RU2602976C2 - Method of producing artificial intermediary language for communication between human and dolphin - Google Patents

Abstract

FIELD: acoustics.

SUBSTANCE: invention relates to methods of a bank of acoustic signals and can be used for studying communication between a dolphin and a person. Method of generating a bank of acoustic signals for communication between a dolphin and person in frequency range typical for communication signals of dolphins, comprises generating a bank of reference acoustic signals from separate sounds of an even-tempered sequence of sounds by formatting sounds on same duration and amplitude, recording on a recording medium and measuring their frequency, amplitude and time characteristics. From the recorded sounds, generating reference sounds by changing their duration and/or amplitude and generating therefrom a bank of reference sounds with measurement of characteristics of each reference sound. Generating reference acoustic signals using combinations of obtained reference sounds and measuring frequency, amplitude, time modulations of each generated signal. Each signal is assigned a unique identification code and information content with subsequent comparison of sounds detected by means of acoustic receivers with reference signals of bank.

EFFECT: high accuracy of recognition of acoustic signals.

1 cl, 1 dwg

Description

The invention relates to communication systems using acoustic signals, and in particular to methods for creating a bank of acoustic signals, and can be used for research in the field of messaging between dolphins and humans.

Numerous studies in the field of physiology, behavior and intelligence of dolphins show that dolphins have a pronounced personality, can represent their future actions and can be trained (https://www.opendemocracy.net/margi-prideaux/dolphins-as-persons).

It is also known that dolphins are capable of responding to various combinations of sound symbols with certain actions, as a result of training, they can proactively use code signals for a number of requests for human actions, and the goal of further experiments was to establish two-way communication between a person and a dolphin using sound symbols. (NL Krushinskaya and T.Yu. Lisitsina, “The Behavior of Marine Mammals.” M.: Nauka, 1983). A number of scientists provide evidence that dolphins, living in the world of sounds, not only use the universal “sound-graphic” language to communicate with each other, but also think in sound images (https://www.opendemocracy.net/margi-prideaux/ dolphins-as-persons).

Communicative sounds of bottlenose dolphins, including whistles, “twitter”, “trills” are marked in the frequency range from 2 kHz to about 20 kHz, they also give alarms and needs for help at lower frequencies (V.V.Kaznadzey, S .A. Kreichi, EN Khakhalkina. Types of dolphin communicative signals and the nature of their organization. Acoustic Journal. Vol. 22, Issue 6, 1976).

The structure of communication signals of dolphins is a complex sound systems with frequency and amplitude modulation. The pronunciation of individual sounds and their combination in the "speech" signals of dolphins is much faster than that of any person. Conducted back in the 60s of the twentieth century, the experiments of J. Lilly (Lilly) revealed developed "vocal" imitative abilities in dolphins. Dolphins make some sounds only after prolonged contact with a person, and they acquire some resemblance to the trainer's voice. A comparison of vibrograms with recording voices of dolphins and humans has confirmed this. If the voice of the trainer sounded with a frequency from 200 to 3000 Hz, then the voice of a dolphin mimicking the trainer had a frequency from 1000 to 8000 Hz (http://www.delfinary.ru/bibl.html).

It is extremely difficult to understand the information content of communication sounds made by dolphins and their combinations. The study of the informational content of dolphin sounds took decades of work by scientists from many institutes and universities, including the Institute for the Search for Extraterrestrial Intelligence (SETI) at Mountain View, the University of California at Davis (https://www.opendemocracy.net/margi-prideaux/dolphins -as-persons).

One way to establish the exchange of human signals with a dolphin is to teach dolphins to exchange messages in the form of commands or signals using technical means.

A known method of forming a set of acoustic signals (signal bank) for exchanging messages between a person and a dolphin using an underwater keyboard sound instrument (project "Wild Dolphin" - Wild Dolphin Project, WDP), (http://www.wilddolphinproject.org/research/dolphin -comrnunication /, http://www.zoopicture.ru/obshhij-vazyk-s-delfmami/), the essence of which is the correlation of a sound acoustic signal with a subject symbol. To do this, use a special underwater keyboard, which consists of ball keys with special graphic symbols. The keyboard, in turn, is connected to a signal system in which each item is associated with a whistle of various keys. When a dolphin makes a sound that matches the tone of a certain key, the researcher gives him the item that the dolphin required. The method of creating a set (bank) of acoustic signals, based on the correlation of the symbol of the object with a specific acoustic signal, adopted as the closest analogue of the invention.

However, it should be noted that to create a bank from graphic or subject symbols and corresponding acoustic signals, you will need a keyboard with a very large number of keys equipped with graphic or subject symbols, and for a dolphin to recognize and remember each symbol using its sonar unit, they must be to form, varying in shape and acoustic properties of the material. In addition, the use of subject symbols is possible only at a short distance, where the dolphin's sonar apparatus is able to distinguish their features. The consequence of this method of forming a bank will be a complex and large set of symbol objects, which will lead to the problem of their identification and standardization during replication. The disadvantage of this method is the use for signal exchange of a keyboard (tactile) instrument, which requires more time for a dolphin to respond to a semantic message than a reaction in an acoustic form. In addition, the overall accuracy of signal recognition is reduced due to the need not only to recognize the acoustic signal, but also an additional procedure for recognition of objects-symbols on the keyboard to supply a response signal.

At the same time, one of the well-recognized by a person acoustic signals are musical sounds and their combinations. It is known that in music, each note in each octave corresponds to a certain and constant frequency of sound vibrations of the main (tonal) sound. Depending on the type of source, the reproduced main sound is saturated with overtones, which determine the "beauty" of the sound of the source. All musical instruments and all voices of people are distinguished by the individuality of the overtones of pronounced sounds, similar to the individuality of skin drawings of human fingers. A human auditory analyzer allows you to recognize a musical instrument by its sound and recognize an invisible acquaintance from the timbre of its voice. Music allows the exchange of sensory-emotional perceptions, but does not carry specific utilitarian information. Combinations of musical sounds performed, for example, on a trumpet, are still used in the army to give commands.

The objective of the invention is to create a bank of acoustic signals for two-way exchange of signals, which are based on the musical sounds of a uniformly-tempered scale.

EFFECT: increased accuracy of recognition of acoustic signals, unlimited filling possibilities.

The problem is solved by creating a bank of acoustic signals in the frequency domain characteristic of communication signals of dolphins, including at the first stage creating a bank of reference acoustic signals from individual sounds of a uniformly-tempered scale by formatting sounds of the same duration and amplitude, recording to a carrier and fixing their frequency , amplitude and temporal characteristics, then from the recorded sounds by changing their duration and / or amplitude, reference sounds are formed and create a bank of reference sounds from them with fixing the characteristics of each reference sound, and then using the combinations of the received reference sounds form reference acoustic signals, fix the frequency, amplitude, time modulations of each generated signal and each of them is assigned a unique identification code and information content with subsequent comparison of the recorded using acoustic sound receivers with reference bank signals.

Significant differences of the proposed method is the formation of a replenished bank of acoustic signals, which are based on individual sounds of a uniformly-tempered scale, which allows combining sounds to obtain a variety of acoustic signals for exchange and use their electronic copies to analyze the acoustic activity of the trained dolphin.

The method allows to standardize the process of generating acoustic signals and uniquely reproduce them, since these signals are characterized by a specific frequency, their syllabic name and musical notation are widely known and international, they can be reproduced using technical means — sound generators or sound cards connected to computers and working under the control of special programs - sound editors, and musical instruments and voice.

In the process of exchange, the signals are emitted, and the received signals (response) are compared with the reference signals contained in the bank, and the acoustic images of the signals are compared that differ from the original ones due to possible frequency distortion, the nature of the timbre and time parameters in the response, which makes it possible to unambiguously determine the content of the response when coincidence.

The emission and reception of an acoustic signal when creating a bank of signals can be carried out using known equipment, for example, broadband acoustic transducers, both underwater and in air.

The fixation of individual sounds and received acoustic signals in the form, for example, of their alphanumeric designation, spectrogram or musical notation, provides a quick perception and understanding by a person of the parameters recorded on the signal carrier, as well as the possibility of identical reproduction of acoustic signals by various technical means, including musical instruments , or even a voice. In addition, a comparative analysis of acoustic signals in graphical form can be one way of detecting acoustic activity and recognizing a response or message.

The assignment of an identification code and semantic content to the reference acoustic signals of the data bank is necessary for their ordering and unambiguous semantic determination when compiling and using the bank of reference acoustic signals. Such a bank has unlimited possibilities of filling with the development of the level of communication and the demand for increasing information units.

The choice of musical sounds as the initial elements of acoustic signals is due to several reasons: firstly, the fact that dolphins have developed imitative abilities, including the ability to reproduce a set of vowels, and in tonalities characteristic of people: from high tenor to bass - males, and from high soprano to contralto - for females (Acoustic Journal, 2014, Volume 50, No. 3, pp. 349-356), which will allow the dolphin to reproduce (“sing”) the given character; secondly, it is known that dolphins respond favorably to the sounds of music, which is important in the process of communication; thirdly, signals consisting of musical sounds can be repeatedly and unambiguously reproduced by digital and analog audio systems using loudspeakers in the air and sonar transducers under water, as well as conventional acoustic instruments, which expands the technical capabilities of training and communication; fourthly, signals and features of their reproduction (radiation, reproduction) can also be represented in unchanged graphic form, in particular in the form of musical notation, which excludes the influence of linguistic differences of people on the reproduced signal (in the case of pronouncing commands by voice) and makes the proposed international way.

The proposed method allows to use modern methods for automated signal processing based on a comparison of incoming sounds received by the acoustic receiver with the reference signals of the data bank for detecting and recognizing the response, which helps to increase the accuracy of signal recognition and allows us to solve the problem.

Known modern achievements in the field of recognition and identification of audio signals. For these purposes, complex algorithms and computer programs for sound analysis are developed, databases of sound fragments of speech or music are formed (for example, WIDI Recognition System, http://www.softkey.ru/catalog). If there is a specific sound fragment in the database, such a fragment, reproduced in a different key and with a different timbre color, can be identified with high probability. Similar means of recognizing audio signals can be used to analyze the response to an audio signal, including recognition and even identification of the source. For example, the vowel-like sounds of a beluga whale dolphin, which are characterized by a formant structure, can be used to identify a specific individual (V. M. Belkovich, S. A. Kreichi. Features of the vowel-like signals of a beluga whale. Acoustic Journal, 2004, Volume 50, No. 3. P. 349 -356).

Using the created bank of acoustic signals for messaging requires the preliminary training of one of the partners (in this case, the dolphin) using standard training techniques, for example, those used in the Wild Dolphin project.

Obviously, the acoustic signals generated in the bank, adapted for the range of hearing and the sound apparatus of dolphins, will be available for reception and playback by dolphins, and the exclusion of the use of graphic or subject symbols, as well as the playback of ready-made signals from the signal bank, allows increasing the speed and distance of signal exchange .

The implementation of the proposed method of creating a signal bank can be carried out using well-known technical means and computer technology both in air and under water.

The drawing shows a diagram of the implementation of the proposed method, where: 1 is a computer, 2 is a bank of reference sounds (a is an example of alphanumeric and musical notation); 3 - bank of reference acoustic signals (b - the content of the recorded characteristics of the reference acoustic signal), 4 - sound card, 5 - acoustic emitter; 6 - acoustic receiver; 7 - dolphin signal bank.

The method can be carried out, for example, as follows.

On computer 1, equipped with a sound editor program, sounds in the form of files in formats without digital compression are generated and stored on a medium (computer hard drive or external drive), for example with the extension “.wav”, corresponding to the sounds of the main scale of the first octave, each lasting, for example, 1 second and the same amplitude (for example, 1 V), while fixing their frequency, amplitude and time characteristics (for example, for the sound “fa” of the first octave - “F1_1_1”). By means of the spectral and temporal analysis options, frequency, amplitude and temporal characteristics of sounds are recorded in graphical form. Further, from the sounds recorded on the carrier by changing their duration or amplitude, reference sounds are formed in the frequency domain of the dolphins' communication activity, while their frequency, amplitude and temporal characteristics are also recorded, for example, for a note reduced in half duration and doubled in amplitude “Fa” of the first octave is “F1_0.5_2” and they form a bank of reference sounds 2. Then, from the bank of 2 reference sounds, reference acoustic signals are formed, which are various combinations of e coupon sounds and assign an identification code to each created acoustic signal, fix the frequency, amplitude and time characteristics of the signal and form a bank of 3 reference acoustic signals by these actions, while each signal or group of signals is assigned information content, which is fixed in the form of text.

In the process of use, the necessary acoustic signals from the bank 3 through the built-in or external sound card 4 and a typical amplification path are reproduced in air or under water using broadband acoustic emitters 5 in the corresponding frequency range with simultaneous execution of actions corresponding to the information content of the acoustic signals. The acoustic signals emitted into the medium and the dolphin's response acoustic activity are recorded using acoustic receivers 6 and, through typical amplification paths and a sound card 4, are recorded onto a carrier in a bank of dolphin signals 7 and then compared with the signals available in the bank of 3 reference signals. To do this, use methods of spectral, correlation analysis, algorithms and music recognition programs that identify acoustic signals in the dolphin response, preferably in real time. Considering that dolphins are capable of abstraction and onomatopoeia, and thus are able to distinguish and recognize similar acoustic signals (acoustic images), despite the differences in timbre coloring, the signals available in Bank 3 can be reproduced using various audio devices, musical instruments, and in a number of cases - by voice.

Thus, the inventive method of creating a databank of acoustic signals based on standardized elements in the form of sounds of a uniformly-tempered scale helps to solve the problem with the achievement of the claimed technical result, namely to increase the accuracy of recognition of acoustic signals. In addition, the use of sound signals instead of graphic or subject symbols to create a data bank allows increasing the speed and distance of signal exchange, as well as expanding the possibilities of filling the bank.

Currently, the inventive method is carried out and tested on laboratory equipment.

Claims (1)

A method of creating a bank of acoustic signals for exchanging messages between a dolphin and a person in the frequency domain characteristic of dolphin communication signals, including creating a bank of reference acoustic signals from individual sounds of a uniformly-tempered scale by formatting sounds for the same duration and amplitude, recording to a carrier and fixing them frequency, amplitude and time characteristics, then from the recorded sounds by changing their duration and / or amplitude to form the reference sound and they create from them a bank of reference sounds with fixing the characteristics of each reference sound, and then, using combinations of the received reference sounds, they form reference acoustic signals, fix the frequency, amplitude, time modulations of each generated signal and each of them is assigned a unique identification code and information content with subsequent comparison sounds recorded using acoustic receivers with reference bank signals.

Images