CN106356053A - Method and device for testing recognition accuracy of voice input method and electronic equipment - Google Patents
Method and device for testing recognition accuracy of voice input method and electronic equipment Download PDFInfo
- Publication number
- CN106356053A CN106356053A CN201610648394.0A CN201610648394A CN106356053A CN 106356053 A CN106356053 A CN 106356053A CN 201610648394 A CN201610648394 A CN 201610648394A CN 106356053 A CN106356053 A CN 106356053A
- Authority
- CN
- China
- Prior art keywords
- sentence
- text
- correct text
- speech data
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 94
- 238000012360 testing method Methods 0.000 title claims abstract description 94
- 238000012545 processing Methods 0.000 claims description 19
- 238000010998 test method Methods 0.000 claims description 12
- 230000003203 everyday effect Effects 0.000 claims description 7
- 230000014509 gene expression Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 10
- 238000004590 computer program Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 241001672694 Citrus reticulata Species 0.000 description 6
- 238000004891 communication Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 3
- 229940104697 arixtra Drugs 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- KANJSNBRCNMZMV-ABRZTLGGSA-N fondaparinux Chemical compound O[C@@H]1[C@@H](NS(O)(=O)=O)[C@@H](OC)O[C@H](COS(O)(=O)=O)[C@H]1O[C@H]1[C@H](OS(O)(=O)=O)[C@@H](O)[C@H](O[C@@H]2[C@@H]([C@@H](OS(O)(=O)=O)[C@H](O[C@H]3[C@@H]([C@@H](O)[C@H](O[C@@H]4[C@@H]([C@@H](O)[C@H](O)[C@@H](COS(O)(=O)=O)O4)NS(O)(=O)=O)[C@H](O3)C(O)=O)O)[C@@H](COS(O)(=O)=O)O2)NS(O)(=O)=O)[C@H](C(O)=O)O1 KANJSNBRCNMZMV-ABRZTLGGSA-N 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 241000196324 Embryophyta Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000006386 memory function Effects 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 210000000697 sensory organ Anatomy 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/01—Assessment or evaluation of speech recognition systems
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Machine Translation (AREA)
Abstract
The application provides a method and a device for testing the recognition accuracy of a voice input method and electronic equipment, wherein the method comprises the following steps: acquiring test data of a test sample set, wherein the test data comprises voice data and a corresponding correct text; recognizing the voice data into text data through a voice input method, comparing the recognized text data with a correct text corresponding to the voice data, and determining the total number of sentences failed in recognition; and determining the identification accuracy of the voice input method according to the total number of the sentences of the correct text and the total number of the sentences which fail to be identified. The proposal not only greatly reduces the calculation base number, but also can ensure that the recognition rate of different voice input methods can be truly reflected to the greatest extent.
Description
Technical field
The application is related to technical field of measurement and test, more particularly, to the recognition accuracy method of testing of phonitic entry method, dress
Put and electronic equipment.
Background technology
The mode of existing general calculating speech recognition accuracy is: recognition accuracy=speech recognition does not have vicious word
The total number of word of number/reading.
Prior art have the disadvantage in that due to number of words more, denominator is very big, and existing computational methods can not be clearly anti-
Mirror the difference of accuracy rate.Too big because of radix on the contrary, lead to result very convergence it is difficult to distinguish.
Content of the invention
The defect aiming to overcome that prior art of the application, proposes one kind and can solve the problem that phonitic entry method identification accurately
The technical scheme of the problem that rate calculating is difficult to differentiate between.
The embodiment of the present application first aspect provides a kind of recognition accuracy method of testing of phonitic entry method, the method bag
Include:
Obtain the test data of test sample set, described test data includes speech data and corresponding correct text;
Described speech data is identified as by text data by phonitic entry method, the text data that will identify that and institute's predicate
The corresponding correct text of sound data is compared, and determines the sentence total quantity of recognition failures;
With the sentence total quantity of described recognition failures, sentence total quantity according to described correct text determines that described voice is defeated
Enter the recognition accuracy of method.
Preferably, described test sample set includes one or more at least one language read by least one accent
The test sample of style, described test sample includes one or more sentences, and described sentence includes phrase, short sentence or long sentence.
Preferably, described diction includes written word style and works and expressions for everyday use style;The sentence of described correct text is total
The acquisition methods of quantity include determining the sentence total quantity of correct text according to the punctuation mark of described correct text.
Preferably, the described punctuation mark according to described correct text determines that the sentence total quantity of correct text includes:
Obtain the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur;
Correct text is determined according to the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur
Sentence total quantity.
Preferably, the described sentence total quantity determining recognition failures includes:
For the arbitrary speech data in the described speech data obtaining, obtain corresponding correct text;
Obtain the text to be appraised and decided obtaining after described phonitic entry method identifies described arbitrary speech data;
The text to be appraised and decided obtaining after the described arbitrary speech data of identification is carried out by language with described corresponding correct text
Sentence compares and determines the quantity identifying sentence failure in described arbitrary speech data;
Quantity according to identifying sentence failure in described arbitrary speech data determines the language identifying described test sample set
The sentence total quantity of sound data failure.
Preferably, described the text to be appraised and decided that obtains and described corresponding correct literary composition after described arbitrary speech data will be identified
Originally carry out Step Into comparison and determine that the quantity identifying sentence failure in described arbitrary speech data includes:
If comparing out incomparable inconsistent between arbitrary sentence of text to be appraised and decided and the corresponding sentence of described correct text comparison
The quantity of unit is not less than first threshold it is determined that described arbitrary sentence recognition failures.
Preferably, described the text to be appraised and decided that obtains and described corresponding correct literary composition after described arbitrary speech data will be identified
Originally carry out Step Into comparison and determine that the quantity identifying sentence failure in described arbitrary speech data includes:
The number of words of corresponding sentence in the corresponding described correct text of arbitrary sentence in text to be appraised and decided described in determination;
According to the corresponding relation of default number of words and Second Threshold, determine the number of words corresponding second with described arbitrary sentence
Threshold value;
If incomparable inconsistent between arbitrary sentence of text to be appraised and decided described in comparing out and corresponding sentence in described correct text
The quantity of comparing unit is not less than described corresponding Second Threshold it is determined that described arbitrary sentence recognition failures.
Preferably, described in the sentence total quantity of the sentence total quantity according to described correct text and described recognition failures determines
The recognition accuracy of phonitic entry method includes:
Calculate the recognition accuracy of described phonitic entry method according to equation below;
The language of recognition accuracy=(the sentence total quantitys of the sentence total quantity-recognition failures of correct text)/correct text
Sentence total quantity.
The embodiment of the present application second aspect provides a kind of recognition accuracy test device of phonitic entry method, described device
Including:
Voice-input device, for the speech data of input test sample set;
Processing equipment, for obtaining the test data of test sample set, described test data includes speech data and right
The correct text answered, after speech data is identified as text data by phonitic entry method, the text data that will identify that with described
The corresponding correct text of speech data is compared, and determines the sentence total quantity of recognition failures;Language according to described correct text
The sentence total quantity of sentence total quantity and described recognition failures determines the recognition accuracy of described phonitic entry method.
Preferably, described test sample set includes one or more at least one language read by least one accent
The test sample of style, described test sample includes one or more sentences, and described sentence includes phrase, short sentence or long sentence.
Preferably, described diction includes written word style and works and expressions for everyday use style;Described processing equipment is used for basis
The punctuation mark of described correct text determines the sentence total quantity of correct text.
Preferably, described processing equipment is used for determining that the sentence of correct text is total according to the punctuation mark of described correct text
Quantity, comprising:
Obtain the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur;
Correct text is determined according to the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur
Sentence total quantity.
Preferably, described processing equipment is used for determining the sentence total quantity of recognition failures, comprising:
For the arbitrary speech data in the described speech data obtaining, obtain corresponding correct text;
Obtain the text to be appraised and decided obtaining after described phonitic entry method identifies described arbitrary speech data;
The text to be appraised and decided obtaining after the described arbitrary speech data of identification is carried out by language with described corresponding correct text
Sentence compares and determines the quantity identifying sentence failure in described arbitrary speech data;
Quantity according to identifying sentence failure in described arbitrary speech data determines the language identifying described test sample set
The sentence total quantity of sound data failure.
Preferably, described processing equipment be used for identifying after described arbitrary speech data the text to be appraised and decided that obtains with described
Corresponding correct text carries out Step Into and compares the quantity determining sentence failure in the described arbitrary speech data of identification, comprising:
If comparing out incomparable inconsistent between arbitrary sentence of text to be appraised and decided and the corresponding sentence of described correct text comparison
The quantity of unit is not less than first threshold it is determined that described arbitrary sentence recognition failures.
Described comparing unit includes sentence is carried out composition division or the minimum unit obtaining by way of participle.When
So it is also possible to by other dividing mode of the prior art.But principle is at least in units of a word, and word can not split into one
The word of individual one is compared, thus reducing comparison number of times.
Preferably, described processing equipment be used for identifying after described arbitrary speech data the text to be appraised and decided that obtains with described
Corresponding correct text carries out Step Into and compares the quantity determining sentence failure in the described arbitrary speech data of identification, comprising:
Determine the number of words of described arbitrary sentence of correct text;
According to the corresponding relation of default number of words and Second Threshold, determine the number of words corresponding second with described arbitrary sentence
Threshold value;
Incomparable inconsistent ratio between arbitrary sentence and the corresponding sentence of text to be appraised and decided of the described correct text comparing out
When described corresponding Second Threshold is not less than to the quantity of unit, determine described arbitrary sentence recognition failures.
The attributes such as the length according to sentence itself, setting identification when one meet or exceed setting error number be judged as failure
Method, closer to the real experiences of user.
The embodiment of the present application third aspect provides a kind of electronic equipment, comprising: voice-input device, processor, storage
Device, communication interface and bus;
Described voice-input device, described processor, described memorizer and described communication interface are connected by described bus
And complete mutual communication;
Described memory storage executable program code;
Described processor is run executable with described by the executable program code of storage in the described memorizer of reading
The corresponding program of program code, for execution method as elucidated before.
The application has the beneficial effect that: carry out the calculating of accuracy rate in this motion in units of sentence, can be more accurate
Evaluate the quality of different phonetic input method.With existing carry out the method for accuracy rate judgement by word compared with, not only greatly reduce meter
Calculate radix, and can guarantee truly to reflect the discrimination height of different phonetic input method as far as possible.
Brief description
The specific embodiment of the application is described below with reference to accompanying drawings, wherein:
Fig. 1 shows that the recognition accuracy method of testing flow process of the phonitic entry method providing in the embodiment of the present application one is illustrated
Figure;
Fig. 2 shows the schematic flow sheet of the step 100 providing in the embodiment of the present application one;
Fig. 3 shows the schematic flow sheet of the step 200 providing in the embodiment of the present application one;
Fig. 4 shows a kind of knot of the recognition accuracy test device of phonitic entry method providing in the embodiment of the present application two
Structure schematic diagram;
Fig. 5 shows a kind of electronic equipment structural representation providing in the embodiment of the present application three.
Specific embodiment
In order that the technical scheme of the application and advantage become more apparent, exemplary to the application below in conjunction with accompanying drawing
Embodiment is described in more detail it is clear that described embodiment is only a part of embodiment of the application, rather than
The exhaustion of all embodiments.And in the case of not conflicting, the embodiment in this explanation and the feature in embodiment can be mutual
Combine.
The calculating that inventor note that the recognition accuracy to phonetic entry in prior art during invention is substantially all
It is to be calculated by word, recognition accuracy=speech recognition does not have the total number of word of vicious number of words/reading.Because calculating basis are big
(i.e. denominator is big), even if identifying that molecule is variant, the result calculated is all very close to it is difficult to embody each phonitic entry method
Recognition accuracy good and bad.
In addition, during realizing the application, inventor also finds that the existing recognition accuracy to phonetic entry is surveyed
Result after examination shows how many character error in a usually test sample, cannot be quick for long paragraph user
Judge it is which mistake actually, have several places mistake, have differences with the actual sense organ of user and experience.
For the problems referred to above, the recognition accuracy method of testing of a kind of phonitic entry method provided herein, device and
Electronic equipment.In the program in units of the sentence in test sample set, according to sentence in the test sample set obtaining
Total quantity and phonitic entry method identify that the total quantity of sentence failure in this test sample set calculates the identification of this phonitic entry method
Accuracy rate, enabling more preferably, more intuitively judge and the recognition accuracy height comparing different input methods.And judging one
Individual sentence whether recognition failures when, according to the attribute of sentence itself, be provided with threshold value, so as to more intuitively reflect that user's is straight
See experience.
The scheme of the embodiment of the present application can be used for intelligent terminal such as smart mobile phone, ipad or vehicle intelligent terminal etc.
Phonitic entry method tests the test it can also be used to the phonitic entry method installed on notebook or desktop computer, etc..
Embodiment one
Fig. 1 is the recognition accuracy method of testing schematic flow sheet of the phonitic entry method of the embodiment of the present application.As Fig. 1 institute
Show, the recognition accuracy method of testing of this phonitic entry method may comprise steps of:
Step 100: obtain test sample set test data, described test data include speech data and corresponding just
Really text;
Step 200: described speech data is identified as by text data by phonitic entry method, the text data that will identify that
Correct text corresponding with described speech data is compared, and determines the sentence total quantity of recognition failures;
Step 300: the sentence total quantity according to described correct text determines institute with the sentence total quantity of described recognition failures
The recognition accuracy of predicate phonetic input method.
In being embodied as, described test sample set include one or more being read by least one accent, at least one
Plant the test sample of diction.Change one to say, preferably described test sample set includes multiple accents, multiple voice style
Speech data.Described test sample includes one or more sentences, and described sentence is phrase, short sentence or long sentence.For example test
Sample " desktop system for mobile phone based on d engine exploitation, move effect experience and redefined ARIXTRA desktop by the three-dimensional extremely dazzled.Impression
Unprecedented unique interaction and visual experience, the challenge speed limit, make your mobile phone unusual from this." in, including comma
Or what fullstop separated is all considered a sentence, in this test sample, altogether include 5 sentences.The people of reading test sample
Member then pauses when comma, fullstop, so that there is punctuation mark between sentence and sentence during identification.Preferably, one
Individual test sample includes multiple accents the are read, speech samples of multiple styles.The reading referring to choose is read in multiple accents
It is not single standard mandarin in the personnel of test sample, but include the reading test sample with Sichuan accent mandarin
Personnel, may have in test sample that river is general, the mandarin with Guangdong accent, etc..For the mandarin with accent
The identification of sample can more realistically reflect the recognition accuracy during input method of different accent this test of librarian use.
Additionally, the sentence that test sample includes can be phrase, short sentence or long sentence, for example " the challenge speed limit " is exactly
One short sentence.Phrase such as " more and more prosperous ", long sentence is primarily referred to as exceeding certain setting value more than number of words in a sentence, for example
One sentence more than 25 words is considered long sentence.
In being embodied as, described diction mainly includes written word style and works and expressions for everyday use style.For example, including written
Locution lattice: the desktop system for mobile phone based on d engine exploitation, the three-dimensional extremely dazzled is moved effect experience and redefined ARIXTRA desktop.Day
Common-use words style: arrived before 10 points.
In being embodied as, first choose the reading word content of non-public publication, choose non-public publication conduct herein
Reading content is also primarily to prevent the impact to speech recognition accuracy for some input method memory functions.Choose at least 10 people
(there are men and women) carries out speech samples input and records acquisition test sample, and described test sample includes common with accent
Words sample.Identifying compared with the identification of existing single standard mandarin to the mandarin sample with accent, more can reflect
The experience of zones of different different places user.Choose personnel when being read, according to formation such as comma, fullstop, branches
Sentence is paused.In identification, sentence quantity typically can be judged according to pause number of times.In being embodied as, described correct literary composition
The acquisition methods of this sentence total quantity include determining the sentence sum of correct text according to the punctuation mark of described correct text
Amount.Furthermore it is preferred that obtaining test sample by way of voice recording, help avoid when different input methods are tested because
The impact that tone testing differences between samples bring.
As shown in Fig. 2 obtaining the schematic flow sheet of the sentence total quantity of correct text, specifically include:
Step 101: obtain the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur;
Step 102: just determined according to the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur
The really sentence total quantity of text.
In being embodied as, provide only a complete correct text for as a test sample collection unification, and according to just
Really the order of text carries out serial number to corresponding speech data, and during identification, corresponding speech data is known by label in order
Wei not text data.It is understood that the mode phase that this set order label carries out speech recognition compares relatively simply, and
Belong to prior art, herein not reinflated elaboration.
For example choose in the test sample set that 10 people are recorded, including ten test samples.In each test equally
Including one or more sentences.The correct text of ten test samples is put into a file, and suitable according to correct text
Ten speech datas that ordered pair is recorded carry out order label.During identification, label is identified in order, and correct with corresponding
Text is compared.It is understood that in order to when comparing sentence need correspondingly, record test sample when each sentence
Pause all standards of comparison, and be not in the situation missing a sentence or many out sentences.Certainly, even
Will be separated for the exact file of ten test samples, first obtain the sentence number of the corresponding correct text of each test sample
Amount, then by plus and by way of obtain correct text sentence total quantity fall within the application protection content.
As shown in figure 3, determining the schematic flow sheet of the sentence total quantity of recognition failures, specifically include:
Step 201: for the arbitrary speech data in the described speech data obtaining, obtain corresponding correct text;
Step 202: obtain the text to be appraised and decided obtaining after described phonitic entry method identifies described arbitrary speech data;
Step 203: the text to be appraised and decided and described corresponding correct text that obtain after described arbitrary speech data will be identified
Carry out Step Into and compare the quantity determining sentence failure in the described arbitrary speech data of identification;
Step 204: the quantity according to identifying sentence failure in described arbitrary speech data determines the described test sample of identification
The sentence total quantity of the speech data failure of set.Tested by summing up the quantity of sentence failure in speech data
The sentence total quantity of the speech data failure of sample set.
In enforcement, described will identify the text to be appraised and decided that obtains and described corresponding correct literary composition after described arbitrary speech data
Originally carry out Step Into comparison and determine that the quantity identifying sentence failure in described arbitrary speech data includes mainly including following two
The mode of kind, test every time can be selected for one of which.
A kind of mode is: if comparing out non-between arbitrary sentence of text to be appraised and decided and the corresponding sentence of described correct text
The quantity of consistent comparing unit is not less than first threshold it is determined that described arbitrary sentence recognition failures.
For example, preset first threshold value is 1, if identifying in a sentence, inconsistent quantity is not less than at 1 it is determined that this language
Sentence recognition failures.In this mode, first threshold is fixing, is typically set at 1, also may be set to other natural numbers.
Another way includes:
The number of words of corresponding sentence in the corresponding described correct text of arbitrary sentence in text to be appraised and decided described in determination;
According to the corresponding relation of default number of words and Second Threshold, determine the number of words corresponding second with described arbitrary sentence
Threshold value;
If incomparable inconsistent between arbitrary sentence of text to be appraised and decided described in comparing out and corresponding sentence in described correct text
The quantity of comparing unit is not less than described corresponding Second Threshold it is determined that described arbitrary sentence recognition failures.
For example, sentence more than n word can be set as long sentence, the Second Threshold that long sentence correspondence sets is as at 2.If identifying
A long sentence in have inconsistent at 3, then this sentence is defined as recognition failures, sets accordingly be less than the sentence of n word as
Short sentence, short sentence corresponds to the Second Threshold setting as 1, if identifying in a short sentence, inconsistent quantity is not less than at 1 it is determined that being somebody's turn to do
Sentence recognition failures, for example wherein n takes 25.Further it will be understood that second decision statement recognition failures mode (with
It is referred to as down the second way) exist with the mode (hereinafter referred to as first kind of way) of the first decision statement recognition failures
Main Differences are, how many second way corresponding can arrange different Second Thresholds according to sentence number of words, and the first
Mode is then that the unified threshold value adopting sets, therefore the second way sentence of different numbers of words is flexibly correspondingly arranged different
Second Threshold, and sentence is judged according to the corresponding Second Threshold of sentence identify whether failure, you can with by changing sentence pair
The Second Threshold answered identifies accurate degrees of fault-tolerance to adjust, thus adapting to different use environments or scene.
Additionally, the comparing unit in both the above decision statement recognition failures mode refers to, carry out composition according to sentence
The minimum unit dividing or being obtained by way of participle.For example according to the mode of participle, " we " belong to a word, sentence
" we do not attend class today " be identified as "YemenToday does not attend class " although mistake is two words, but a word mistake, know
Wei not mistake at.Illustrate, sentence is divided or the method for participle belongs to prior art, simply directly will herein
The prior art takes back application, thus not reinflated elaboration.
In being embodied as, the quantity of sentence and the described test sample collection of identification in the test sample set that described basis obtains
The recognition accuracy that the quantity of sentence failure in conjunction calculates described phonitic entry method includes:
Calculate the recognition accuracy of described phonitic entry method according to equation below;
The language of recognition accuracy=(the sentence total quantitys of the sentence total quantity-recognition failures of correct text)/correct text
Sentence total quantity.
Embodiment two
Based on same application design, in the embodiment of the present application, additionally provide a kind of recognition accuracy test of phonitic entry method
Device, because the principle of this device solve problem is similar to the method for the data processing in embodiment one, the therefore reality of this device
Apply the enforcement of the method for may refer to, repeat no more in place of repetition.
Fig. 4 shows the recognition accuracy test device structural representation of the phonitic entry method in the embodiment of the present application.As
Shown in Fig. 4, the recognition accuracy test device of this phonitic entry method specifically includes that
Voice-input device 401, for the speech data of input test sample set;
Processing equipment 402, for obtaining the test data of test sample set, described test data include speech data and
Corresponding correct text, after speech data is identified as text data by phonitic entry method, the text data that will identify that and institute
State the corresponding correct text of speech data to compare, determine the sentence total quantity of recognition failures;According to described correct text
The sentence total quantity of sentence total quantity and described recognition failures determines the recognition accuracy of described phonitic entry method.
In being embodied as, described test sample set includes one or more at least one read by least one accent
The test sample of diction, described test sample includes one or more sentences, and described sentence includes phrase, short sentence or length
Sentence.
In being embodied as, described diction includes written word style and works and expressions for everyday use style;Described processing equipment 402 is used
In the sentence total quantity determining correct text according to the punctuation mark of described correct text.
In being embodied as, described processing equipment 402 is used for determining correct text according to the punctuation mark of described correct text
Sentence total quantity, comprising:
Obtain the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur;
Correct text is determined according to the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur
Sentence total quantity.
In being embodied as, described processing equipment 402 is used for determining the sentence total quantity of recognition failures, comprising:
For the arbitrary speech data in the described speech data obtaining, obtain corresponding correct text;
Obtain the text to be appraised and decided obtaining after described phonitic entry method identifies described arbitrary speech data;
The text to be appraised and decided obtaining after the described arbitrary speech data of identification is carried out by language with described corresponding correct text
Sentence compares and determines the quantity identifying sentence failure in described arbitrary speech data;
Quantity according to identifying sentence failure in described arbitrary speech data determines the language identifying described test sample set
The sentence total quantity of sound data failure.
In one embodiment, described processing equipment 402 be used for by identify described arbitrary speech data after obtain wait appraise and decide
Text carries out Step Into with described corresponding correct text and compares the number determining sentence failure in the described arbitrary speech data of identification
Amount, comprising:
If comparing out incomparable inconsistent between arbitrary sentence of text to be appraised and decided and the corresponding sentence of described correct text comparison
The quantity of unit is not less than first threshold it is determined that described arbitrary sentence recognition failures.
In another enforcement, described processing equipment 402 is used for treating core by obtain after the described arbitrary speech data of identification
Determine text and described corresponding correct text to carry out Step Into and compare determining and identifying sentence failure in described arbitrary speech data
Quantity, comprising:
Determine the number of words of described arbitrary sentence of correct text;
According to the corresponding relation of default number of words and Second Threshold, determine the number of words corresponding second with described arbitrary sentence
Threshold value;
Incomparable inconsistent ratio between arbitrary sentence and the corresponding sentence of text to be appraised and decided of the described correct text comparing out
When described corresponding Second Threshold is not less than to the quantity of unit, determine described arbitrary sentence recognition failures.
Embodiment three
The embodiment of the present application additionally provides a kind of electronic equipment, as shown in figure 5, this electronic equipment 500 specifically includes that voice
Input equipment 505, processor 501, memorizer 502, communication interface 503 and bus 504;
Described voice-input device 505, described processor 501, described memorizer 502 and described communication interface 503 are passed through
Described bus 504 connects and completes mutual communication;
Described memorizer 502 stores executable program code;
Described processor 501 by read described memorizer 502 in storage executable program code run with described
The corresponding program of executable program code, for executing a kind of recognition accuracy method of testing of phonitic entry method;Wherein, institute
The recognition accuracy method of testing of predicate phonetic input method specifically includes that
Obtain the test data of test sample set, described test data includes speech data and corresponding correct text;
Described speech data is identified as by text data by phonitic entry method, the text data that will identify that and institute's predicate
The corresponding correct text of sound data is compared, and determines the sentence total quantity of recognition failures;
With the sentence total quantity of described recognition failures, sentence total quantity according to described correct text determines that described voice is defeated
Enter the recognition accuracy of method.
Preferably, described test sample set includes one or more at least one language read by least one accent
The test sample of style, described test sample includes one or more sentences, and described sentence includes phrase, short sentence or long sentence.
Preferably, described diction includes written word style and works and expressions for everyday use style;The sentence of described correct text is total
The acquisition methods of quantity include determining the sentence total quantity of correct text according to the punctuation mark of described correct text.
Preferably, the described punctuation mark according to described correct text determines that the sentence total quantity of correct text includes:
Obtain the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur;
Correct text is determined according to the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur
Sentence total quantity.
Preferably, the described sentence total quantity determining recognition failures includes:
For the arbitrary speech data in the described speech data obtaining, obtain corresponding correct text;
Obtain the text to be appraised and decided obtaining after described phonitic entry method identifies described arbitrary speech data;
The text to be appraised and decided obtaining after the described arbitrary speech data of identification is carried out by language with described corresponding correct text
Sentence compares and determines the quantity identifying sentence failure in described arbitrary speech data;
Quantity according to identifying sentence failure in described arbitrary speech data determines the language identifying described test sample set
The sentence total quantity of sound data failure.
Preferably, described the text to be appraised and decided that obtains and described corresponding correct literary composition after described arbitrary speech data will be identified
Originally carry out Step Into comparison and determine that the quantity identifying sentence failure in described arbitrary speech data includes:
If comparing out incomparable inconsistent between arbitrary sentence of text to be appraised and decided and the corresponding sentence of described correct text comparison
The quantity of unit is not less than first threshold it is determined that described arbitrary sentence recognition failures.
Preferably, described the text to be appraised and decided that obtains and described corresponding correct literary composition after described arbitrary speech data will be identified
Originally carry out Step Into comparison and determine that the quantity identifying sentence failure in described arbitrary speech data includes:
The number of words of corresponding sentence in the corresponding described correct text of arbitrary sentence in text to be appraised and decided described in determination;
According to the corresponding relation of default number of words and Second Threshold, determine the number of words corresponding second with described arbitrary sentence
Threshold value;
If incomparable inconsistent between arbitrary sentence of text to be appraised and decided described in comparing out and corresponding sentence in described correct text
The quantity of comparing unit is not less than described corresponding Second Threshold it is determined that described arbitrary sentence recognition failures.
Preferably, described in the sentence total quantity of the sentence total quantity according to described correct text and described recognition failures determines
The recognition accuracy of phonitic entry method includes:
Calculate the recognition accuracy of described phonitic entry method according to equation below;
The language of recognition accuracy=(the sentence total quantitys of the sentence total quantity-recognition failures of correct text)/correct text
Sentence total quantity.
The embodiment of the present application additionally provides a kind of application program, and wherein, this application program is used for operationally executing this Shen
Please a kind of phonitic entry method described in embodiment recognition accuracy method of testing.
This application program can run in the electronic equipment of the embodiment of the present application offer.
The embodiment of the present application additionally provides a kind of storage medium, and wherein, this storage medium is used for storing application program, described
Application program is used for the recognition accuracy method of testing of the operationally phonitic entry method described in execution the embodiment of the present application.
Those skilled in the art are it should be appreciated that embodiments herein can be provided as method, system or computer program
Product.Therefore, the application can be using complete hardware embodiment, complete software embodiment or the reality combining software and hardware aspect
Apply the form of example.And, the application can be using in one or more computers wherein including computer usable program code
The upper computer program implemented of usable storage medium (including but not limited to disk memory, cd-rom, optical memory etc.) produces
The form of product.
The application is the flow process with reference to method, equipment (system) and computer program according to the embodiment of the present application
Figure and/or block diagram are describing.It should be understood that can be by each stream in computer program instructions flowchart and/or block diagram
Flow process in journey and/or square frame and flow chart and/or block diagram and/or the combination of square frame.These computer programs can be provided
The processor instructing general purpose computer, special-purpose computer, Embedded Processor or other programmable data processing device is to produce
A raw machine is so that produced for reality by the instruction of computer or the computing device of other programmable data processing device
The device of the function of specifying in present one flow process of flow chart or multiple flow process and/or one square frame of block diagram or multiple square frame.
These computer program instructions may be alternatively stored in and can guide computer or other programmable data processing device with spy
Determine in the computer-readable memory that mode works so that the instruction generation inclusion being stored in this computer-readable memory refers to
Make the manufacture of device, this command device realize in one flow process of flow chart or multiple flow process and/or one square frame of block diagram or
The function of specifying in multiple square frames.
These computer program instructions also can be loaded in computer or other programmable data processing device so that counting
On calculation machine or other programmable devices, execution series of operation steps to be to produce computer implemented process, thus in computer or
On other programmable devices, the instruction of execution is provided for realizing in one flow process of flow chart or multiple flow process and/or block diagram one
The step of the function of specifying in individual square frame or multiple square frame.
Although having been described for the preferred embodiment of the application, those skilled in the art once know basic creation
Property concept, then can make other change and modification to these embodiments.So, claims are intended to be construed to including excellent
Select embodiment and fall into being had altered and changing of the application scope.
Claims (10)
1. a kind of recognition accuracy method of testing of phonitic entry method is it is characterised in that the method includes:
Obtain the test data of test sample set, described test data includes speech data and corresponding correct text;
Described speech data is identified as by text data by phonitic entry method, the text data that will identify that and described voice number
Compare according to corresponding correct text, determine the sentence total quantity of recognition failures;
Sentence total quantity according to described correct text determines described phonitic entry method with the sentence total quantity of described recognition failures
Recognition accuracy.
2. the method for claim 1 is it is characterised in that described test sample set includes:
The test sample of one or more at least one dictions read by least one accent, described test sample includes
One or more sentences, described sentence includes phrase, short sentence or long sentence.
3. method as claimed in claim 2 is it is characterised in that described diction includes:
Written word style and works and expressions for everyday use style;
The punctuation mark that the acquisition methods of the sentence total quantity of described correct text are included according to described correct text determines correct
The sentence total quantity of text.
4. method as claimed in claim 3 is it is characterised in that the described punctuation mark according to described correct text determines correctly
The sentence total quantity of text includes:
Obtain the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur;
Determine the sentence of correct text according to the number of times that the comma in correct text, fullstop, branch, question mark and exclamation mark occur
Total quantity.
5. the method as any one of Claims 1-4 is it is characterised in that the sentence of described determination recognition failures is total
Amount includes:
Obtain the text to be appraised and decided obtaining after described phonitic entry method identifies described arbitrary speech data;
The text to be appraised and decided obtaining after the described arbitrary speech data of identification is carried out Step Into ratio with described corresponding correct text
To the quantity determining sentence failure in the described arbitrary speech data of identification;
Quantity according to identifying sentence failure in described arbitrary speech data determines the voice number identifying described test sample set
Sentence total quantity according to failure.
6. method as claimed in claim 5 is it is characterised in that obtain after described described arbitrary speech data by identification treats core
Determine text and described corresponding correct text to carry out Step Into and compare determining and identifying sentence failure in described arbitrary speech data
Quantity includes:
If comparing out incomparable inconsistent comparing unit between arbitrary sentence of text to be appraised and decided and the corresponding sentence of described correct text
Quantity be not less than first threshold it is determined that described arbitrary sentence recognition failures, described comparing unit includes sentence is become
The minimum unit that graduation is divided or obtained by way of participle.
7. method as claimed in claim 5 is it is characterised in that obtain after described described arbitrary speech data by identification treats core
Determine text and described corresponding correct text to carry out Step Into and compare determining and identifying sentence failure in described arbitrary speech data
Quantity includes:
The number of words of corresponding sentence in the corresponding described correct text of arbitrary sentence in text to be appraised and decided described in determination;
According to the corresponding relation of default number of words and Second Threshold, determine second threshold corresponding with the number of words of described arbitrary sentence
Value;
If incomparable inconsistent between arbitrary sentence of text to be appraised and decided described in comparing out and corresponding sentence in described correct text compare
The quantity of unit is not less than described corresponding Second Threshold it is determined that described arbitrary sentence recognition failures, described comparing unit bag
Include and sentence is carried out with composition division or the minimum unit being obtained by way of participle.
8. a kind of recognition accuracy test device of phonitic entry method is it is characterised in that described device includes:
Voice-input device, for the speech data of input test sample set;
Processing equipment, for obtaining the test data of test sample set, described test data includes speech data and corresponding
Correct text, after speech data is identified as text data by phonitic entry method, text data and the described voice that will identify that
The corresponding correct text of data is compared, and determines the sentence total quantity of recognition failures;Total according to the sentence of described correct text
The sentence total quantity of quantity and described recognition failures determines the recognition accuracy of described phonitic entry method.
9. device as claimed in claim 8 it is characterised in that described test sample set include one or more by least one
Plant the test sample of at least one diction that accent is read, described test sample includes one or more sentences, institute's predicate
Sentence includes phrase, short sentence or long sentence.
10. device as claimed in claim 9 is it is characterised in that described diction includes written word style and works and expressions for everyday use
Style;Described processing equipment is used for determining the sentence total quantity of correct text according to the punctuation mark of described correct text.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610648394.0A CN106356053A (en) | 2016-08-09 | 2016-08-09 | Method and device for testing recognition accuracy of voice input method and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610648394.0A CN106356053A (en) | 2016-08-09 | 2016-08-09 | Method and device for testing recognition accuracy of voice input method and electronic equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106356053A true CN106356053A (en) | 2017-01-25 |
Family
ID=57844495
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610648394.0A Pending CN106356053A (en) | 2016-08-09 | 2016-08-09 | Method and device for testing recognition accuracy of voice input method and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106356053A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109215640A (en) * | 2017-06-30 | 2019-01-15 | 深圳大森智能科技有限公司 | Audio recognition method, intelligent terminal and computer readable storage medium |
CN109410915A (en) * | 2017-08-15 | 2019-03-01 | 中国移动通信集团终端有限公司 | The appraisal procedure and device of voice quality, computer readable storage medium |
CN110223689A (en) * | 2019-06-10 | 2019-09-10 | 秒针信息技术有限公司 | The determination method and device of the optimization ability of voice messaging, storage medium |
CN110245576A (en) * | 2019-05-21 | 2019-09-17 | 深圳壹账通智能科技有限公司 | Detection method, device, equipment and the storage medium of OCR recognition accuracy |
CN113707148A (en) * | 2021-08-05 | 2021-11-26 | 中移(杭州)信息技术有限公司 | Method, device, equipment and medium for determining accuracy rate of voice recognition |
CN113849604A (en) * | 2021-09-27 | 2021-12-28 | 广东纬德信息科技股份有限公司 | NLP-based power grid regulation and control method, system, equipment and storage medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1811915A (en) * | 2005-01-28 | 2006-08-02 | 中国科学院计算技术研究所 | Estimating and detecting method and system for telephone continuous speech recognition system performance |
-
2016
- 2016-08-09 CN CN201610648394.0A patent/CN106356053A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1811915A (en) * | 2005-01-28 | 2006-08-02 | 中国科学院计算技术研究所 | Estimating and detecting method and system for telephone continuous speech recognition system performance |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109215640A (en) * | 2017-06-30 | 2019-01-15 | 深圳大森智能科技有限公司 | Audio recognition method, intelligent terminal and computer readable storage medium |
CN109215640B (en) * | 2017-06-30 | 2021-06-01 | 深圳大森智能科技有限公司 | Speech recognition method, intelligent terminal and computer readable storage medium |
CN109410915A (en) * | 2017-08-15 | 2019-03-01 | 中国移动通信集团终端有限公司 | The appraisal procedure and device of voice quality, computer readable storage medium |
CN109410915B (en) * | 2017-08-15 | 2022-03-04 | 中国移动通信集团终端有限公司 | Method and device for evaluating voice quality and computer readable storage medium |
CN110245576A (en) * | 2019-05-21 | 2019-09-17 | 深圳壹账通智能科技有限公司 | Detection method, device, equipment and the storage medium of OCR recognition accuracy |
CN110223689A (en) * | 2019-06-10 | 2019-09-10 | 秒针信息技术有限公司 | The determination method and device of the optimization ability of voice messaging, storage medium |
CN113707148A (en) * | 2021-08-05 | 2021-11-26 | 中移(杭州)信息技术有限公司 | Method, device, equipment and medium for determining accuracy rate of voice recognition |
CN113707148B (en) * | 2021-08-05 | 2024-04-19 | 中移(杭州)信息技术有限公司 | Method, device, equipment and medium for determining speech recognition accuracy |
CN113849604A (en) * | 2021-09-27 | 2021-12-28 | 广东纬德信息科技股份有限公司 | NLP-based power grid regulation and control method, system, equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106356053A (en) | Method and device for testing recognition accuracy of voice input method and electronic equipment | |
CN107908635B (en) | Method and device for establishing text classification model and text classification | |
CN105706163B (en) | For detecting the method and system of voice input phrase confusion risk | |
Preston et al. | Developing a weighted measure of speech sound accuracy | |
CN104464757B (en) | Speech evaluating method and speech evaluating device | |
CN110532522A (en) | Error-detecting method, device, computer equipment and the storage medium of audio mark | |
CN109360550A (en) | Test method, device, equipment and the storage medium of voice interactive system | |
CN110276023A (en) | POI changes event discovery method, apparatus, calculates equipment and medium | |
CN109947651B (en) | Artificial intelligence engine optimization method and device | |
CN109166569B (en) | Detection method and device for phoneme mislabeling | |
Klessa et al. | Annotation Pro+ TGA: automation of speech timing analysis. | |
CN109697988A (en) | A kind of Speech Assessment Methods and device | |
WO2023278980A1 (en) | Interface to natural language generator for generation of knowledge assessment items | |
US7475016B2 (en) | Speech segment clustering and ranking | |
WO2021012495A1 (en) | Method and device for verifying speech recognition result, computer apparatus, and medium | |
CN106250755A (en) | For generating the method and device of identifying code | |
Oliveira et al. | An Extensible Framework to Implement Test Oracle for Non-Testable Programs. | |
Hacine-Gharbi et al. | Prosody based Automatic Classification of the Uses of French ‘Oui’as Convinced or Unconvinced Uses | |
TR202022040A1 (en) | A METHOD OF MEASURING TEXT SUMMARY SUCCESS THAT IS SENSITIVE TO SUBJECT CLASSIFICATION AND A SUMMARY SYSTEM USING THIS METHOD | |
US10546080B1 (en) | Method and system for identifying potential causes of failure in simulation runs using machine learning | |
Maxwell et al. | Homogeneity vs Heterogeneity in Indian English: Investigating Influences of L1 on f0 Range. | |
CN113724738B (en) | Speech processing method, decision tree model training method, device, equipment and storage medium | |
CN113704452B (en) | Data recommendation method, device, equipment and medium based on Bert model | |
Davel et al. | Verifying pronunciation dictionaries using conflict analysis | |
CN115438129A (en) | Structured data classification method and device and terminal equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170125 |
|
RJ01 | Rejection of invention patent application after publication |