US2705260A - Phonetic printer of spoken words - Google Patents

Description

March 29, 1955 M, v. KALFAIAN PHOIZIETIC PRINTER OF SPOKEN WQRDS 2 Sheets-Sheet l Filed Dec. 3, 1952 I ONE CYCLE OF I I Fl/AIDANE/VTM.

sou/v0 "'a FUNDAMENTAL 110 CM D E N T UV w W PHONE TIC PRINTER Ll/IITER I1 IX ER TRANSPOSED 0R I/VAL SPEECH-WAVE FRE FREQUENCY TRANSPOSED NEGATIVE LFEED-BACK 4 6PEE CAI-WAVE ORIGINAL SPEECH-WAVE March 29, 1955 M. y. YKALFAIAN 5 2,705,260

\' PHONETIC PRINTER 0F SPOKEN WORDS Filed Dec. 3, 1952 2 sheets-sheet 2 SECTION OF 18 24 RED 19 23 GREEN g E- 20 35 E 6 BLUE w I: 23 2.6 21 8 it 1 wa 1 E .2-

0: 29 g a L:

2 in Z w E a g u E INVENTOR. k

United States Patent PHONETIC PRINTER 0F SPOKEN WORDS Meguer V. Kalfaian, Los Angeles, Calif.

Application December 3, 1952, Serial No. 323,873

Claims. (Cl. 178-31) This invention relates to the analysis of speech waves, and more particularly to those waves that are responsible for the intelligibility of phonetic characters in spoken sounds. Its main object is to provide methods and means for the analysis and selection of various frequency components that occur simultaneously during propagation of phonetic sounds, for the purpose of translating phonetic sounds into visible intelligible indicia. A corollary object is to provide methods and means to translate those simultaneously selected frequency components into discrete signals for the actuation of character-printing keys, for example the keys of a modified electric typewriter, or slotted code bars such as in teletypewriter devices, so that spoken words may be translated into visual words. In the course of translating those simultaneous signals into singular discrete signals, a novel ratio meter is provided herein, which is capable of compensating for most all of the variables, such as, pitch, formants, etc., that occur in speech waves, due to different speakers voices. This ratio meter had been described in my Patent No. 2,613,273 issued October 7, 1952; but it is particularly contemplated to be used in conjunction with the system of speech wave analysis described in my copending application Serial No. 268,243 filed January 25, 1952, wherein, methods and means had been disclosed for shifting the variable frequency bands contained in speech waves to standard bands, whereby selection of phonetic sounds could be more accurately achieved for translating spoken sounds into visible intelligible indicia.

In ordinary speech, each phonetic sound is produced in substantially replica wave trains that repeat successively at a fundamental (pitch) frequency, during propa-; gation of the articulated sound. The succession of these wave trains is effected by fairly regular puffs of air from the glottis, which are set into vibration in the momentarily formed resonant cavities of the vocal system. Each wave train contains all the phonetic information necessary, and its specific wave shape is formed by the.

number of frequency components that are produced in these cavities, and their relations with regard to frequency positions and relative amplitude levels one with another. The basic frequency components composing a pure phonetic sound are independent of characterization components, the latter of which are mainly produced by the larynx. The composite structure of these latter.components is inconsistent in form, and it varies in a complex manner with the varying pitch of the speakers voice. However, the frequency ratios of the basic components, and their relative amplitudes, remain substantially constant with respect to the fundamental frequency; even though the frequency locations of all these components may change in the entire spectrum band of the voice. Thus, the human intelligence interprets phonetic sounds by measuring the ratios of basic frequency components, and their relative amplitudes, with respect to the fundamental frequency, without regard to the characterization components; the latter of which is interpreted as a form of voice quality.

To define characterization complexities, as an example, let one speaker pronounce certain phonetic sound (in natural voice) at first and second fundamental frequencies. The listener can easily recognize the characteristic quality of the sound to be of the same speaker. But when the sound at the first fundamental is recorded and reproduced at the second fundamental (by speeding or retarding the movement of reproduction), the

2,705,260 Patented Mar. 29, 1955 listener can easily detect the phonetic sound but cannot recognize the characteristic quality of the voice. The concept of this peculiar condition has been advanced by actual recordings of male and female voices on magnetic tape. The consonant sounds had been both preceded and succeeded by vowel sounds, and the reproduction speed had been varied randomly. In all cases, the phonetic sound had been recognizable by a group of listeners.

A graphical comparison between the high pitched and low pitched voices of the same phonetic sound, as shown in Fig. 1, indicates that the high pitched wave is purer in sound than the lower pitched wave. Moreover these two waves indicate that the characterization frequency components are mostly outside the regions of the basic bandwidth of the high pitched sound; these two conditions hold true in the average, for different phonetic sounds (it is perhaps due to this peculiar condition why the human intelligence can easily separate the characterization components from the basic components of phonetic sounds for recognition). Thus, since the high pitched sound contains all the basic components of phonetic information, it is possible to first transpose the frequency positions of all the components during propagation of the speech waves, so that they will all be based on a single reference fundamental frequency, and set a boundary, within which to select the basic frequency components for analysis. Thus it may be seen that through the process of. frequency transposition, or standardization, and characterization-frequency filtration thereof, the true sounds of phonetics may be derived from the original speech waves, for final and more accurate analysis than by devices heretofore devised. In order to further advance the operational accuracy of phonetic selection, however, this invention contemplates to utilize the ratio meter, as described in my above noted patent, in conjunction with the system of frequency transposition, as described in my above noted 'co-pending application. When utilized in conjunction with the fre quency transposed speech waves, the ratio meter will provide a large number of predetermined adjustments, so that most of the undesired variables that still exist in the frequency transposed waves will be reduced to negligible values for practical purposes. The ratio meter will provide, by further modification, convenient output signals, representative of phonetic sounds, to operate printing mechanisms of different types, for example, the code bars such as used in teletypewriter devices.

In the drawings: Fig. 1 illustrates graph waveforms of speech waves; Fig. 2 is partly schematic and partly block diagram of the speech wave analyzer in accordance with the invention; Fig. 3 is an amplitude control device of the speech waves, as used in conjunction with the arrangement of Fig. 2; and Figures 4, 5, 6 are modifications of the ratio meter utilized in conjunction with the arrangement of Fig. 2.

In Fig. 2, the original speech waves in block 1 are first frequency transposed in block 2, and the various frequency components of importance are selected therefrom, by the band-pass filters, such as 1111 to f4 inclusive. The oscillatory output waves of these pass band circuits are rectified, as shown, and further, the rectified outputs are independently applied upon the electrostatic defleeting plates of a cathode ray tube 3, which is utilized as the ratio meter. As described in the foregoing, each phonetic sound is determined by the number of frequency components that are produced simultaneously, and their relations with regard to frequency positions and amplitude levels one with another. Thus, in an exemplary embodiment of the invention, assuming that a certain phonetic sound had contained four simultaneous frequency components, such as ft to f4, of various amplitudes, then the electron beam 4 will be deflected angularly (from its normal central position), whose orientation is a function of the ratio of the differences of voltageamplitudes applied upon the four beam-deflecting plates. Accordingly, if a signal target 5 were placed in that path of the beam, as shown, a distinct output signal would be produced by the target, representative of the original phonetic sound. Similarly, other target sectors could be placed at predetermined positions, each of which, when the beam strikes it, would produce an output signal,

water so as to render the technical application of water possible. Therefore, it is of no consequence which hydroxides or salts are employed provided that they are sutliciently water-soluble and neutral, i. e. they must not form stable addition products with the nitrogen compounds to be separated and that they do not undergo reaction with the nitrogen compounds. Especially suitable salts are, for instance, common salt, sodium sulphate, sodium carbonate, sodium phosphate, sodium acetate, sodium formate as well as the corresponding potassium salts and alkali hydroxides, such as sodium and potassium hydroxide. Further substances which may be employed, are described, for instance, in British specification No. 475,818. The said salt solution may contain according to the special requirements only small amounts of the salt or quantities up to saturation. On using alkali hydroxides, solutions containing from about to about 40% of the hydroxide are preferred.

Which of the nitrogen compounds is preferably absorbed depends on the nature of the absorbent applied. Thus, the invention permits of adapting the process to the prevailing conditions of the various absorbents in the single steps of the reaction. On the other hand, it is possible to apply the absorbents in combination in the same step as far as they agree as to their separating activity. For instance, the weak acids may be employed in combination with neutral solvents boiling not substantially lower than the weak acid applied and being indiflierent to the weak acid as well as to the nitrogen compounds and yielding homogeneous mixtures with the weak acid. Suitable solvents are for instance o-dichlorobenzene, 1.2.4-trichlorobenzene, nitrobenzene, tetralin, dekalin, higher boiling aliphatic or aromatic hydrocarbons as far as they are still liquid under the reaction conditions applied, as well as higher boiling ethers, alcohols, ketones and polyalcohols.

The application of mixtures of the weak acids with the organic solvents is especially advantageous in the separation of ammonia from mixtures containing methyl amines and in the separation of a mixture consisting of monoand dimethylamine. Furthermore, it is possible in the separation of trimethylamine from methylamine mixtures being free of ammonia to increase the separating activity of the weak acids by addition of water. Of course, water must not be added in quantities exceeding saturation at the temperatures employed.

The process according to the invention may be advantageously carried out by a continuous method by feeding the reaction mixture, if desired under pressure, in a reaction tower counter-currently to the flow of the absorbent. By appropriately adjusting the flow velocity and the temperature one or more nitrogen compounds are selectively dissolved in the weak acids or in the said other absorbents applied whereas the nitrogen compounds not absorbed escape as vapours at the top of the reaction tower. The absorbed compounds are expelled from the absorbent as described above. By repeating the process once or several times each of the components contained in the starting mixture may be obtained in pure form.

The process herein described is substantially different from that disclosed in German Patent 615,527. German Patent 615,527 comprises the separation of trimethylamine and ammonia by treatment with acids in quantities insufiicient for neutralization. The resultant salts cannot be decomposed again by merely heating or by reducing the pressure.

The invention is further illustrated by the following examples, without being restricted thereto.

Example 1 A mixture of 62.5% by volume of ammonia and 37.5% by volume of trimethylamine is passed through a liquid mixture of by Weight of phenol and 75% by weight of o-dichlorobenzene. At the beginning the mixture is completely absorbed. After saturation of the absorbent a mixture of 90% by volume of ammonia and 10% by volume of trimethylamine escapes. The mixture of ammonia and trimethylamine dissolved in the absorbent is expelled again by heating to 170 C. The mixture consists of 33% by volume of ammonia and 67% by volume of trimethylamine. By repeating the process several times, each of the two components is obtained in pure form.

Example 2 A mixture of ammonia and dimethylamine is introduced into a molten mixture of aand fi-naphthol, the proportion of the mixtures being 1:1. After saturation of the naphthol melt at about C. with the bases a gas mixture consisting of 68% by volume of ammonia and 32% by volume of dimethylamine escapes. By repeating the process several times, each of the two components is obtained in pure form.

Example 3 400 parts by weight of a solvent mixture consisting of 25% by weight of phenol and 75 by weight of o-dichlorobenzene is saturated with a mixture consisting of 78% by volume of trimethylamine and 22% by volume of ammonia. 108 parts by weight of the mixture are totally absorbed. Thereupon pure trimethylamine is introduced into the saturated solution through a glass frit. The escaping gas mixture consists of 50% by volume each of ammonia and trimethylamine. As soon as the content of ammonia in the escaping gas decreases feeding of pure trimethylamine is stopped. By heating the solution 112 parts by weight of a 96.5% trimethylamine are obtained.

Example 4 M-cresol and a gas mixture of approximately equal parts by volume of ammonia, dimethylamine, and trimethylamine are contacted in countercurrent in an ab sorption tower packed with Raschig rings, said absorption tower having a length of 2.50 m. and a diameter of 3 cm. 45 liters of the aforesaid mixture and 120 grams of m-cresol are charged each hour. The gas escaping at the top of the tower consists of 99% ammonia whereas the mixture of methylarnines expelled from the absorbent is almost free from ammonia.

Example 5 The mixture of dimethylamine and trimethylamine set free on heating the sump obtained according to Example 4 is contacted with m-cresol in an absorption tower as indicated in Example 4. About 48 liters of the mixture of the methylamines and 90 grams of m-cresol are charged each hour. 98% trimethylamine escapes at the top of the reaction tower whereas a 90% dimethylamine is obtained by heating the sump solution.

Example 6 A mixture consisting of 55% by volume of ammonia, 15% by volume each of mono-, di-, and trimethylamine is contacted in countercurrent with a technical cresol mixture (30 grams per hour) in an absorption tower packed with Raschig rings, said absorption tower having a diameter of 25 mm. and a height of 2.50 m.; the throughput of said mixture amounts to 30 liters per hour. The nonabsorbed gas contains 100% of the amount of ammonia charged and of the trimethylamine charged and is free from monoand dimethylamine.

The mixture absorbed by the cresol and containing besides small amounts of trimethylamine, the whole monoand dimethylamine is contacted after expelling from the solvent with a mixture consisting of 1 part by weight of phenol and 3 parts by weight of o-dichlorobenzene in the same reaction tower and in similar manner.

monomethylamine escapes at the top of the reaction tower whereas 92% dimethylamine is obtained from the sump solution.

Example 7 A mixture of 49% by volume of ammonia and 17% by volume each of mono-, di-, and trimethylamine at a rate of 29 liters per hour is contacted, in countercurrent, at room temperature with a caustic soda solution of 10% strength in an absorption tower packed with Raschig rings and having a height of 2.50 m. and a diameter of 25 mm. The gas mixture is fed at a point in the middle of the tower, the sump of the absorption tower is heated to 45 C. When charging 70 cm. of caustic soda solution per hour 100% trimethylamine is taken off from the top of the tower. The dissolved nitrogen compounds are practically free from trimethylamine.

The dissolved mixture of nitrogen compounds is expelled by heating and contacted in a similarly constructed tower with a technical cresol mixture of such an amount that the monoand dimethylamine contained in the mixture are dissolved whereas pure ammonia escapes at the top of the tower.

may also be achieved by different arrangements, as for example, by utilizing two opposing Windings in each relay, such as shown in the above mentioned Patent No. 2,540,660, or in the arrangement given in Patent No. 2,556,975, June 12, 1951, the disclosure of which is hereby made part of this invention, as if fully included herein.

The monochrome photographic film 12 may also be arranged in colors, such as shown in Fig. 5. In this case, the chromatic film 17 is interposed between a plurality of photocells 18, 19, 20, 21, etc., and the phosphor screen 22 of cathode ray tube 23. In front of each photocell there is placed a color filter, 24, 25, 26, 27, etc, each of which admits light of a prearranged color to its associated photocell. Thus, when the electron beam 28 luminesces the phosphor screen 22 at an area where the film 17 passes only red light, the photocell 18 becomes excited, while the others remain idle. These selected signals may then be applied upon a phonetic prmter, for mechanical actuation.

For coding purposes, each section on the color film 17, representative of a phonetic sound, may be arranged to have a color code, such as shown in the sect1onal drawing of the color film 17. In this drawing, three primary colors (red, green and blue) are shown, which collectively represent a prearranged phonetic sound. In order that the beam may cover all the primary colors in a given area, the dilferent colors may be arranged in thin stripes, as shown, so that the beam W111 always see all the colors in that given section. In the case that 32 phonetic characters are utilized for final printing, only five different colors, and five photocells are needed. The outputs of these photocells are then applied upon the code bars, for example of a teletypewriter device, for final printing. For automatic volume control, the film may be arranged with a circular ring of another color, so as to efiect excitation of a photocell, which may be used for controlling the amplitude level of the speech waves.

The coding system just described, may also be adapted to the target sectors of the ratio meter shown in Fig. 2. In such an arrangement, for example, target sector may be divided into narrow strips, and their output signals applied upon the code bars of a teleprinter. Because of the narrowness of the target strips, they may be physically arranged as shown in Fig. 6. In this arrangement, the target sectors 29 are mounted on an insulator base 30 (shown in cross sectional view), and on the other side of this insulator there are mounted electrical connectors 32, by way of metal pins 31. The electrical connectors are in the form of metal rings 33, 34, 35 and 36 (shown in front view), the outputs of which are applied upon the code bars of a teleprinter. In one mode of manufacturing process, the connector rings may first be mounted on the surface of the insulator base 30, with the said metal pins passing therethrough, and the target strips mounted on the other side of the insulator; touching the pins. For mounting the target strips, the whole surface of the insulator may be coated with metal, and cut into sections with a die, or by photoetching method.

While I have described particular embodiments of my invention, numerous substitutions of parts, adaptations and modifications are possible without departing from the spirit and scope thereof. To demonstrate one possible modification of the arrangements given herein, it is possible to utilize greater number of frequency selection from the total spectrum band of the voice frequencies, with lesser number of beam-deflecting means of the ratio meter, described herein, by sub-dividing and mixing the outputs of the band-pass filters before applying upon the beam-deflecting means of the ratio meter. This is possible due to the many prearrangements that may be made with the ratio meter. With such operating conditions, the frequency bands associated with some consonant sounds, for example, the sounds s, 2, which comprise bands of higher frequencies than other sounds, may be separated from other sounds more easily than previously described.

What I claim is:

1. In speech wave analysis where each phonetic sound may be represented by substantially a specific combination of frequency components having definite amplitude relations one with another, the system of deriving and reducing said combinations into singular discrete signals representative of said phonetic sounds, which comprises in combination means for producing speech waves; means for sub-dividing said waves into a plurality of predetermined frequency bands, and a plurality of pass-band means therefor, for selecting same; means for rectifying the output oscillations of said pass-band means, whereby to obtain a plurality of substantially unidirectional signals, any collective combination of which representing a phonetic sound contained in the speech waves; means for producing a ray; a plurality of ray-deflecting means arranged around said ray; means for individually applying said plurality of simultaneous signals upon said plurality of deflecting means for deflecting the ray angularly from plurality of directions to a mean angular position whose orientation is a function of the ratio of differences of said plurality of angular deflections; a plurality of target sectors in the path of said ray, at predetermined directions from the undeflected path, each adapted to produce an output signal when the ray strikes it, whereby indicating when the ratio of said differences controlling the deflection corresponds to the direction of that target sector, so that each of said target sector will produce a singular discrete signal representing a combination of frequency components aforesaid.

2. The system as set forth in claim 1, wherein, each of said target sectors is sub-divided into a plurality of narrow coded-sectors, so narrow that the total number of coded-sectors comprising a code will intercept the stream of said ray simultaneously; and means for applying the outputs of said coded-sectors to code bars of a printing device comprising coded bars, thereby to effect translation of phonetic sounds into visible intelligible indicia.

3. The system as set forth in claim 1 which includes a phonetic printer; and means for controlling the operation of the keys of said printer by said discrete signals, whereby to translate said phonetic sounds into visible intelligible indicia.

4. Apparatus as set forth in claim 1, which includes an amplitude-control target disposed around the periphery of said plurality of target sectors, in the path of said ray; means for deriving an output signal from the amplitude-control target whenever said ray strikes it; and means for applying last said signal upon the source of said speech waves for controlling its output amplitude level, whereby the amplitude level of the speech wave may be automatically controlled to keep the angular deflection of said ray within the apparatus limits of said plurality of target sectors.

The system as set forth in claim 1, which includes means for deriving output signals during transient periods of one phonetic sound to the other in the speech waves; and means utilizing last said signals for suppressing the production of said discrete signals, thereby avoiding erroneous production of said discrete signals during said transient periods.

References Cited in the file of this patent UNITED STATES PATENTS 2,137,888 Fuller Nov. 22, 1938 2,195,081 Dudley Mar. 26, 1940 2,375,044 Skellett May 1, 1945 2,575,909 Davis et al. Nov. 20, 1951 2,602,836 Foster et a]. July 8, 1952 2,613,273 Kalfaian Oct. 7, 1952 2,646,465 Davis et al. July 21, 1953

Images