System and method for speech to text translation using cores of a natural liquid architecture system

Abstract

A system and method for speech-to-text translation. The method includes determining, based on at least one audio input in a first language, at least one original language concept; identifying, based on the determined at least one original language concept, the first language of the at least one audio input; determining, for each original language concept, a matching translated concept, wherein each matching translated concept is associated with a second language, wherein the second language is different from the first language; generating a textual output based on the determined at least one translated concept.

Description

TECHNICAL FIELD

The present disclosure relates generally to pattern recognition in speech translation and, more particularly, to pattern recognition in audio analysis for speech translation.

BACKGROUND

Sound files, like images, may be indexed by their titles. Unfortunately, if a sound file is simply an embedded or linked audio file on a Web page, there may be no additional information about it. The audio files may have some descriptive information included, such as the source. Other metadata can be included in audio files, but such inclusion requires more effort on the part of the content producer and, as in the case of images, the metadata may be incomplete or insufficient.

To fully index the content of audio files generally requires having a transcript of the session in a computer-readable text format that enables text-indexing. With voice recognition software, some automated indexing of audio files is possible and has been successfully used. However, it is widely known that such transcripts rarely match what was spoken exactly. The difficulty is compounded if the spoken words are sung and the search is for the song in a specific tune, or a search for a tune regardless of the words.

Analysis of audio signals is desirable for a wide variety of reasons such as speaker recognition, voice command recognition, dictation, instrument or song identification, and the like. In some instances, it may be desirable to convert human speech from one language to one or more other languages in real-time or at a later time. Particularly, a user listening to an audio signal may wish to hear the contents of the file in another language. Currently real-time speech translation is largely performed by human translators, as any machine-based translation algorithm does not provide reliable results.

It would be therefore advantageous to provide a solution that would overcome the challenges noted above.

SUMMARY

A summary of several example embodiments of the disclosure follows. This summary is provided for the convenience of the reader to provide a basic understanding of such embodiments and does not wholly define the breadth of the disclosure. This summary is not an extensive overview of all contemplated embodiments, and is intended to neither identify key or critical elements of all embodiments nor to delineate the scope of any or all aspects. Its sole purpose is to present some concepts of one or more embodiments in a simplified form as a prelude to the more detailed description that is presented later. For convenience, the term “some embodiments” or “certain embodiments” may be used herein to refer to a single embodiment or multiple embodiments of the disclosure.

Certain embodiments disclosed herein include a method for speech-to-text translation. The method comprises: determining, based on at least one audio input in a first language, at least one original language concept; identifying, based on the determined at least one original language concept, the first language of the at least one audio input; determining, for each original language concept, a matching translated concept, wherein each matching translated concept is associated with a second language, wherein the second language is different from the first language; generating a textual output based on the determined at least one translated concept.

Certain embodiments disclosed herein also include a non-transitory computer readable medium having stored thereon causing a processing circuitry to execute a process, the process comprising: determining, based on at least one audio input in a first language, at least one original language concept; identifying, based on the determined at least one original language concept, the first language of the at least one audio input; determining, for each original language concept, a matching translated concept, wherein each matching translated concept is associated with a second language, wherein the second language is different from the first language; generating a textual output based on the determined at least one translated concept.

Certain embodiments disclosed herein also include a system for speech-to-text translation. The system comprises: a processing circuitry; and a memory, the memory containing instructions that, when executed by the processing circuitry, configure the system to: determine, based on at least one audio input in a first language, at least one original language concept; identify, based on the determined at least one original language concept, the first language of the at least one audio input; determine, for each original language concept, a matching translated concept, wherein each matching translated concept is associated with a second language, wherein the second language is different from the first language; generate a textual output based on the determined at least one translated concept.

BRIEF DESCRIPTION OF THE DRAWINGS

The subject matter that is regarded as the disclosure is particularly pointed out and distinctly claimed in the claims at the conclusion of the specification. The foregoing and other objects, features, and advantages of the disclosed embodiments will be apparent from the following detailed description taken in conjunction with the accompanying drawings.

FIG. 1 is a flowchart illustrating a method for speech-to-text translation according to an embodiment.

FIG. 2 is a schematic diagram of a speech-to-text translator according to an embodiment.

FIG. 3 is a block diagram depicting the basic flow of information in the signature generator system.

FIG. 4 is a diagram showing the flow of patches generation, response vector generation, and signature generation in a large-scale speech-to-text system.

FIG. 5 is a network diagram utilized to describe the various disclosed embodiments.

DETAILED DESCRIPTION

It is important to note that the embodiments disclosed herein are only examples of the many advantageous uses of the innovative teachings herein. In general, statements made in the specification of the present application do not necessarily limit any of the various claimed inventions. Moreover, some statements may apply to some inventive features but not to others. In general, unless otherwise indicated, singular elements may be in plural and vice versa with no loss of generality. In the drawings, like numerals refer to like parts through several views.

A system and method for speech-to-text translation. Signatures are generated for audio inputs in a first language. Based on the generated signatures, original language concepts representing portions of the audio are determined. The original language concepts are associated with the first language. Matching translated concepts associated with a desired second language are identified. Textual output is generated based on the identified translated concepts.

FIG. 1 is an example flowchart 100 illustrating a method for speech-to-text translation according to an embodiment. In an embodiment, the method may be performed by the speech-to-text translator 200, FIG. 2.

At S110, at least one audio input is received. Each audio input may be, but is not limited to, a digital representation of an audio signal, a direct feed from one or more microphones, a combination thereof, and the like. In an embodiment, a plurality of audio inputs from a single source is received. As a non-limiting example, a plurality of audio inputs may be received from a plurality of microphones directed at a single podium with one or more speakers.

At S120, at least one signature is generated for the received audio inputs. Each signature may be generated based on an entire audio input, a portion of an audio input, or both. In an embodiment, the signatures are generated as described further herein below with respect to FIGS. 3 and 4. In another embodiment, each generated signature may be stored in, e.g., a database.

In an embodiment, S120 includes generating the signatures via a plurality of at least partially statistically independent computational cores, where the properties of each core are set independently of the properties of the other cores. In another embodiment, S120 includes sending the multimedia content element to a signature generator system and receiving the plurality of signatures. The signature generator system includes a plurality of at least statistically independent computational cores as described further herein. The signature generator system may include a large ensemble of randomly and independently generated heterogenous computational cores, mapping data-segments onto a high-dimensional space in parallel and generating compact signatures for classes of interest.

Each signature represents a concept, and may be robust to noise and distortion. Each concept is a collection of signatures representing multimedia content elements and metadata describing the concept, and acts as an abstract description of the content to which the signature was generated. As a non-limiting example, a ‘Superman concept’ is a signature-reduced cluster of signatures describing elements (such as multimedia elements) related to, e.g., a Superman cartoon: a set of metadata providing a textual representation of the Superman concept. As another example, metadata of a concept represented by the signature generated for a picture showing a bouquet of red roses is “flowers”. As yet another example, metadata of a concept represented by the signature generated for a picture showing a bouquet of wilted roses is “wilted flowers”.

At S130, the at least one generated signature is compared to a plurality of previously generated signatures to determine at least one matching signature. The plurality of previously generated signatures may be stored in, e.g., a signature database. In an embodiment, if no matching signature is determined, S130 may result in a null value indicating that a translation for the terms represented by the concept is not available.

At S140, at least one cluster is identified based on the determined matching signatures. Each cluster includes a group of signatures, where each signature in the group at least partially matches each other signature in the group. A matching portion of signature that is common to all signatures of the cluster is a concept represented by the cluster. Clustering of signatures is described further in U.S. Pat. No. 8,386,400 assigned to the common assignee, which is hereby incorporated by reference.

The clustering process may map a certain content-universe onto a hierarchical structure of clusters. The content-elements of the content-universe are mapped to signatures as appropriate. The signatures of all of the content-elements are matched to each other and, consequently, such matching generates an inter-match matrix. Generation of the inter-match matrix leads to a set of clusters. This results in a highly compressed representation of the content-universe.

At S150, an original language concept is identified for each cluster. The original language concepts may be identified based on previously generated concepts, or based on concepts generated in response to identification of clusters.

At S160, a first language (e.g., Hebrew, English, Spanish, etc.) is determined based on the identified first concepts. The language may be determined by different classification techniques. One such example is a statistical approach based on prevalence of certain function words (such as the word “the” in the English language). Another example is to create a language n-gram model from a training audio file for each language which the system may detect. For any audio for which a language needs to be determined, a similar model is made, and that model is compared to each stored language model. The most likely language is the one with the model that is most similar to the model from the audio needing to be identified.

At S170, a matching translated concept is determined for each identified original language concept. Each of the matching translated concepts is associated with a second language. In an embodiment, S170 includes comparing a signature representing each identified original langauge concept to a plurality of previously generated signatures. Matching may be performed, for example, by statistically identifying proximity of signatures or concepts to each other. In the above example, the concept of “tree” may often appear in proximity to words such as “green”, “brown”, “tall”, and so on in the English language. The concept of “arbre” may often appear in proximity to words such as “vert”, “brun” and “grand” in the French language. It is therefore statistically possible to match “tree” to “arbre” with a degree of certainty determined, for example, by a threshold. Proximity may be based on whether such words appear within the same sentence, paragraph, and the like. Proximity may be, for example, audio detected within a window of time before or after the concept. In another embodiment, proximity may additionally be determined by considering placement of the second concept within written text.

In an embodiment, a translated concept may only be matched if it is associated with a desired second language. As an example, if the desired second language is English, concepts that are similar may only be provided as a match if such concepts are associated with the English language. Association with a language may be determined based on, e.g., metadata associated with the concepts. Which language is desired as the translated language may be determined by, but is not limited to, user preferences provided by a user, a user profile based on previously identified concepts by that user, and so on. In some embodiments, a plurality of translated concepts, each translated concept associated with a distinct language, may be provided, thereby allowing for translation into multiple different languages.

In another embodiment, the desired second language may be indicated in a user profile of a user to view the translated text. The user profile may be generated and modified based on a user's impressions with respect to multimedia content elements. Impressions may be determined based on, but is not limited to, a user gesture; adjustment to computer volume by a user, time spent viewing, interacting with, or listening to a multimedia content element; key strokes entered while viewing or listening to a multimedia content element; and so on. A user gesture may be, but is not limited to, a mouse click, a mouse scroll a tap, a swipe, and any other gesture on a device having a touch screen display or a pointing device. User profiles and user impressions are discussed in more detail in U.S. patent application Ser. No. 13/856,201 assigned to common assignee, which is hereby incorporated by reference for all that it contains.

As a non-limiting example of matching based on a user profile, past interactions with multimedia content featuring English language text and audio demonstrate a positive impression of English language content (i.e., that the user interacts with English language content, suggesting that the user can read English content), thereby causing a subsequently generated user profile to associate the user with the English language as an English language speaker. When the user later listens to an audio file containing Italian speech, the concepts of the words in the audio file are determined and matched respective of related second concepts associated with the English language.

At S180, a textual output is generated based on the determined translated concepts. In an embodiment, S180 includes retrieving (e.g., from a translation database) textual content associated with each matching translated concept, where the texual output is generated using the retrieved textual content. The textual output may be caused to be displayed on, e.g., a user device.

At S190, it is determined whether additional audio content is to be translated and, if so, execution continues with S130; otherwise, execution terminates. Multiple translations may allow for, e.g., translating the same audio inputs to multiple languages.

FIG. 2 is an example schematic diagram of a speech-to-text translator 200 according to an embodiment. The speech-to-text translator 200 includes a processing circuitry 210 coupled to a memory 220, a storage 230, an audio input interface 240, and a network interface 250. In an embodiment, the components of the speech-to-text translator 200 may be communicatively connected via a bus 205.

The processing circuitry 210 may be realized as one or more hardware logic components and circuits. For example, and without limitation, illustrative types of hardware logic components that can be used include field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), Application-specific standard products (ASSPs), system-on-a-chip systems (SOCs), general-purpose microprocessors, microcontrollers, digital signal processors (DSPs), and the like, or any other hardware logic components that can perform calculations or other manipulations of information. In an embodiment, the processing circuitry 210 may be realized as an array of at least partially statistically independent computational cores. The properties of each computational core are set independently of those of each other core, as described further herein above.

The memory 220 may be volatile (e.g., RAM, etc.), non-volatile (e.g., ROM, flash memory, etc.), or a combination thereof. In one configuration, computer readable instructions to implement one or more embodiments disclosed herein may be stored in the storage 230.

In another embodiment, the memory 220 is configured to store software. Software shall be construed broadly to mean any type of instructions, whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise. Instructions may include code (e.g., in source code format, binary code format, executable code format, or any other suitable format of code). The instructions, when executed by the processing circuitry 610, cause the processing circuitry 210 to perform the various processes described herein. Specifically, the instructions, when executed, cause the processing circuitry 210 to perform speech-to-text translation based on audio inputs received from the audio input interface 240 as described herein. The audio input interface 240 may be used to receive different signals, a single signal from a plurality of locations, or any combination thereof.

The storage 230 may be magnetic storage, optical storage, and the like, and may be realized, for example, as flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVDs), or any other medium which can be used to store the desired information.

The network interface 250 allows the speech-to-text translator 130 to communicate with the signature generator system 140 for the purpose of, for example, sending multimedia content elements, receiving signatures, and the like. Further, the network interface 250 allows the speech-to-text translator 130 to receive audio inputs from the audio input interface 240.

It should be understood that the embodiments described herein are not limited to the specific architecture illustrated in FIG. 2, and other architectures may be equally used without departing from the scope of the disclosed embodiments. In particular, the speech-to-text translator 130 may further include a signature generator system configured to generate signatures as described herein without departing from the scope of the disclosed embodiments.

FIGS. 3 and 4 illustrate the generation of signatures for the multimedia content elements by the SGS 140 according to an embodiment. An exemplary high-level description of the process for large scale matching is depicted in FIG. 3. In this example, the matching is for a video content.

Video content segments 2 from a Master database (DB) 6 and a Target DB 1 are processed in parallel by a large number of independent computational Cores 3 that constitute an architecture for generating the Signatures (hereinafter the “Architecture”). Further details on the computational Cores generation are provided below. The independent Cores 3 generate a database of Robust Signatures and Signatures 4 for Target content-segments 5 and a database of Robust Signatures and Signatures 7 for Master content-segments 8. An exemplary and non-limiting process of signature generation for an audio component is shown in detail in FIG. 3. Finally, Target Robust Signatures and/or Signatures are effectively matched, by a matching algorithm 9, to Master Robust Signatures and/or Signatures database to find all matches between the two databases.

To demonstrate an example of the signature generation process, it is assumed, merely for the sake of simplicity and without limitation on the generality of the disclosed embodiments, that the signatures are based on a single frame, leading to certain simplification of the computational cores generation. The Matching System is extensible for signatures generation capturing the dynamics in-between the frames. In an embodiment the server 130 is configured with a plurality of computational cores to perform matching between signatures.

The Signatures' generation process is now described with reference to FIG. 4. The first step in the process of signatures generation from a given speech-segment is to breakdown the speech-segment to K patches 14 of random length P and random position within the speech segment 12. The breakdown is performed by the patch generator component 21. The value of the number of patches K, random length P and random position parameters is determined based on optimization, considering the tradeoff between accuracy rate and the number of fast matches required in the flow process of the server 130 and SGS 140. Thereafter, all the K patches are injected in parallel into all computational Cores 3 to generate K response vectors 22, which are fed into a signature generator system 23 to produce a database of Robust Signatures and Signatures 4.

In order to generate Robust Signatures, i.e., Signatures that are robust to additive noise L (where L is an integer equal to or greater than 1) by the Computational Cores 3 a frame ‘i’ is injected into all the Cores 3. Then, Cores 3 generate two binary response vectors: {right arrow over (S)} which is a Signature vector, and {right arrow over (RS)} which is a Robust Signature vector.

For generation of signatures robust to additive noise, such as White-Gaussian-Noise, scratch, etc., but not robust to distortions, such as crop, shift and rotation, etc., a core Ci={n_i} (1≤i≤L) may consist of a single leaky integrate-to-threshold unit (LTU) node or more nodes. The node n_i equations are:

$V_i=\sum _j{w}_{\mathrm{ij}}{k}_j$ $n_i=\theta \left(\mathrm{Vi}-{\mathrm{Th}}_x\right)$

where, θ is a Heaviside step function; w_ij is a coupling node unit (CNU) between node i and image component j (for example, grayscale value of a certain pixel j); kj is an image component ‘j’ (for example, grayscale value of a certain pixel j); Th_x is a constant Threshold value, where ‘x’ is ‘S’ for Signature and ‘RS’ for Robust Signature; and Vi is a Coupling Node Value.

The Threshold values Th_x are set differently for Signature generation and for Robust Signature generation. For example, for a certain distribution of Vi values (for the set of nodes), the thresholds for Signature (Th_S) and Robust Signature (Th_RS) are set apart, after optimization, according to at least one or more of the following criteria:

1 For: V_i>Th_RS 1−p(V>Th_S)−1−(1−ε)^l<<1i.e., given that l nodes (cores) constitute a Robust Signature of a certain image I, the probability that not all of these I nodes will belong to the Signature of same, but noisy image, Ĩ is sufficiently low (according to a system's specified accuracy).

2 p(V_i>Th_RS)≈l/L

i.e., approximately l out of the total L nodes can be found to generate a Robust Signature according to the above definition.

3: Both Robust Signature and Signature are generated for certain frame i.

It should be understood that the generation of a signature is unidirectional, and typically yields lossless compression, where the characteristics of the compressed data are maintained but the uncompressed data cannot be reconstructed. Therefore, a signature can be used for the purpose of comparison to another signature without the need of comparison to the original data. The detailed description of the Signature generation can be found in U.S. Pat. Nos. 8,326,775 and 8,312,031, assigned to the common assignee, which are hereby incorporated by reference.

A Computational Core generation is a process of definition, selection, and tuning of the parameters of the cores for a certain realization in a specific system and application. The process is based on several design considerations, such as:

(a) The Cores should be designed so as to obtain maximal independence, i.e., the projection from a signal space should generate a maximal pair-wise distance between any two cores' projections into a high-dimensional space.

(b) The Cores should be optimally designed for the type of signals, i.e., the Cores should be maximally sensitive to the spatio-temporal structure of the injected signal, for example, and in particular, sensitive to local correlations in time and space. Thus, in some cases a core represents a dynamic system, such as in state space, phase space, edge of chaos, etc., which is uniquely used herein to exploit their maximal computational power.

(c) The Cores should be optimally designed with regard to invariance to a set of signal distortions, of interest in relevant applications.

A detailed description of the Computational Core generation and the process for configuring such cores is discussed in more detail in U.S. Pat. No. 8,655,801, referenced above.

FIG. 5 is an example network diagram 500 utilized to describe the various disclosed embodiments. The network diagram 500 includes a user device 520, the speech-to-text translator (STTT) 200, a database 530, and a plurality of audio capturing devices (ACDs) 550-1 through 550-n (hereinafter referred to individually as an audio capturing device 550 and collectively as audio capturing devices 550, merely for simplicity purposes) communicatively connected via a network 510. The network 510 may be, but is not limited to, the Internet, the world-wide-web (WWW), a local area network (LAN), a wide area network (WAN), a metro area network (MAN), and other networks capable of enabling communication between the elements of the network diagram 500.

The user device 520 may be, but is not limited to, a personal computer (PC), a personal digital assistant (PDA), a mobile phone, a smart phone, a tablet computer, a wearable computing device and other kinds of wired and mobile appliances, equipped with browsing, viewing, listening, filtering, managing, and other capabilities that are enabled as further discussed herein below. The user device 520 may have installed thereon an agent 525 such as, but not limited to, a web browser, an application, and the like. The application 525 may be configured to receive and display textual content.

The database 530 may store signatures, multimedia content elements, or both. Each of the multimedia content elements stored in the database 530 may be associated with one or more of the stored signatures. In some implementations, multiple databases (not shown), each storing signatures, multimedia content elements, or both, may be utilized.

The speech-to-text translator 200 is configured to obtain audio inputs in a first language from, e.g., the database 530, the audio capturing devices 550, or a combination thereof, and to generate textual outputs in a second language as described further herein above. The speech-to-text translator 200 may be configured to store the textual outputs in the database 530, to send the textual outputs to the user device 520, or both.

In an embodiment, the speech-to-text translator 200 is communicatively connected to a signature generator system (SGS) 540, which is utilized by the speech-to-text translator 200 to perform the various disclosed embodiments. Specifically, the signature generator system 540 is configured to generate signatures to multimedia content elements and includes a plurality of computational cores, each computational core having properties that are at least partially statistically independent of each other core, where the properties of each core are set independently of the properties of each other core.

The signature generator system 540 may be communicatively connected to the The signature generator system 540 may be communicatively connected to the speech-to-text translator 200 directly (as shown), or through the network 510 (not shown). In another embodiment, the speech-to-text translator 200 may further include the signature generator system 540, thereby allowing the speech-to-text translator 200 to generate signatures for multimedia content elements directly (as shown), or through the network 510 (not shown). In another embodiment, the speech-to-text translator 200 may further include the signature generator system 540, thereby allowing the speech-to-text translator 200 to generate signatures for multimedia content elements.

The audio capturing devices 550 are configured to capture audio inputs to be translated. The audio capturing devices 550 may be, but are not limited to, microphones. Alternatively or collectively, audio inputs to be translated may include audio inputs stored in the database 530.

It should be noted that using signatures for determining the context ensures more accurate identification of trending multimedia content than, for example, based on metadata alone.

It should be noted that only one user device 520 and one agent 525 are described herein above with reference to FIG. 5 merely for the sake of simplicity and without limitation on the disclosed embodiments. Multiple user devices may receive textual outputs generated by the speech-to-text translator 530 without departing from the scope of the disclosure.

The various embodiments disclosed herein can be implemented as hardware, firmware, software, or any combination thereof. Moreover, the software is preferably implemented as an application program tangibly embodied on a program storage unit or computer readable medium consisting of parts, or of certain devices and/or a combination of devices. The application program may be uploaded to, and executed by, a machine comprising any suitable architecture. Preferably, the machine is implemented on a computer platform having hardware such as one or more central processing units (“CPUs”), a memory, and input/output interfaces. The computer platform may also include an operating system and microinstruction code. The various processes and functions described herein may be either part of the microinstruction code or part of the application program, or any combination thereof, which may be executed by a CPU, whether or not such a computer or processor is explicitly shown. In addition, various other peripheral units may be connected to the computer platform such as an additional data storage unit and a printing unit. Furthermore, a non-transitory computer readable medium is any computer readable medium except for a transitory propagating signal.

All examples and conditional language recited herein are intended for pedagogical purposes to aid the reader in understanding the disclosed embodiments and the concepts contributed by the inventor to furthering the art, and are to be construed as being without limitation to such specifically recited examples and conditions. Moreover, all statements herein reciting principles, aspects, and embodiments of the invention, as well as specific examples thereof, are intended to encompass both structural and functional equivalents thereof. Additionally, it is intended that such equivalents include both currently known equivalents as well as equivalents developed in the future, i.e., any elements developed that perform the same function, regardless of structure.

It should be understood that any reference to an element herein using a designation such as “first,” “second,” and so forth does not generally limit the quantity or order of those elements. Rather, these designations are generally used herein as a convenient method of distinguishing between two or more elements or instances of an element. Thus, a reference to first and second elements does not mean that only two elements may be employed there or that the first element must precede the second element in some manner. Also, unless stated otherwise, a set of elements comprises one or more elements.

As used herein, the phrase “at least one of” followed by a listing of items means that any of the listed items can be utilized individually, or any combination of two or more of the listed items can be utilized. For example, if a system is described as including “at least one of A, B, and C,” the system can include A alone; B alone; C alone; A and B in combination; B and C in combination; A and C in combination; or A, B, and C in combination.

Claims

1. A method for speech-to-text translation, comprising: determining, based on at least one audio input in a first language, at least one original language concept;identifying, based on the determined at least one original language concept, the first language of the at least one audio input;determining, for each original language concept, a matching translated concept, wherein each matching translated concept is associated with a second language, wherein the second language is different from the first language; andgenerating a textual output based on the determined at least one translated concept.

2. The method of claim 1, further comprising: selecting the second language based on past interactions of a user with multimedia.

3. The method of claim 1, wherein each matching translated concept is statistically proximate to one of the at least one original language concept.

4. The method of claim 3, wherein each signature is robust to noise and distortion.

5. The method of claim 3, wherein each signature is generated by a signature generator system, wherein the signature generator system includes a plurality of at least partially statistically independent computational cores, wherein the properties of each core are set independently of the properties of each other core.

6. The method of claim 1, further comprising: generating at least one signature for the at least one audio input, wherein the at least one original language concept is determined further based on the generated at least one signature.

7. The method of claim 6, wherein determining the at least one original language concept further comprises: determining at least one previously generated signature that matches the generated at least one signature, wherein each matching signature represents one of the at least one original language concept.

8. The method of claim 7, wherein each concept is a collection of signatures and metadata representing the concept.

9. The method of claim 1, comprising selecting the second language based on a user profile.

10. The method according to claim 1 wherein each concept is a collection of signatures and is an abstract description of contents for which the signatures of the concept were generated.

11. The method according to claim 1 herein the at least one input audio comprises a speech segment; wherein the method comprises generating at least one signature for the speech segment by breaking down the speech segment to multiple patches of random length and of random position in the speech segment.

12. A non-transitory computer readable medium having stored thereon instructions for causing a processing circuitry to execute a process, the process comprising: determining, based on at least one audio input in a first language, at least one original language concept; identifying, based on the determined at least one original language concept, the first language of the at least one audio input; determining, for each original language concept, a matching translated concept, wherein each matching translated concept is associated with a second language, wherein the second language is different from the first language; and generating a textual output based on the determined at least one translated concept.

13. A system for speech-to-text translation, comprising: a processing circuitry; and a memory connected to the processing circuitry, the memory containing instructions that, when executed by the processing circuitry, configure the system to: determine, based on at least one audio input in a first language; at least one original language concept, identify, based on the determined at least one original language concept, the first language of the at least one audio input; determine, for each original language concept, a matching translated concept, wherein each matching translated concept is associated with a second language, wherein the second language is different from the first language; and generate a textual output based on the determined at least one translated concept.

14. The system of claim 13, wherein each matching translated concept is statistically proximate to one of the at least one original language concept.

15. The system of claim 14, wherein each signature is robust to noise and distortion.

16. The system of claim 14, further comprising: a signature generator system, wherein each signature is generated by the signature generator system, wherein the signature generator system includes a plurality of at least partially statistically independent computational cores, wherein the properties of each core are set independently of the properties of each other core.

17. The system of claim 13, wherein the system is further configured to: generate at least one signature for the at least one audio input, wherein the at least one original language concept is determined further based on the generated at least one signature.

18. The system of claim 17, wherein the system is further configured to: determine at least one previously generated signature that matches the generated at least one signature, wherein each matching signature represents one of the at least one original language concept.

19. The system of claim 18, wherein each concept is a collection of signatures and metadata representing the concept.

20. The non-transitory computer readable medium according to claim 12 wherein each concept is a collection of signatures and is an abstract description of contents for which the signatures of the concept were generated.

21. The non-transitory computer readable medium according to claim 12 wherein the at least one input audio comprises a speech segment; wherein the method comprises generating at least one signature for the speech segment by breaking down the speech segment to multiple patches of random length and of random position in the speech segment.

22. The system of claim 13, wherein the system is further configured to select the second language based on past interactions of a user with multimedia.

23. The system of claim 13, wherein the system is further configured to select the second language based on a user profile.

24. The system according to claim 13 wherein each concept is a collection of signatures and is an abstract description of contents for which the signatures of the concept were generated.

25. The system according to claim 13 wherein the at least one input audio comprises a speech segment; wherein the system comprises a signature generator system that is configured to generate at least one signature for the speech segment by breaking down the speech segment to multiple patches of random length and of random position in the speech segment.