SU493781A1 - A device for reproducing sound signals - Google Patents

Description

(one

The invention relates to the field of automation and computing.

Devices for recognizing and reproducing sound signals are known, comprising a multi-track recording unit, a loudspeaker with a low-frequency amplifier and a series-connected harmonic generator, an amplifier and a comb filter.

The purpose of the invention is to create a device for synthesizing and reproducing sound command combinations, providing high quality sound reproduction with high speed and reliability.

This is achieved by introducing a pulse generator, two frequency dividers, a function converter, a switch, a multiplier unit, and an adder. A first frequency divider connected to a pulse generator is connected to a second frequency divider connected to a functional converter, to a functional converter connected to a switch and a multi-track recording unit, and to a switch. A switch is connected to a multiplier, a multiplier is connected to a comb filter and an adder connected to a low-frequency amplifier.

FIG. 1 shows a block diagram of the proposed device; in fig. 2 - structure 2

turnar diagram of the functional converter.

It contains loudspeaker 1 (speaker), low frequency amplifier 2, adder 3, multiplying unit 4, comb filter 5, amplifier 6, 7 harmonic generator, switch 8, functional converter 9, dividers 10 and 11 frequencies with amplitude output, pulse generator 12 , block 13 multi-track recording.

Auxiliary equipment serving for learning (tuning and debugging) of a two-argument functional converter 9 includes the following elements:

microphone 14 with amplifier, comb-band narrowband filter 15, block 16 of amplitude detectors, block 17 of multi-track recording (magnetic drum or disc), comparing device 18, control device 19 and

alarm 20 end of the setting of the values of the scale factors.

Pulse generator 12 (Fig. 1) is connected to the input of the frequency divider 11, the output of the I-plot x (t) divider to the input of the divider 10

frequency, the first input of the converter 9 and the switch 8. The output of the frequency divider 10 (diagrams y (t) is connected to the second input of the converter 9, the converter outputs are connected to the input of the switch 8 and the multi-track magnetic recording unit 13. The outputs of the switch 8 (1Л) are connected to one of

the inputs of the cells (links) of the multiplying unit 4, their second inputs are with the outputs (, ..., ") of the filter 5. The input of the filter is connected to the output of amplifier 6, and the input of this amplifier is connected to the generator of 7 harmonics. The plots of the generated by him are shown on the graph U (t). The outputs of the cells (links) of the multiplying unit 4 are connected to the inputs (1, ..., n of the adder 3 connected to the low-frequency amplifier 2 loaded on, for example, a speaker.

In the proposed device, all command speech can be synthesized by combining the sounds of standard syllables, the number of which is limited. Each sound syllable is represented as a limited family of sound images, a family of sound images as a limited family of temporal functions, which are reproduced using an analog two-argument functional transducer.

In general, with the help of such a device, the following sequence of actions is carried out.

The individual standard syllables converted by the microphone 14 are spectrally divided into frequency bands and their envelopes are separated as a family of temporal functions, which are then simultaneously recorded (Fig. 2).

The two-argument functional converter 9 is separately trained in the family of functions of each syllable sound, and the scale factors (ao, ai, ..., ai) are separately recorded for each syllable (Fig. 2).

When reproducing the necessary sequence of syllable sounds, each temporal function of the family reproduced by the transducer is distributed to synthesize the sound images (Fig. 1).

First, we consider the process of learning a functional converter for given functions (determining the scale coefficients yo, ai, ..., un, and writing them into block 13 using auxiliary equipment).

The standard syllables are separately fed to the microphone 14, and then in the filter bank 15, the sound of each syllable is spectrally divided into corresponding narrow frequency bands. (The number of filters in the comb is chosen from the conditions for obtaining the minimum distortion at each frequency). The envelopes of the time functions of the syllables obtained after divided bandpass filtering and detection using block 16 are recorded in block 13 of a multi-track magnetic recording (drums, discs, etc.) or similar information storage device. At the same time, from the external generator, 12 series of pulses is fed to the input of the frequency divider 11 (see the diagrams of its output signal in FIG. 2). The output signal is then fed to the input of the converter. From the output of the divider frequency 10

(see the diagrams of its output signal in Fig. 2) the signal is fed to the inputs x of the converter.

Due to such a connection of frequency dividers and the form of their output voltages, the converter provides a search of the coordinates of the reproduced families of temporal functions, i.e., at each moment of time, due to the duration of the signal x (t), with a certain fixed value

(.; const) interrogates the coordinates of the families of temporary functions in terms of y, for another specific value of Xi const, again coordinates of the temporary functions are performed in terms of g /, etc. Transmitter output signals

arrive at the switch 8, which distributes them and delivers them to the device 18, where the signals previously recorded in block 13 are compared, which are also fed to the device 18.

If there are differences in the amplitudes of the compared signals (compared functions), the difference signal is fed to the control device 19, which is connected to the inputs of the scaling factors ao, ai, ..., and „converter. The control unit changes the scaling factors until the difference signal is reduced to zero. At this moment, the signaling device 20 of the end of the adjustment of the scale factor values is triggered, and the coefficient values are recorded in block 13. Next, the next standard syllable is pronounced in front of the microphone 14, which is also recorded in block 13 in the same way.

Thus, block 13 plays the role of a library of standard sounds (in fact, the amplitude values of the scale factors ao, ui, ..., un are stored in it). When reproducing human speech (voice) using a speaker, it is trained in the described manner. A two-argument functional converter with blocks of the structural scheme 10, 11, 13, 12, 8 is connected as indicated in FIG. 1. If it is necessary to reproduce this or that sound, the values of the scale factors ao, ui, ..., a "are entered from block 13 into the converter. Transform: the caller is polled on the inputs l: and g / due to the fact that the signals from the dividers 10 and 11 frequencies

arrive at the coordinate inputs xn of the functional converter. Since the values of the argument x correspond to one frame of the sound image, and the value of the argument y corresponds to the points of the curves of the family, the scan

provides simultaneous reproduction of all curves of one family of temporary functions on the divided outputs.

The signal from the output of the functional converter is selectively distributed by the switch 8 to the elementary cells (links) of the multiplying unit 4. The signal of a certain harmonic carrier arrives at the other inputs of the cells (links) from the outputs of filter 5. The harmonic generator 7 generates an asymmetric form voltage, which is amplified by the amplifier 6 and fed to the filter 5. From the filter outputs (1, ..., l), the voltage of each harmonic passes to one of the inputs of the cells (links) of the 4 as a carrier oscillation. Consequently, in block 4 amplitude modulation takes place. The amplitude-modulated signal from block 4 is fed to adder 3, then amplified by low-frequency amplifier 2 and reproduced by loudspeaker 1.

Subject invention

A device for reproducing sound signals comprising a multi-track recording unit, a low-frequency amplifier speaker and connected in series

a harmonic generator, an amplifier and a heating filter, characterized in that, in order to improve the quality of sound reproduction, it contains a pulse generator, two frequency dividers, a functional converter, a switch, a multiplier unit and an adder, the first frequency divider connected to a pulse generator connected to the second frequency divider connected to the functional converter, to the functional converter connected to the switch and the multi-track recording unit, and to the switch, connected nnomu a multiplier unit connected

to comb filter and adder connected to a low frequency amplifier

17

(I am ulll

Ij:

15