to eliminate a huge number of various interferences and noises that arise when working with sound, acoustic or vibration signals.

DAC32000 has 8 input channels and 6 output channels, which provides up to 16 different processing modes. The maximum number of adaptation factors provided by this device in the working band of 8 kHz is 32768.

The device has a liquid crystal display to display the input signal levels and the parameter values ​​of various processing modes.

2. Commercial equipment.

This group includes the following devices:

“Filter-N”, “AF-512”, “Cinderella-Micro 2”, “MicroDac”.

A characteristic feature of these devices is their single-channel nature (there is no reference channel), which does not allow them to effectively combat non-stationary broadband interference such as white noise (an example of white noise is the noise of a tape during a pause).

Nevertheless, it is possible to achieve quite noticeable suppression of stationary interference (network, harmonic, “smooth” music, etc.).

The best representative in this group (in terms of price/quality ratio) is the “Cinderella-Micro 2” equipment from the TsRT company, which, at a significantly lower price than its analogue “MicroDac” from the DAC company, provides not only practically the same technical characteristics, but has additional functional capabilities — a ring repeater for 11 seconds and AGC for input and output.

A characteristic feature of the above-mentioned products is the ability to operate from a battery and low power consumption, which is optimal for using this equipment, for example, in a car.

Both “MicroDac” and “Cinderella-Micro 2” are distinguished by a minimum number of controls, which allows even a non-professional user to successfully use this equipment.

The “AF-512” equipment is somewhat inferior in technical parameters to the devices described above (see Table 1), but has the lowest price among the equipment in this group.

3. Hardware and software systems.

Hardware and software systems (HSS) are a powerful tool in the fight against interference and noise, since they use the computing and interface capabilities of the PC on the basis of which they are created.

A distinctive feature of the APCS is the presence of advanced software. Table 2 presents the main technical and functional parameters of the hardware and software systems.

Table 2. Hardware and software systems.

n/a — no data;

* — see further in the text;

** — with the cost of a PC.

As a rule, the APC includes (see table) the following components:

• an input/output unit that ensures coordination with external signal sources. Usually implemented as a separate device, connected to the processing board by a connecting cable.
• a digital signal processing board installed in a PC slot, ensuring both direct processing and data transmission to the PC.
• A personal computer that displays and stores the results of processing.
• software that allows the operator to select various modes and methods of signal processing, as well as to listen, observe spectrograms, etc.

Since the APC is intended for professional users, the main requirement for the APC is the ability to implement an adaptive filtering algorithm in conjunction with other algorithms (low-pass filter, high-pass filter, pulse interference suppressor, equalizer, etc.) in real time.

Let's take a closer look at the APKs presented in Table 2.

1. “Fair”. Developer and manufacturer: FSB of Russia.

The unique software, operating in the MS DOS environment, in addition to the capabilities listed in Table 2, also allows you to:

• adjust the recording level and volume of the signal being listened to;
• process speech signals and simultaneously output the processed signal to headphones and analog outputs in mono, pseudo-stereo and stereo modes;
• save filter parameters on the PC disk drive and load them into the signal processing board memory;
• display spectra and time realizations of processed signals on the screen;
• compile transcripts of processed audio recordings using the built-in text editor.

High technical characteristics, a wide choice of processing algorithms, a flexible and convenient software interface indicate that among the presented APCS “Yarmarka” is the most successful choice for users professionally working in the field of increasing the intelligibility of noisy speech signals.

The disadvantages of this equipment include the lack of software that works in the Windows 95 environment.

2. “IKAR-1K”. manufacturer — the company “CRT”.

In its technical and functional characteristics, it is close to the APCS “Yarmarka”, but is designed to solve a wider range of problems:

• bioacoustics;
• linguistic research;
• hearing research;
• voice and speech pathology;
• voice speaker identification and verification;
• diagnostics of a person’s psychophysiological state by voice;
• increasing intelligibility.

The software includes the following components:

• SIS 4.5 sound recording analysis and processing package;
• real-time adaptive digital filtering program;
• program of multi-band (up to 2000 bands) equalizer and spectrum analyzer of real time;
• program of operational documentation of sound recordings “Caesar-SOFT”.

Like all multifunctional systems covering a wide range of tasks, this complex is not without shortcomings associated with the cumbersomeness of the interface, which does not allow the operator to quickly set the desired processing mode, as well as to see the results of processing.

3. PCDF4096. Manufacturer — DAC.

Technical specifications and functionality are close to those of the above-described APCS.

The software includes several packages:

• DFWS(Digital Filter Work Station) — a workstation for a digital filter designer. Allows you to implement various digital filters in real time, including adaptive ones (both single and dual-channel).
• PC-EDGE (Enhanced Digital Graphic Equalizer) — an improved digital graphic equalizer. Allows you to implement an equalizer with any user-defined amplitude-frequency characteristic.
• PC-SIF (Spectral Inverse Filter) — implementation of a filter inverse to the signal spectrum.

It should be noted that the PCDF4096 equipment, without having significant advantages over domestic APCS in terms of technical, functional and software capabilities, however, has a significantly higher cost.