KR102020773B1 - Multimedia Speech Recognition automatic evaluation system based using TTS - Google Patents

Abstract

The present invention relates to an automatic multimedia speech recognition evaluation system using a speech synthesis engine, and more particularly, to an automatic multimedia speech recognition evaluation system using a speech synthesis engine, which uses a TTS engine (speech synthesis engine) to play and evaluate new sentence patterns in real time without recording in order to improve the problems, of an existing automatic speech recognition evaluation system which sequentially plays pre-recorded speech data to evaluate the recognition rate of a speech recognition apparatus, such as enormous time and cost due to the continuous construction of a speech recording DB and the possibility of not recognizing other sentences with the same meaning, thereby outputting accurate and fast performance and functional verification results. The automatic multimedia speech recognition evaluation system according to the present invention comprises: an automatic speech recognition test device (100); a TTS driving device (200); and a speech recognition engine device (300).

Description

Multimedia Speech Recognition automatic evaluation system based using TTS}

The present invention relates to a system for automatically evaluating multimedia speech recognition using a speech synthesis engine, and more particularly, to a speech recording DB of a general speech recognition automatic evaluation system for sequentially reproducing pre-recorded speech data to evaluate a recognition rate of a speech recognition apparatus. In order to improve the problems such as huge time and cost, and the possibility of not recognizing other sentences with the same meaning, it is possible to use TTS engine (voice synthesis engine) at any time to evaluate and play new sentences in real time without recording. By providing a speech recognition automatic evaluation system, the present invention relates to a multimedia speech recognition automatic evaluation system using a speech synthesis engine that can output accurate and fast performance and function verification results.

Speech Recognition technology allows a computer to analyze a user's voice input through a microphone, extract features, recognize a result that is close to a pre-entered word or sentence as a command, and perform an operation corresponding to the recognized command. It is a technique to do.

Conventional voice recognition system has a terminal voice recognition method in which a voice recognition engine is stored inside a terminal such as a vehicle and a mobile device, and a cloud-based server voice recognition for smartphone internet voice search and various information processing is suitably adapted to each service purpose. Has been used.

In particular, the existing speech recognition automation evaluation system has been progressed mainly on the speech DB automatic output device for measuring the speech recognition rate, or the adjusting device for adjusting the noise environment.

However, recently, voice recognition requires integrated performance verification for a hybrid method in which both the terminal voice recognition and cloud-based server voice recognition are simultaneously operated. This is necessary.

For example, in the case of a voice recognition system for a vehicle, a method in which a multilingual native speaker is directly boarded in a vehicle in a real vehicle high-speed driving environment, instructs a user to utter a predetermined command, and the inspector is accompanied by a passenger to manually check the recognition result.

In other words, the invention of the existing automatic speech recognition test usually does not take into consideration the actual vehicle environment, but includes a speech recognition system on a PC, and automatically inputs recognition target vocabularies, and batches the results, and creates a test environment. The focus was on technology such as a volume control device that automatically adjusts the ratio of visual noise and voice.

However, the actual vehicle test method has several hundreds of native speakers, problems with induction and management up to the test site, safety problems due to high-speed driving conditions, efficiency reductions due to the record of recognition results, huge result data purification and analysis time. In the face of various real-world problems such as excessive requirements and non-repeatable tests, it was difficult to perform enough tests to produce statistically realistic results.

On the other hand, the prior art related to the present invention will be described in more detail.

Since human speech is required to automatically evaluate speech recognition, it is time-consuming and expensive to bring many people directly to the test site for testing.

Therefore, after recording voice, a method of outputting and testing a recording sound source is generally utilized.

However, if the recording of the sound source also increases the number of languages to be tested, and the test target command is continuously changed, a new sound source DB must be constructed, which is a very difficult problem.

In particular, when natural language recognition is generalized, all variable sentences cannot be recorded when natural random sentences need to be tested. Therefore, a new automatic speech recognition evaluation system is required.

For example, the speech recognition automatic evaluation system sequentially reconstructs the prerecorded voice recording data to evaluate the recognition rate of the speech recognition apparatus.

Efficient long-term testing can be performed efficiently without the tester's work, but only limited voice recording data can be used to create a suitable voice recording data when new features are added, and it is impossible to find a sentence with a high probability of recognition failure. This rises drastically.

For example, if a speech recognition device that only recognizes media and call-related voice commands is upgraded to recognize navigation-related voice commands, the automatic speech recognition evaluation system must build navigation-related voice recording data. .

In addition, even if all the patterns of the voice recording data are successfully recognized, other sentences having the same meaning, for example, 'guide to the office' and 'find the office' for the intention of utterance of 'office search' Other sentences are more likely to fail recognition.

Therefore, there is a need to develop a new system for discovering a sentence having a high probability of recognition failure by testing a technique for reducing the voice recording data construction cost and testing various types of sentences.

In particular, in the case of natural language recognition, it is necessary to continuously build a voice recording DB, which requires enormous time and cost, and a system for improving it is urgently needed.

(Previous Document) Republic of Korea Patent Publication No. 10-2013-0029635

Therefore, the present invention has been proposed in view of the problems of the prior art as described above, and a first object of the present invention is to recognize a large amount of time and cost required for continuous construction of a voice recording DB for natural language recognition, and to recognize other sentences having the same meaning. In order to improve the problems such as the possibility of not being able to do so, TTS engine (voice synthesis engine) is used to provide a voice recognition automatic evaluation system that can evaluate and play new sentences in real time at any time without recording.

A second object of the present invention is to provide a significant advantage in recognition accuracy even in a variable test environment when evaluating natural language recognition, compared to conventional simulation test methods.

In order to achieve the problem to be solved by the present invention, a multimedia speech recognition automatic evaluation system using a speech synthesis engine,

When the test script information including the voice recognition correct answer value is received, the TTS driving apparatus 200 is driven and controlled, and the voice recognition result value provided from the voice recognition engine device 300 is compared with the voice recognition correct answer value of the test script. The automatic voice recognition automatic test apparatus 100 for determining whether the recognition is successful and outputting the recognition success result;

TTS driving device 200 for converting the test script provided from the automatic voice recognition automatic test device 100 to the voice, and outputs the converted voice,

And a voice recognition engine device 300 for acquiring a voice output through the TTS driving device to generate a voice recognition result value and to provide the voice recognition automatic test device 100.

Through the automatic speech recognition system for multimedia recognition using the speech synthesis engine according to the present invention having the above-described configuration and operation, it is very useful in evaluating natural language recognition and provides a considerable advantage even in a variable test environment than the conventional simulation test method. It becomes possible.

In addition, it is possible to reduce the cost of constructing voice recording data and to test sentences of various types of sentences to find a sentence having a high probability of recognition failure.

In particular, in the case of natural language recognition, it eliminates the enormous time and cost required to continuously build a voice recording DB. The instantaneous efficiency, efficiency, and cost savings effect can be evaluated by real-time playback of new sentence patterns without recording at any time. Will be provided at the same time.

1 is an overall configuration diagram schematically showing a multimedia speech recognition automatic evaluation system using a speech synthesis engine according to an embodiment of the present invention.
Figure 2 is a block diagram of the automatic speech recognition automatic test apparatus 100 of the automatic multimedia speech recognition evaluation system using a speech synthesis engine according to an embodiment of the present invention.
Figure 3 is a block diagram of the TTS driving apparatus 200 of the automatic multimedia speech recognition system using a speech synthesis engine according to an embodiment of the present invention.
Figure 4 is a block diagram of a speech recognition engine 300 of the automatic speech recognition system for multimedia recognition using a speech synthesis engine according to an embodiment of the present invention.

The following merely illustrates the principles of the invention. Therefore, those skilled in the art, although not explicitly described or illustrated herein, can embody the principles of the present invention and invent various devices that fall within the spirit and scope of the present invention.

In addition, all conditional terms and embodiments listed herein are in principle clearly intended to be understood only for the purpose of understanding the concept of the invention and are not to be limited to the specifically listed embodiments and states. do.

In describing the present invention, terms such as first and second may be used to describe various components, but the components may not be limited by the terms.

For example, without departing from the scope of the present invention, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component.

When a component is referred to as being connected or connected to another component, it may be understood that the component may be directly connected to or connected to the other component, but there may be other components in between. .

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention, and singular forms may include plural forms unless the context clearly indicates otherwise.

In this specification, the terms including or including are intended to designate that there exists a feature, a number, a step, an operation, a component, a part, or a combination thereof described in the specification, and one or more other features or numbers, It can be understood that it does not exclude in advance the possibility of the presence or addition of steps, actions, components, parts or combinations thereof.

The following merely illustrates the principles of the invention. Therefore, those skilled in the art, although not explicitly described or illustrated herein, can embody the principles of the present invention and invent various devices that fall within the spirit and scope of the present invention.

In addition, all conditional terms and embodiments listed herein are in principle clearly intended to be understood only for the purpose of understanding the concept of the invention and are not to be limited to the specifically listed embodiments and states. do.

Means for solving the problems of the present invention are as follows.

That is, the multimedia speech recognition automatic evaluation system using the present inventors speech synthesis engine,

When the test script information including the voice recognition correct answer value is received, the TTS driving apparatus 200 is driven and controlled, and the voice recognition result value provided from the voice recognition engine device 300 is compared with the voice recognition correct answer value of the test script. The automatic voice recognition automatic test apparatus 100 for determining whether the recognition is successful and outputting the recognition success result;

TTS driving device 200 for converting the test script provided from the automatic voice recognition automatic test device 100 to the voice, and outputs the converted voice,

It is characterized in that it comprises a voice recognition engine device 300 for obtaining the voice output through the TTS driving device to generate a voice recognition result value to provide to the voice recognition automatic test device 100.

At this time, the voice recognition automatic test device 100,

A test script input unit 110 for receiving test script information including a voice recognition correct answer value for a voice recognition test;

The test script information input through the test script input unit is stored in the test information storage unit 130, and converted into a format that can be driven by the TTS engine unit 220 to convert the converted information into the TTS information storage unit 210. A script analysis unit 120 for storing and processing;

A test information storage unit 130 including a test script provided through the script analysis unit, a voice recognition correct answer value of the test script, and a test number;

To compare the speech recognition correct answer value of the test script information stored in the test storage unit with the speech recognition result value provided from the speech recognition engine device 300 to determine whether the recognition is successful, and to provide the recognition success information to the speech recognition result output unit. Voice recognition result processing unit 140;

And a voice recognition result output unit 150 for outputting a recognition success result provided by the voice recognition result processing unit.

At this time, the script analysis unit 120,

The reference information corresponding to the voice recognition correct answer value of the test script, the speaker information necessary for the TTS synthesis, and the test number are extracted and stored in the test information storage unit.

At this time, the TTS driving device 200,

A TTS information storage unit 210 for storing the converted information converted into a driveable format provided by the script analyzing unit 120;

A TTS engine unit 220 for extracting the converted information stored in the TTS information storage unit 210 and converting the converted information into voice (TTS synthesis) to provide the test voice output unit 230;

And a test voice output unit 230 for outputting the converted voice provided from the TTS engine unit 220.

At this time, the voice recognition engine device 300,

A voice recognition engine 310 for obtaining a voice output from the test voice output unit 230 to generate a voice recognition result value and providing the voice recognition result acquisition unit to the voice recognition result acquisition unit;

A voice recognition result acquisition unit 320 for storing a voice recognition result value provided through the voice recognition engine unit and providing the voice recognition result value to the communication unit 330;

And a communication unit 330 for transmitting the voice recognition result value provided through the voice recognition result acquisition unit to the voice recognition result processing unit 140.

Hereinafter, it will be described in detail through an embodiment of the automatic multimedia speech recognition system using the speech synthesis engine according to the present invention.

1 is an overall configuration diagram schematically showing a multimedia speech recognition automatic evaluation system using a speech synthesis engine according to an embodiment of the present invention.

As shown in FIG. 1, a voice recognition automatic test apparatus 100, a TTS driving apparatus 200, and a voice recognition engine apparatus 300 are largely configured.

Specifically, the voice recognition automatic test apparatus 100 controls the driving of the TTS driving apparatus 200 when receiving test script information including a voice recognition correct answer value.

In order to control the driving of the TTS driving apparatus 200, the TTS driving apparatus 200 is converted into a format that can be driven, and when the converted information is provided to the TTS driving apparatus 200, the corresponding TTS driving apparatus 200 is Therefore, it is converted into voice.

The technology for converting the TTS driving apparatus into a format that can be driven is a general technology, and is described in, for example, a system and method for converting a text, which is Korean Patent Publication No. 10-2006-0051151, into speech. The description is omitted.

At this time, by comparing the speech recognition result value provided from the speech recognition engine device 300 with the speech recognition correct answer value of the test script, it is determined whether the recognition is successful and outputs the result of the recognition success.

For example, if the voice recognition correct answer value of the test script information is' hello 'and the voice recognition result value is' hello', the result is' successful recognition ', and if the voice recognition result value is' If it is not the same, it will output the result of 'recognition failure' because it is not the same.

In addition, the TTS driving apparatus 200 converts the test script provided from the automatic voice recognition test apparatus 100 into voice, and outputs the converted voice.

For example, you might get a test script called "hello", turn it into a voice, and output the converted voice.

That is, it outputs a voice of 'hello'.

In addition, the voice recognition engine device 300 obtains the voice output through the TTS driving apparatus, generates a voice recognition result value, and provides the voice recognition automatic test apparatus 100.

For example, the output 'hello' is to be obtained, and 'hello' is generated as a voice recognition result value to be provided to the automatic voice recognition test apparatus 100.

At this time, the speech recognition automatic test device 100 compares the information 'hello', which is the speech recognition result value provided from the speech recognition engine device 300, with the speech recognition correct answer value of the test script 'hello' to determine whether the recognition success. In the example, the result is 'recognition success' because it matches.

Hereinafter, the voice recognition automatic test apparatus 100, the TTS driving apparatus 200, and the voice recognition engine apparatus 300 will be described in detail with reference to the accompanying drawings.

2 is a block diagram of an automatic speech recognition automatic test apparatus 100 of a multimedia speech recognition automatic evaluation system using a speech synthesis engine according to an embodiment of the present invention.

As shown in FIG. 2, the automatic voice recognition test apparatus 100 includes a test script input unit 110, a script analysis unit 120, a test information storage unit 130, a voice recognition result processing unit 140, and a voice. It is configured to include a recognition result output unit 150.

Specifically, the test script input unit 110 performs a function of receiving test script information including a voice recognition correct answer value for a voice recognition test.

The test script information may be input from various multimedia devices.

The test script information is a text file and includes a test script, a voice recognition correct answer value of the test script, and a test number.

The test script information may mean, for example, '#test no-1 # test script-love # Speech recognition correct value-love'.

In addition, the script analyzing unit 120 stores and processes the test script information input through the test script input unit in the test information storage unit 130, and converts the test script information into a format that can be driven by the TTS engine unit 220. The information storage unit 210 stores the converted information.

That is, the script analyzing unit 120 stores and processes the test script information input through the test script input unit in the test information storage unit 130. For example, '#test no-1 # test script-love # Speech It will store the recognition correct value-love.

In addition, the conversion information is converted into a format that can be driven by the TTS engine unit 220 to store the converted information in the TTS information storage unit 210.

The conversion information refers to a drive format capable of converting a text-to-speech (TTS) in the TTS engine unit 220, and converts the conversion information into a TTS information storage unit 210 by processing of the script analysis unit 120. Will be saved).

For example, the body of a test script can be parsed before being converted to speech.

The test script can be parsed into sections, chapters, pages, paragraphs, sentences and / or portions thereof, such as fragments (eg, based on punctuation and other grammar rules), or words or letters. have.

Each part can be analyzed to determine if it has one or more specific attributes that can represent a context (eg, linguistic context).

For example, is the text part indented, bullet point preceded, italic, bold, underlined, double underlined, subscripted, superscripted, no specific punctuation, or include specific punctuation? , Have a specific font size compared to other font sizes in the text, are all uppercase letters, title characters, are justified in a certain way (for example, right-aligned, centered, left-aligned or justified), headings Whether it is at least part of a header, at least part of a header or footer, at least part of a table of contents (TOC), at least part of a footnote, has a different attribute, or has any combination of the attributes described above. have.

The conversion of the text portion to speech can be controlled based on this attribute, for example, by setting one or more conversion parameter values for the text.

For a given text part, any of the conversion parameters, such as volume, intonation rate, voice accent, voice pulsation, syllable emphasis, pause before and / or after that part, other parameters, and any suitable combination thereof Values for may be set.

Values for any of these parameters can be set, which are sent to the TTS engine section with the given text portion.

For example, a programming call can be made in a standard Speech API (SAPI) for each piece of text, including setting values for specific SAPI parameters.

The text can be selected by the user, for example, a word processing (eg Microsoft® Word) document, a spreadsheet (eg Excel ™) document, a presentation (eg PowerPoint®) ) A digital document, such as a document, an email (eg, Outlook®) message, or another type of document.

Alternatively, the text may be part of a document such as, for example, part of any of the above.

The test information storage unit 130 may store the test script included in the test script information provided through the script analysis unit, the voice recognition correct answer value of the test script, and the test number.

In addition, the script analysis unit 120,

Reference information corresponding to the voice recognition correct answer value of the test script, speaker information necessary for TTS synthesis, and a test number may be extracted and stored in the test information storage unit.

For example, reference information may refer to site information or company information that provides a voice recognition correct answer value, and speaker information necessary for TTS synthesis may include adult female, adult male, entertainer 1, entertainer 2, young female child, and the like. The same may mean various speaker information, and in the case of a test number, it may mean a unique number that can be identified to summarize success and failure.

In this case, the speech recognition result processing unit 140 compares the speech recognition correct answer value of the test script stored in the test storage unit with the speech recognition result value provided from the speech recognition engine device 300 to determine whether the recognition is successful or not. It performs a function for providing to the voice recognition result output unit.

For example, the voice recognition correct value 'love' is extracted from the '#test no-1 # test script-love # Speech recognition correct value-love' stored in the test storage unit, and the voice recognition engine 300 The 'love' is compared in the 'speech recognition mode execution signal-speech recognition mode execution signal #test no-1 # STT-love' which is the provided speech recognition result.

In the above case, it is determined whether the recognition is successful because it matches, and this is provided to the voice recognition result output unit.

Accordingly, the voice recognition result output unit 150 outputs a result of the recognition success provided by the voice recognition result processing unit.

For example, 'Test No. 1 love is recognized as a result of voice recognition, recognition succeeded.' Will be displayed on the screen.

On the other hand, according to an additional aspect, the automatic voice recognition test apparatus 100 is a result processing unit for obtaining the recognition success information from the voice recognition result processing unit 140 and aggregates the recognition result of the test script input at the set time; It can be configured to include more.

For example, the recognition success rate for each test script input for 1 hour can be counted. If the test number 1-success, test number 2-failure, test number 3-success, test number 4- failure, the recognition success rate is 50. Will be%.

On the other hand, according to another additional aspect, the voice recognition automatic test apparatus 100 may further comprise a voice recognition evaluation setting unit for setting the volume value, the noise volume value of the output voice for voice evaluation. .

For example, the voice recognition evaluation setting unit sets various environments of the voice output for the voice recognition test.

The set values are provided to the test voice output unit 230 to perform the voice output at the set values.

For example, setting values such as the evaluation output voice volume -10 and the noise volume -2 are provided.

As described above, the reason for setting the environment variable is that the environment to which the actual speech recognition is applied is not an environment without noise, so that the speech recognition is provided by providing a setting value to set the state suitable for the actual environment. To increase the success rate.

3 is a block diagram of a TTS driving apparatus 200 of a multimedia voice recognition automatic evaluation system using a voice synthesis engine according to an embodiment of the present invention.

As shown in FIG. 3, the TTS driving apparatus 200 includes a TTS information storage unit 210, a TTS engine unit 220, and a test voice output unit 230.

Specifically, the TTS information storage unit 210 stores the converted information converted into a driveable format provided by the script analysis unit 120.

In this case, the TTS engine 220 extracts the conversion information stored in the TTS information storage unit 210 and converts the converted information into voice (TTS synthesis) to provide the test voice output unit 230.

The test number 'love' is converted into voice, that is, TTS synthesized and provided to the test voice output unit, and the test voice output unit 230 outputs the converted voice provided from the TTS engine unit 220. Will be.

For example, the TTS engine unit 220 converts the voice data into an audio file having a format that can be output from the test voice output unit.

That is, since the format of the audio file that can operate in a specific function is different depending on the manufacturer of the test audio output unit and the model of the test audio output unit, detailed description of the format of the audio file is omitted.

At this time, the test voice output unit 230 is characterized in that the audio reproduction means, for example, it means a speaker 270.

The test voice output unit 230 transmits a voice recognition mode execution signal to the voice recognition engine 300 when the voice is output.

Therefore, the voice recognition engine 300 operates automatically at the time when the voice is output to recognize the output voice.

4 is a block diagram of a speech recognition engine device 300 of the automatic speech recognition system for multimedia recognition using a speech synthesis engine according to an embodiment of the present invention.

As shown in FIG. 4, the voice recognition engine device 300 includes a voice recognition engine 310, a voice recognition result acquisition unit 320, and a communication unit 330.

Specifically, the voice recognition engine 310 performs a function of obtaining a voice output from the test voice output unit 230 to generate a voice recognition result value and providing the result to the voice recognition result acquisition unit.

In order to obtain the voice output from the test voice output unit 230, the voice recognition engine may further comprise a voice acquisition device.

For example, as a voice acquisition device, this means a microphone 370.

In this case, the voice recognition engine 310 obtains the voice output from the test voice output unit 230 through the microphone to generate a voice recognition result value.

The speech recognition engine refers to a speech to text (STT) engine, and performs a function of converting a speech acquired by a microphone into text, and the converted text means a speech recognition result.

For example, when the voice acquired through the microphone is 'love', when the STT conversion is performed through the voice recognition engine, the voice recognition value 'love' is generated.

The voice recognition result acquisition unit 320 stores the voice recognition result provided through the voice recognition engine and provides the voice recognition result to the communication unit 330.

For example, the voice recognition result of '#speech recognition mode execution signal-speech recognition mode execution signal # test no-1 # STT-love' is provided to the communication unit.

Accordingly, the communication unit 330 transmits the voice recognition result value provided through the voice recognition result acquisition unit to the voice recognition result processing unit 140.

That is, the voice recognition result value '#speech recognition mode execution signal-speech recognition mode execution signal # test no-1 # STT-love' is transmitted.

As described above, if the automatic speech recognition system using the speech synthesis engine is used, it provides instantaneous efficiency and efficiency and cost reduction effects that can be evaluated by real-time playback of new sentence patterns without recording.

That is, in the case of natural language recognition, which is a problem of the prior art, it eliminates enormous time and cost required to continuously build a voice recording DB.

According to the present invention, it is possible to provide a significant advantage in evaluating natural language recognition and a considerable advantage even in a variable test environment over the conventional simulation test method.

In addition, it is possible to reduce the cost of constructing voice recording data and to test sentences of various types of sentences to find a sentence having a high probability of recognition failure.

In particular, in the case of natural language recognition, it eliminates the enormous time and cost required to continuously build a voice recording DB. The instantaneous efficiency, efficiency, and cost savings effect can be evaluated by real-time playback of new sentence patterns without recording at any time. Will be provided at the same time.

In addition, while the above has been shown and described with respect to preferred embodiments of the present invention, the present invention is not limited to the specific embodiments described above, the technical field to which the invention belongs without departing from the spirit of the invention claimed in the claims. Of course, various modifications can be made by those skilled in the art, and these modifications should not be individually understood from the technical spirit or the prospect of the present invention.

100: automatic voice recognition device
200: TTS driving device
300: voice recognition engine device

Claims (5)

In the automatic evaluation system of multimedia speech recognition using speech synthesis engine,

When the test script information including the voice recognition correct answer value is received, the TTS driving apparatus 200 is driven and controlled, and the voice recognition result value provided from the voice recognition engine device 300 is compared with the voice recognition correct answer value of the test script. The automatic voice recognition automatic test apparatus 100 for determining whether the recognition is successful and outputting the recognition success result;
TTS driving device 200 for converting the test script provided from the automatic voice recognition automatic test device 100 to the voice, and outputs the converted voice,
It is configured to include a voice recognition engine device 300 for obtaining the voice output through the TTS driving device to generate a voice recognition result value to provide to the voice recognition automatic test device 100,

The voice recognition automatic test device 100,
A test script input unit 110 for receiving test script information including a voice recognition correct answer value for a voice recognition test;
The test script information input through the test script input unit is stored in the test information storage unit 130, and converted into a format that can be driven by the TTS engine unit 220 to convert the converted information into the TTS information storage unit 210. A script analysis unit 120 for storing and processing;
A test information storage unit 130 including a test script provided through the script analysis unit, a voice recognition correct answer value of the test script, and a test number;
Speech recognition result output unit 150 compares the speech recognition correct answer value of the test script information stored in the test information storage unit with the speech recognition result value provided from the speech recognition engine device 300 to determine whether the recognition is successful. Speech recognition result processing unit 140 for providing;
Automatic speech recognition system using a speech synthesis engine, characterized in that it comprises a; voice recognition result output unit for outputting the recognition success result provided by the speech recognition result processing unit 140.
delete The method of claim 1,
The script analysis unit 120,
Automatic speech recognition system using a speech synthesis engine, characterized in that for extracting the reference information corresponding to the correct answer value of the speech recognition of the test script, the speaker information and the test number necessary for the synthesis of the TTS and store it in the test information storage.
The method of claim 1,
The TTS driving device 200,
A TTS information storage unit 210 for storing the converted information converted into a driveable format provided by the script analyzing unit 120;
A TTS engine unit 220 for extracting the converted information stored in the TTS information storage unit 210 and converting the converted information into voice (TTS synthesis) to provide the test voice output unit 230;
And a test voice output unit 230 for outputting the converted voice provided from the TTS engine 220. The automatic voice recognition system for multimedia recognition using a speech synthesis engine.
The method of claim 1,
The voice recognition engine device 300,
A voice recognition engine 310 for acquiring a voice output from the test voice output unit 230 of the TTS driving apparatus 200 to generate a voice recognition result value and to provide the voice recognition result acquisition unit to the voice recognition result acquisition unit;
A voice recognition result acquisition unit 320 for storing a voice recognition result value provided through the voice recognition engine 310 and providing the voice recognition result value to the communication unit 330;
Multimedia speech recognition using a speech synthesis engine, characterized in that it comprises a; communication unit 330 for transmitting the speech recognition result value provided through the speech recognition result acquisition unit 320 to the speech recognition result processing unit 140 Automatic evaluation system.

Images