Transcoding (conversion). Adaptive transcoding: What is it? "Rosex" wants to create Russian chips for Bluetooth, Wi-Fi, NFC and Internet of things

From the video satellite is transmitted either in the MPEG-2 codec or in H.264 (it is AVC or MPEG-4 Part10). As a rule, MPEG-4 Part 10 is reduced to MPEG-4, but it is important not to confirm with MPEG-4 Part 2, which is absolutely not compatible and does not look like H.264 and used in old IP cameras.

Audio is transmitted to MPEG Audio Layer 2 (abbreviated MP2) or in AC3 (A / 52).

And it is important to understand that today H264 is usually compressed with Intra-Refresh, i.e. There are no reference frames in the video stream (IDR or KEYFRAME). This compression method allows you to smooth out the bitter jumps.

As a result, none of the audio or video options transmitted from the satellite is not played on the iPhone. Only H264 is played in the browser.

When transmitting via the Internet, as a rule, you can safely compress the video from MPEG2 in H264 with a three-time reduction in traffic.

When transferring HD channels via the Internet today you have to compress the flow into several different qualities: from HD with maximum quality to standard SD to compensate for overloaded channels.

As a result, the video from the satellite to provide high-quality OTT service should be transode into other codecs and quality.

It is important not to confuse the transcoding with the overpacking. Transcoding - an extremely resource-intensive operation, including:

• unpacking stream to coded video / audio
• decoding to raw video / audio
• resizing and other parameters
• coding back
• packaging in transport for stream

Packaging and unpacking relatively light operations, the styling server can handle up to 1000 channels on one computer. You can transcode on one computer from 1 to 30 channels, depending on the size and power of the computer.

For transcoding, you can use specialized dedicated devices, a central processor or video card: an external or built-in processor.

Specialized devices we will not consider, because in our mass this is either a computer with some program, or extremely expensive and very specialized equipment, or either simply unreasonably expensive device implemented exclusively by the manufacturer's marketing effort and not allowing -Lo significant results.

H.264.

To process video on CPU there are several different programs, but by and large today there are only two libraries that make sense to use for compression in the H.264 codec on the CPU: this is free LIBX264 and paid MainConcept. Everything else is either worse, or much worse, and both on the output, and on the use of resources.

Working with MainConcept in this article will not be considered, only LIBX264 will be mentioned

H.264 codec is a de facto standard for the video, because it is supported in all modern devices, except for some of the devices from Google.

There is practically no alternative. Today, H.265 appeared and develops, he already has a lot of support, but so far work with it is an investment in the future.

Google codecs: VP8 and VP9 are more desire for Google to drag the blanket on themselves, rather than something really useful. The resulting quality is worse, there is no support for hardware decoding, and therefore the price of the device is growing.

When encoding video, you need to understand what you have to balance between such parameters:

• delay inside an encoder in frames
• using CPU (how many milliseconds is required to compress one frame)
• weekend quality pictures (as far as pixel and which colors)
• output bitrate

For all types of ether, the use of CPU is absolutely critical. If the encoder settings require full CPU download or more, the video will not have time to be encoded in real time and therefore the video streaming will disappear.

For a VOD of such a hard restriction, there is no time for an hour in length, it is quite possible to encode three hours if you want to lower the bitrate. At the same time, for the essential video, it is usually still trying to use not all the power of the processor, to handle on one computer not 4 channels, and 10.

As for the delay inside the encoder, it is critical for video conferencing, but completely uncritted for IPTV. Even 5 seconds of delay in broadcasting television do not change the quality of service.

Bitrate and quality communication is quite clear: the more information about the picture we transmit, the better it will be displayed. Improve the quality of the picture by reducing the bit rate, usually can be able to select more productive compression tools that require greater delay and more clocks.

Understanding this challenging relationship is needed in order to better perceive the assurances that "our encoder is the best encoder in the world." It is necessary to compare at least 4 parameters, but as a result, everything comes down to: how much money is worth it and a month transcoding of one channel with the desired quality and output bit rate.

Flussonic Media Server for Transcoding

A separate package to Flussonic Media Server is a transcoder.

Flussonic Media Server can decode video from UDP / HTTP MPEG-TS, RTMP sources and encode it in several qualities and sizes.

This feature becomes needed when there is a need to show videos not only on consoles, but also on tablets: there the choice of available codecs is significantly less than on the console.

It is important to note that in order to play the video on the iPhone, it is necessary even the H264 from the satellite transodate, because as a rule, an intra-refresh coding mode is used for a smooth bitrate, creating a video that is not played on an iPhone.

Flussonic Media Server is more convenient than VLC or other options for organizing transcoding, because it is managed by one configuration file and automatically monitors the transcription state. VLC also requires writing a large number of monitoring scripts to track the transcoding state.

The next important feature of Flussonic Media Server for transcoding is automatic overbalance of streams when one of the servers fall. If one of 20 transcoders break at night, then the remaining transcoders can be configured to automatically capture the flows for transcoding, and the streamer it will take streams from backup transcoders.

Adaptive transcoding: What is it?

Such a term is called individual language mediation, which is carried out by a specialist of the translation agency. When adaptive transcoding, the information is translated from one language to another with simultaneous transformation according to the laws of intersective interaction.
Typically, adaptive transcoding requires attention to which language group or a specific form of informative change is present in the context. Therefore, adaptive transcoding allows you to select a transfer option that corresponds to the content of the original text. In this case, the text of the translation cannot be used for a 100% replacement of source text.
Translation has always been based on linguistic mediation. The source and end text should be equal and identical in meaning. Such a semblance of texts is required to achieve mutual understanding determined by linguistic features of communication.
Adaptive transcoding is characterized by a parapertoral character and allows you to comprehensively transform the text, which includes not only the usual translation, but also adapting text. The essence of adaptive transcoding consists in the method of making texts of various forms when orienting to the permissible style and the nature of the information and the required volume. The main information that is contained in the texts is carefully selected and regrouping.
These communicative text formats are distinguished by their own valid volume and defined rules for presentation of the material. Performance of translation facilitates the perception of text in accordance with them.

Need for transcoding video

Today, digital video compression technologies are important in almost all types of video applications. The significance of such parameters as compression and data compatibility is even more increasing due to the increasing trend towards the convergence of communications.
The most famous digital video applications include DVD, high-definition television (HDTV), video telephony / provision of teleconferencing and, recently, video surveillance. Each of these technologies has its own history of development, respectively, in each of them their own compression algorithms exist.
Transcoding plays two important roles. First, it provides communications between existing and newly emerging devices. For example, many existing video conferencing systems are based on the video data coding standard H.263. Newer video conferencing systems use the basic profile H.264 / AVC. Thus, to provide communication between these systems, there is a real-time trans-video transaction. Secondly, information networks, especially the Internet, have limited bandwidth when transmitting video. So, most of the videos are currently stored on DVD disks in MPEG2 format. Restrictions on the bandwidth in video services on request and streaming video on IP networks require the transformation of these video data into the format with a greater degree of compression. This is achieved by transcoding video in real time before transfer. In general, as a result of transcoding, up to 50% of the network bandwidth without loss of video quality is released.
Transcoding in video conferencing

So, one of the applications of transcoding is a video conferencing system. Consider a typical transcoding scheme used in such systems (Fig. 1). One signal processor (DSP2) decodes the input video stream and generates a recovered video frame, which is transmitted to another digital signal processor (in this example it is DSP1) through the Rapidio serial interface (SRIO). DSP1 encodes a reconstituted video frame to the desired format. Usually, on one side of the video conference, equipment based on H.263 standard is used, while the other side uses equipment based on H.264 standard.
The host processor controlling network traffic interacts with multiple digital signal processors (in this case with four) via the connection via the PCI bus.
The key feature of the interaction of processors in this example is their connection through the SRIO interface. Since the data transmitted between digital signal processors is uncompressed video, as a rule, with a frequency of 30 frames / s, then the requirements for the bandwidth of the communication line between devices are very high.
If you take a video in the standard NTSC resolution (720 per 480 pixels) YUV 4: 2: 0, then the size of each frame will be 720 × 480 × 1,5 \u003d 518400 bytes. Accordingly, at a frequency of 30 frames per second, the bandwidth of the line should be about 124 Mbps.
The SRIO interface selection dictates requirements for the speed of transmitting video data and supporting the flexible switching structure. SRIO supports three data rates: 1.24 Gbps, 2.5 Gbps and 3,125 Gb / s. In this interface, Serdes technology is used to restore the data stream clock synchronization and uses 8-b / 10-b encoding. This sequential interface specification supports ports with one line (1x) and with four lines (4x). The physical level of the SRIO interface defines the acknowledgment mechanism, which is used when establishing communication between devices, as well as error detection procedure based on cyclic redundant code. The physical level of the interface establishes and the priority of the packets used during routing within the commuting matrix.
To fully take advantage of SRIO bandwidth, processors must have these interfaces. Such processors offer TEXAS INSTRUMENTS. For example, the TMS320C6455 signal processor is equipped with a built-in SRIO interface, which provides four simultaneous compounds and has a peak data transfer rate of 20 Gb / s in both directions.
TMS320C6455 processor

In addition to the SRIO interface, the C6455 processor has an additional set of important functions that make it an ideal transaction device. These functional features can be combined into four main blocks.
The presence of a large number of high-speed I / O interfaces. System developers use different solutions, so the digital signal processor for video data processing applications should provide I / O ports to connect the system modules at the board level. As mentioned earlier, in C6455 there is a built-in SRIO port for communication between devices.
Other I / O variants in C6455 is an Ethernet Controller Controller (Ethernet Media Access Controller, EMAC) at a speed of 1 Gb / s, a 32-bit memory controller with a double data exchange rate (DDR2-500), as well as 66-MHz Tire for connecting peripheral devices (PCI). The built-in ATM interface (UTOPIA 2) allows you to use the C6455 processor in the telecommunications infrastructure.
Effective movement of data inside the chip. Singlery architecture for efficient data movement is one of the main advantages of the C6455 processor compared to its predecessors. In video data processing applications, digital signal processors operate as driven host processor devices. Therefore, there are high throughput, a small delay and the ability to parallel data transmission between the leading and driven devices. These requirements have identified the device architecture: peripherals, internal memory and processor core interact with each other through an efficient switch (Switched Central Resource - SCR) C6455 processor.
An optimal organization of data flow is also important. It was possible to improve the use of the memory bus with a bit of 256 bits and internal direct access to the memory (INTERNAL DIRECT MEMORY ACCESS, IDMA). IDMA provides data to the background in the background between the two internal memory levels, as well as to the bus peripherals and back.
Large amount of intraralium. Intracury Static SRAM RAM works much faster than the dynamic external SDRAM memory, and its volume is much less due to the high cost of manufacture. For typical applications in the video field, the intraccury memory serves mainly two targets: 1) Stores often used code and data, 2) loads / unloads temporary data before and after processing. As a rule, the greater the amount of available ingredic memory, the higher the performance of the application. In the C6455 digital signal processor, there are useful two megabytes of static RAM.
Software compatibility (software). Backward compatibility is important, since many programs for video applications have been developed long before the wide application of transcoding. To use the existing software on new processors, it is advisable to improve the DSP performance without changing its command set, and due to the processor kernel architecture. In the C6455 processor, two architectural innovations are implemented. The first is associated with the introduction of a cyclic buffer, which potentially improving the efficiency of software conveyorization of code for short cycles. The second is the application of 16-bit versions of the initial 32-bit commands, which significantly reduces the size of the program code and, thus, reduces the ratio of "non-payments" when cache-memory appeals.
Prototype of the transcoding system

Transcoding is also necessary for data transmission with DVDs on an IP network, for example, in the company's training system, video applications on request and broadcast broadcast video. In this case, the initial video format is MPEG2, and as a target format is used mainly WMV9. Note that the ability to program digital signal processors makes it easy to maintain almost any combination of source / target video format.
For video transcoding, you need to solve many technical issues, such as format conversion, a decrease in the bitrate of the video stream and its temporary and spatial resolution. Therefore, various schemes of intellectual transcoding of video data were developed. Their main principle is to the maximum possible reuse of information that is contained in the input video stream.
This section discusses the prototype of a video transcoding system, which is suitable for any transcoding schemes through the use of architecture based on the flexible hardware / software infrastructure. To satisfy various target video transcoding scenarios, the simplest transcoding scheme is selected, in which the video stream is fully decoded, and then encoded again in accordance with new restrictions.
The data stream in the system begins on the left side of the circuit (Fig. 2), with a compressed in MPEG2 format video file, which is stored on the hard disk, and ends on a flat-panel display, where the video is played by the Windows Media Player program. In this demonstration version, the video has a standard NTSC resolution (720 per 480 pixels) and is transcoded at a speed of 30 frames per second.
The flow receiver module operating on the DSP1, buffers the MPEG2 stream and organizes the input data for the MPEG2 decoder module. Data receiving operation is made using network development tools library (Network Development Kit, NDK), which is essentially a TCP / IP stack. The ASF Package Module operating on the DSP2 processor generates packets in ASF format from the data compressed in the WMV9 module. The DSP2 also has an HTTP server based on NDK, which processes the stream transmission requests from the Windows Media Player program and transmits the ASF packets into it. Windows Media Player decodes the ASF packets and shows the video on the screen.
One of the most interesting and complex aspects of data flow transmission is the interaction of two digital signal processors via the SRIO interface. When transmitting each video frame, the following occurs. After DSP1 completes the video edge transmission, it sends a data packet that is called Doorbell in the SRIO protocol specification. The Doorbell package generates a system interruption in the DSP2 processor notifying the presence of a frame. In response, DSP2 launches the encoding process in the WMV9 format. After completing the frame coding, DSP2 sends the Doorbell Package DSP1 processor. In this case, the DSP1 generates an interrupt to report the readiness of the DSP1 processor to continue the transmission of the next frame. In practice, the buffer circuit with an altered switching is used so that the coding / decoding and data transmission operations are performed in parallel.
The graph of a graphical user interface (GUI) provides built-in control and monitoring functions. SRIO communication channel activity and Gigabit Mac (GMAC) communication channels are displayed in real time. When the MPEG-2 data flow channel is transmitted, the average transmission rate is 8 Mbps, which is typical for coding with a standard resolution with a frequency of 30 frames per second. When transmitted over the ASF packet channel, the average transmission rate is 4 Mbps. This shows that WMV9 format is able to release approximately 50% bandwidth, providing similar video quality. For the SRIO communication channel, the average data transfer rate is 124 Mbps.

Thus, the capabilities of the Digital signal processor C6455 TI in combination with the SRIO interface, as well as the demonstration of the described prototype of the transcoding system based on C6455 processors indicate that the complex video transmission task in IP networks can be successfully solved and currently and in the future .

Transcoding means converting a file encoded by one method using another method. Transcoding can be made from Lossless in Losslese from Lossless in Lossy, from Lossy in Lossy, as well as from Lossy in Lossless.

For transcoding, use a converter, such as FOOBAR2000.

Lossy -\u003e Lossy

Each time a lossy encoder is encoded, the quality deteriorates. Also, there is no way to return former quality, even if you encode mp3 128 kbps in mp3 320 kbps (or in any other format with high quality).

For this reason, transcoding between Lossy formats is extremely recommended. The quality of the resulting file will be worse than the source file. Nevertheless, the reasons for such aole can be the following:

• Lowering a bitrate or conversion to another format for use with portable players, in which quality may not care too much.
• Savings on the hard disk. Uncompressed data with Audio CD have a bit rate of 1411 kbps (605 MB / hour); Lossless encoders allow you to reduce the flow on average up to 700 kbps (300 MB / hour). Lossy encoders, like Vorbis, MPC and AAC usually provide transparent sound on bitrates in the area of \u200b\u200b150-170 Kbps (69 MB / hour). For MP3 (in the case of using LAME), transparency is usually achieved with a bitrate of ~ 192 Kbps (82 MB / hour). For a large musical collection in such cases, savings relative to lossless compression may be significant.

Lossless -\u003e lossless

Unlike the aforementioned Lossy transcoding, in this case quality not lost. Thus, you can transform from one lossless format to another as many times as you need (for example, to increase the degree of compression or provide compatibility with specific programs / devices).

Lossless -\u003e Lossy

Archiving music in Lossless saves the possibility of further transcoding of music in another Lossy format (for example, in the event of new versions of the encoders). For example, if at the moment Lossy format x is transparent to 192 kbps, and after three years, Y format will appear, which will be transparent to 128 kbps - is hardly in this case transcoding [Email Protected] Kbps B. [Email Protected] Kbps will give an acceptable result, in contrast to encoding from Lossless. This is explained by the fact that for X format, since it is encoded with losses, some information regarded as not critical, is deleted, while in the opinion of the Y encoder, it is quite a modet to be important. As a result, the coding Y will be significantly distorted.

If you encode in Lossy from the source lossless, it is strongly recommended save the original lossless file. In this case, if the coding results are unsatisfactory, it is possible to recode the material again.

Please note: Some converters have the Options Options Options. Check it to be disabled.

Lossy -\u003e lossless

Quite often people think that they can improve sound quality by transcoding Lossy in Lossless (for example, MP3 in FLAC). In fact, the Lossy transformation in Lossless is absurd, because as soon as the material passed Lossy coding, by definition, the losses have already occurred, and they are irreversible. So, although you can transform from the Lossy format in Lossless (which, by the way, happens while playing a lossy file), the sound does not change, i.e., actually only its storage format changes.

If you ever have to perform such conversion, you must specify lossy origin in the file name (as well as, desirable, in tags) so that everyone who used this file immediately saw that it is received not from the original (lossless) source .

In which cases is used Lossless transcoding in Lossy:

1. Archiving audio, the source of which is an outdated or proprietary Lossy format, without loss of quality.
2. Editing audio that cannot be changed directly in the form of lossy.
3. As an intermediate format for encoding Lossy -\u003e Lossy.

To cut off the content issued for the original Lossless, users and administrators of file sharing services often use spectral analysis or special programs. Also these ways are used by customers of CDs to find out whether there was no encoding with losses in the process of creating and distribution of the material (which sometimes really takes place).

Information from sponsor

Remeshkov.ru: Online Strap Store for Watch. Hell you can buy a watch strap from leading manufacturers (Longines, Omega, Corum, etc.). Also on the store's website you can find out how to measure the watch strap.

Types of Language Mediation: Translation and Adaptive Transcoding.

Hell. transkode. - This is a type of language mediation, in which transcoding (transfer) of information from one language to another occurs (which takes place and when translating), and its transformation (adaptation) in order to set out it in another form determined by the intellectual communication task. Specificity of A.TransKode. Determined by the orientation of language mediation to a specific group receptor group or for a given form of conversion of information contained in the original.

A.T., like translation, is a special representation of the content of the original to the translating language, but unlike the translation, the text created is not intended for a full replacement of the original.

A.T., focused on the specified volume and nature of information, is carried out by drawing up annotations, abstracts, summary and other forms of transmission of information related to the selection and rearrangement of information contained in the input text. For each of these forms, an approximate volume and rules for presenting the material that facilitate the perception of transmitted information are set.

23. Types of semantic variation.

Within one method of describing the situation, various types of semantic variation are possible. The choice of a meaningful category, on the basis of which the situation will be described, does not fully determine the organization of the transmitted information.

Main types Seismant. Varyir. As part of the third type of equivalent: the most common types of varying: 1) the degree of detail of the description; 2) the method of combining the described features in the report; 3) the direction of relations between the signs; 4) Distribution of individual features in the message. Degree of detail. descriptions. Description of the situation in the chosen way may carried out. with large or smaller details. As a result, some signs in some posts will be named, and in others will remain only implied. The method of combining the described features. Along with the phenomena, common to all languages, each language imposes its limitations on the possibility of a combination of individual concepts as part of the message. Direction of relations between signs. When describing a situation with different points of view, synonymous message. Could be related to conversion relationships: (Professor takes the exam in students. - Students pass the exam by the professor). The extreme case of such a difference is the relationship of the opposite: (he always remembers about it. "He never forgets about it)." Converse rephrase is elected by a translator for stylistic considerations. Distribution of individual features in the message. Equivalent messages related to the same method of describing the situation may differ from each other and the distribution of features in separate parts of the message. The possibility of combining and the sequence of describing the signs is sometimes unequal in different languages.

24. Equivalence of the connotative value of correcelated words. The equivalence of connotative values \u200b\u200bin correlated words in the original and translation also implies playback in the translating an associative-shaped component of this value. The semantics of some words includes additional information associated with certain associations in consciousness of speaking. For the inhabitants of many countries, snow is not just a kind of precipitation, but also the whiteness of white, with which it is customary to compare other white (snow-white) items (hair, sugar, lingerie, etc.). Chalk is also white, but with him you can compare only the color of the pale face. Russian slip It is used for a figurative description of a person, and in the semantics words needledenoting the thing much more subtle, there is no component causing such associations. Thanks to the value component, the word produces a special impact on the receptor, its semantics perceived with more readiness, attracts attention, causes an emotional attitude. The preservation of the formation of the original may be a prerequisite for achieving the equivalence of translation. Here you can note three different degrees of the proximity of the form of two languages:

25. Translation of the associative component of the word

26. Pragmatic adaptation of source text.

In the process of transfer, along with comparison of various language systems, there are compared of different crops. As a rule, texts addressed to the original language carrier are calculated only on its perception. Do they completely come completely and? Specific features of his psychology, accessible by the amount of information, the characteristics of the surrounding socio-cultural sphere. In the process of translation, the text is redirected by an input to the recipient, which has a different volume of background knowledge. In this case, there is a pragmatic adaptation of the source text, i.e. Making certain amendments to socio-cultural, psychological and other differences between the recipients of the original text and the text of the translation.

Realities- These are the concepts relating to life, everyday life, traditions, history, material and spiritual culture of this nation.

Among the techniques used for the transfer of realities should first of all stay on transliteration and calculation.

Unconscious or unfamiliar realities require inclusion in the text of additional explanatory elements.

Explanatory elements do not have to follow after the realities, they can precede it. Such a reception as it will prepare the reader to meet with an invaluable word, makes this word more convenient for the perception of transliteration and tracing, accompanied by explanatory elements, are used in cases where the translator seeks to cause the reader a sense of national flavor or the concepts that are denoted by the realities are the subject of communication and therefore Can be omitted. To this method, however, should not be resorted too often. The fact is that the repeated use of deployed explanatory comments within the same text is unreasonably expanding its volume, makes it cumbersome and verbose. The translator should first make sure that transliteration with the explanatory comment is really necessary. If there is no need for particular need, it may decide on the removal of Russian realities and the use of another method of translating a functional analogue. The essence of this technique is that the same substantive situation is depicted in a translating language based on various, although interrelated signs. Semantic shift, which occurs with such a translation, is similar to those described in Section. 1. Here are the same translation transformation - generalization, concretization and metonymia.

27. Types of pragmatic text adaptation:

Pragmatics transfer - Impact on the course and result of the translation process, the need to reproduce the pragmatic potential of the original and ensure the desired impact on the recipient.

Pragmatic adaptation - Changes introduced into the text of the translation in order to achieve the necessary reaction from the recipient. However, the belonging of the transfer receptor to another language team, to other culture, often leads to the fact that the equivalent translation is pragmatically inadequate. In this case, the translator has to resort to pragmatic adaptation of the translation, entering into its text the necessary changes.

In translation practice, four types of such adaptation are most common. The first type of pragmatic adaptation is to ensure an adequate understanding of the message receptors. If the change in the above translations provided an adequate understanding of the transmitted message, then the second type of pragmatic adaptation is intended to achieve the correct perception of the original content, convey to the transfer receptor emotional impact of the source text. . In contrast to the previous ones in this case, the translator is not focused on averaged, but on a specific receptor and on a specific situation of communication, seeking to ensure the desired impact. Therefore, such adaptation is usually associated with a significant deviation from the source message. The fourth type of pragmatic adaptation can be described as a solution of "extra-superfolding".

28. Modeling the translation process

Modeling - This is a scientific receipt consisting in schematic reproduction of an object or direct observation, or differing great complexity.

Model - This is an auxiliary object that replaces the studied object presented in the most general form.

The real translation process is carried out in the translator's brain and is unavailable for direct observation and research. Therefore, the study of the translation process is made indirectly by developing various theoretical models, with a greater or less approximity of describing the translation process as a whole or any of its direction.

Model translation called conditional opisinging a number of thought operations, performing whichvodter can transfer the entire original or some part of it. In the linguistic theory of translation models of the translation present the process of translation in the form of a number of thought operations over linguistic or speech units, i.e. In the form of linguistic operations, the choice of which is determined by the original characteristics of the original and the corresponding phenomena in the translation language.

The translation model is conditional, since it does not necessarily reflect the real actions of the translator during the process of creating the text of the translation. Most of these models have limited explanationand does not pretend that on their basiscan really be realized to transfer any text withwhat is the degree of equivalence.The tasks of the model concludeonly to describe the sequence of actions, withusing which you can solve this translationchu under the set conditions for the translation process. Translation models disclose separate parties to the functioning of the linguistic translation mechanism. Although the translator can achieve the necessary result and in any way that does not coincide with any of the conversion models known to us, knowledge of such models can help him in solving difficult translation problems.