CN114519358A - Translation quality evaluation method and device, electronic equipment and storage medium - Google Patents
Translation quality evaluation method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN114519358A CN114519358A CN202210145951.2A CN202210145951A CN114519358A CN 114519358 A CN114519358 A CN 114519358A CN 202210145951 A CN202210145951 A CN 202210145951A CN 114519358 A CN114519358 A CN 114519358A
- Authority
- CN
- China
- Prior art keywords
- text
- translation
- language
- source language
- determining
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
- G06F40/51—Translation evaluation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
- G06F40/58—Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Machine Translation (AREA)
Abstract
The invention provides a translation quality evaluation method, a translation quality evaluation device, electronic equipment and a storage medium, wherein the method comprises the following steps: determining source language voice and target language text; performing source language standard translation on the source language voice to obtain a source language standard text, and determining a voice evaluation result based on the source language standard text; determining a text evaluation result based on the target language text; and determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result. The translation quality evaluation method, the translation quality evaluation device, the electronic equipment and the storage medium provided by the embodiment of the invention simultaneously take the quality of the source language voice and the quality of the target language text into consideration, thereby improving the accuracy of the translation quality evaluation.
Description
Technical Field
The present invention relates to the field of natural language technologies, and in particular, to a translation quality evaluation method and apparatus, an electronic device, and a storage medium.
Background
With the market demand for translation further expanding, strong demand is put forward for instant translation, and people need instant translation to facilitate direct communication. However, the current machine translation capability still does not reach the translation level of professional translators, and how to accurately evaluate and measure the translation quality of instant translation under different scenes becomes increasingly important.
The existing evaluation of instant translation quality mainly evaluates a translation, and determines the translation quality according to the similarity between the text of the original voice and the text of the translation. However, due to the diversity of language expression and the richness of language directions, the reliability of the translation quality evaluation is still insufficient.
Disclosure of Invention
The invention provides a translation quality evaluation method, a translation quality evaluation device, electronic equipment and a storage medium, which are used for solving the defect of poor reliability of translation quality evaluation in the prior art.
The invention provides a translation quality evaluation method, which comprises the following steps:
determining source language voice and target language text;
performing source language standard translation on the source language voice to obtain a source language standard text, and determining a voice evaluation result based on the source language standard text;
determining a text evaluation result based on the target language text;
and determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result.
According to the translation quality evaluation method provided by the invention, the determining of the speech evaluation result based on the source language standard text comprises the following steps:
determining a word level voice evaluation result based on the translation accuracy of the segmented words in the source language standard text to the source language voice;
And/or determining sentence level voice evaluation results based on semantic accuracy and/or expression fluency of clauses in the source language standard text;
determining a speech evaluation result based on the word-level speech evaluation result and/or sentence-level speech evaluation result.
According to the translation quality evaluation method provided by the invention, the translation accuracy of the segmented words in the source language standard text to the source language speech is determined based on the following steps:
determining a first technical field based on the technical field of word segmentation in the transcribed text of the source language voice;
and determining the translation accuracy of the segmented words in the source language standard text to the source language voice based on the first technical field.
According to the translation quality evaluation method provided by the invention, the determining of the text evaluation result based on the target language text comprises the following steps:
determining a word level text evaluation result based on the translation accuracy of the participles in the target language text to the source language standard text;
and/or performing source language standardized translation on the target language text to obtain a translation standard text, and determining a sentence level text evaluation result based on the semantic accuracy and/or expression fluency of the clauses in the translation standard text;
Determining a text evaluation result based on the word-level text evaluation result and/or sentence-level text evaluation result.
According to the translation quality evaluation method provided by the invention, the translation accuracy of the participle in the target language text to the source language standard text is determined based on the following steps:
determining a second technical field based on the technical field of word segmentation in the target language text;
and determining the translation accuracy of the participles in the target language text to the source language standard text based on the second technical field.
According to the translation quality evaluation method provided by the invention, the semantic accuracy of the clause is determined based on the following steps:
determining a target standard language order, wherein the target standard language order is based on a standard language order of a source language in the first technical field and/or the standard language order of a target language in the second technical field;
and based on the target standard language sequence, performing language sequence adjustment on clauses in the source language standard text or the translation standard text, and acquiring semantic accuracy of the clauses in the source language standard text or the translation standard text after the language sequence adjustment.
According to the translation quality evaluation method provided by the invention, the expression fluency of the clauses is determined based on the following steps:
And determining the expression fluency of the clauses in the source language standard text or the translation standard text based on the relevance of the clauses in the source language standard text or the translation standard text clause.
The present invention also provides a translation quality evaluation apparatus, including:
the voice and text determining unit is used for determining source language voice and target language text;
the voice evaluation unit is used for performing source language standard translation on the source language voice to obtain a source language standard text and determining a voice evaluation result based on the source language standard text;
the text evaluation unit is used for determining a text evaluation result based on the target language text;
and the translation quality determining unit is used for determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result.
The present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the translation quality evaluation method according to any of the above methods when executing the program.
The present invention also provides a non-transitory computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a translation quality assessment method as described in any of the above.
The present invention also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of the translation quality assessment method according to any of the above.
According to the translation quality evaluation method, the translation quality evaluation device, the electronic equipment and the storage medium, the source language standard text is obtained by performing source language standard translation on the source language speech, and the speech evaluation result is determined based on the source language standard text; determining a text evaluation result based on the target language text; and determining the translation quality of the target language text based on the speech evaluation result and the text evaluation result, wherein the quality of the source language speech and the quality of the target language text are considered simultaneously, so that the accuracy of the translation quality evaluation is improved.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a translation quality assessment method provided by the present invention;
FIG. 2 is a flow chart of a speech assessment method provided by the present invention;
FIG. 3 is a schematic flow chart of a translation accuracy determining method according to the present invention;
FIG. 4 is a schematic flow chart of step 130 of the translation quality evaluation method provided by the present invention;
FIG. 5 is a second flowchart of the translation accuracy determination method according to the present invention;
FIG. 6 is a flow chart of a sentence semantic accuracy determination method according to the present invention;
FIG. 7 is a schematic structural diagram of a translation quality evaluation apparatus provided in the present invention;
fig. 8 is a schematic structural diagram of an electronic device provided in the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The existing instant translation quality evaluation mainly aims at evaluating a translated text, and the translation quality is determined according to the similarity between the text of the original voice and the text of the translated text. However, due to the diversity of language expression and the richness of language direction, the confidence of translation quality evaluation is still insufficient.
In view of the above, the embodiment of the present invention provides a translation quality evaluation method. Fig. 1 is a schematic flow chart of a translation quality evaluation method provided by the present invention, and as shown in fig. 1, the method includes:
Specifically, the source language speech is speech data used for translation, the source language speech may be obtained through sound pickup equipment, where the sound pickup equipment may be a smart phone, a tablet computer, or a smart electrical apparatus such as a sound, a television, an air conditioner, and the like, the sound pickup equipment may further amplify and reduce noise of the source language speech after the source language speech is picked up by a microphone array, and in addition, the source language speech may be a speech segment formed after the sound pickup is finished, or a speech stream in a real-time sound pickup process, which is not specifically limited in this embodiment of the present invention.
The target language text is the translated text to be evaluated, which needs to be subjected to quality evaluation, and the target language text is obtained by translating the source language voice. The target language text may be a translation obtained by machine translation, or may be a translation obtained by manual translation by a translator, which is not specifically limited in this embodiment of the present invention.
The source and target languages herein may be languages in different countries and regions, e.g., speech translation may support multi-national languages such as mandarin, japanese, russian, english, etc. The source language speech may also support various dialects, such as various dialects corresponding to the mandarin language, e.g., sichuan language, yue language, southern Fujian language, Wei language, Shanghai language, etc.
And 120, performing source language standard translation on the source language speech to obtain a source language standard text, and determining a speech evaluation result based on the source language standard text.
Specifically, considering the diversity of language expression, taking the source language as mandarin as an example, because the mandarin standard degrees of people in various regions are different, the input speech has dialect accents in different degrees, thereby affecting the translation quality of the target language text; and the language expression habit of each person is also different, which also influences the translation quality of the target language text. Therefore, the translation quality evaluation method provided by the embodiment of the invention considers the speech quality of the input speech, and can also perform speech evaluation on the source language speech before performing comprehensive evaluation on the translation quality.
And performing source language standard translation on the source language voice, namely translating the source language voice into a corresponding standard language to obtain a source language standard text. For example, Sichuan speech or non-standard Mandarin speech may be translated into standard Mandarin text, where the resulting standard Mandarin text is the source language standard text.
And then, evaluating the source language speech according to the obtained source language standard text to obtain a speech evaluation result, wherein the source language standard text only relates to the standardized correction on the accent of the source language speech, and the speech evaluation result based on the source language standard text can directly reflect the speech quality of the source language speech after the accent interference is removed, such as whether the ideographical meaning of the source language speech is clear or not, whether the word is ambiguous or not, and the like.
The speech evaluation result can evaluate the accuracy of the source language speech by taking the participle as a unit from the semantics of the source language speech and the source language standard text participle, so as to obtain a speech word-level evaluation result; and evaluating the semantic accuracy and/or expression fluency of the source language voice by taking the clauses as units from the overall semantics of the source language voice and the source language standard text clauses, thereby obtaining a sentence level evaluation result of the voice.
And step 130, determining a text evaluation result based on the target language text.
Specifically, the text evaluation may be performed on the target language text to obtain a text evaluation result, and the text evaluation result may represent the translation quality of the target language text. The text evaluation result can evaluate the translation accuracy of the participles in the target language text by taking the participles as units from the semantics of the participles of the source language standard text and the target language text, so as to obtain a word level evaluation result of the text; and evaluating the semantic accuracy and/or expression fluency of the target language text clause by taking the clause as a unit from the overall semantics of the source language standard text and the target language text clause, thereby obtaining a sentence level evaluation result of the text.
And step 140, determining the translation quality of the target language text based on the speech evaluation result and the text evaluation result.
Specifically, the speech evaluation result and the text evaluation result may be fused to obtain a translation quality evaluation result of the target language text. Here, corresponding weights may be set for the speech evaluation result and the text evaluation result, and then the speech evaluation result and the text evaluation result may be fused in a weighted manner, or the speech evaluation result and the text evaluation result may be fused through a multi-layer perceptron network or other forms of fusion networks, which is not limited in this embodiment of the present invention.
Furthermore, word level evaluation results of the voice and/or sentence level evaluation results of the voice, which are obtained by performing quality evaluation from different granularities, word level evaluation results of the text and/or sentence level evaluation results of the text, which are obtained by performing quality evaluation from different granularities, can be fused, so that the accuracy of translation quality evaluation is further improved.
According to the method provided by the embodiment of the invention, the source language standard text is obtained by performing source language standard translation on the source language speech, and the speech evaluation result is determined based on the source language standard text; determining a text evaluation result based on the target language text; and determining the translation quality of the target language text based on the speech evaluation result and the text evaluation result, wherein the quality of the source language speech and the quality of the target language text are considered simultaneously, so that the accuracy of the translation quality evaluation is improved.
Based on the above embodiment, fig. 2 is a flowchart of the speech evaluation method provided by the present invention, and as shown in fig. 2, the step 120 of determining a speech evaluation result based on the source language standard text includes:
and/or step 122, determining sentence level voice evaluation results based on semantic accuracy and/or expression fluency of the clauses in the source language standard text;
and step 123, determining a voice evaluation result based on the word-level voice evaluation result and/or sentence-level voice evaluation result.
Specifically, it is considered that the pronunciation of the input speech person may be less standard, such as with local accent, or the input speech contains domain vocabulary of a certain domain, thereby affecting the translation quality evaluation result of the target language text. Thus, the quality assessment of the source language speech can be performed from different granularities. The speech evaluation result can evaluate the translation accuracy of the segmented words in the source language standard text to the source language speech by taking the segmented words as units from the semantics of the segmented words in the source language speech and the source language standard text, so as to obtain a word-level speech evaluation result.
The translation accuracy can represent the possibility that when the speech segments in the source language are translated in sequence, the speech segments are translated into the standard text segments at the corresponding positions.
It should be noted that, here, word-level speech evaluation results may be determined based on the translation accuracy of each segmented word in the source language speech and the source language standard text; the word-level speech evaluation result may also be determined based on the source language speech and some segmented words in the source language standard text, for example, some key segmented words or domain vocabularies, which is not specifically limited in this embodiment of the present invention.
For example, the source language speech is "are you at dry? Now ", corresponding to the expression habits and grammatical structure of chinese, the corresponding source language standard text expression is" what are you doing now? ". The word-level speech assessment results may then be determined based on the translation accuracy of the participles "you", "now", "what" in the source language standard text for each participle "you", "at", "heavily" and "now" in the source language speech. The higher the translation accuracy of the word segmentation is, the better the word-level speech evaluation result is; the lower the translation accuracy of the segmented word, the worse the word-level speech evaluation result.
The speech evaluation result can also evaluate the semantic accuracy and/or expression fluency of the source language speech by taking the clause as a unit from the overall semantics of the source language standard text clause, thereby obtaining the sentence-level evaluation result of the speech.
The semantic accuracy can represent the clarity of the semantics expressed by the source language speech, the higher the semantic accuracy is, the clearer the semantics expressing the source language speech is, the lower the possibility of ambiguity exists, and the better the sentence level evaluation result of the speech is; the lower the semantic accuracy, the less clear the semantic meaning of the source language speech, the greater the possibility of ambiguity, and the worse the sentence-level evaluation result of the speech. For example, the source language speech is: "where do you now? "or" where are you? Now, the accurate meaning of the voice can be clearly expressed; "you are now" belonging to the source language speech with low semantic accuracy, and the accurate meaning of the speech cannot be clearly expressed, thereby affecting the translation evaluation result of the target language.
The expression fluency can represent the fluency of the whole expression of the source language speech, and reflects the degree that the expression of the source language speech conforms to the expression habit of the corresponding source language. For example, the expression in the source language speech is finished at one time, the katoon rarely occurs, the word order of the main and predicate guest accords with the expression habit of the corresponding language, and the expression fluency is considered to be high. The higher the expression fluency is, the better the sentence level evaluation result of the voice is; the lower the fluency of expression, the worse the sentence-level assessment result of speech. It should be noted that sentence-level speech evaluation results can be determined based on semantic accuracy and/or expression fluency of each clause in the source language standard text; sentence-level speech evaluation results can also be determined based on partial sentences in the source language standard text, such as semantic accuracy and/or expression fluency of some key sentences.
Determining a voice evaluation result according to the word-level voice evaluation result, or determining a voice evaluation result according to the sentence-level voice evaluation result; or the word-level voice evaluation result and the sentence-level voice evaluation result are fused to determine a voice evaluation result.
According to the method provided by the embodiment of the invention, the speech evaluation result is determined based on the word-level speech evaluation result and/or sentence-level speech evaluation result, and the speech quality is evaluated from different dimensions, so that the evaluation reliability of the translation quality of the target language text is improved.
Based on any of the above embodiments, fig. 3 is a schematic flow chart of the method for determining translation accuracy provided by the present invention, and as shown in fig. 3, in step 121, the translation accuracy of a term in a standard text of a source language for a speech of the source language is determined based on the following steps:
and 320, determining the translation accuracy of the words in the source language standard text to the source language voice based on the first technical field.
Specifically, considering that each technical field has a field vocabulary, the same speech segmentation may correspond to different translated texts in different fields, so that when the translation quality of the target language text is evaluated, the evaluation may be performed based on the technical field to which the source language speech belongs.
The transcription text of the source language speech is the text obtained by performing speech transcription on the source language speech. The speech recognition model can be used for speech transcription, and the source language speech is input into the speech recognition model to obtain a transcription text of the source language speech.
The technical field of the segmentation in the transcribed text can be determined by taking the segmentation as a unit according to the probability of the occurrence of the segmentation in the field lexicon, for example, the field lexicon can comprise daily conversations, business exchanges and academic exchanges, and the technical field of the segmentation is determined according to the field lexicon in which the probability of the occurrence of the segmentation is the highest. The first technical field is the technical field determined by the technical field of word segmentation in the transcribed text based on the source language speech.
For example, the transcription text of the speech in the source language is: where do you are? Now. The corresponding participles are respectively as follows: you, where, now, can thus determine that the first technical field is the "daily dialog" field.
And determining the translation accuracy of the participles in the source language standard text to the source language voice in the technical field corresponding to the participles. Specifically, under the technical field corresponding to the word segmentation, the standard source language text is subjected to standardized translation decoding to obtain the translation accuracy of the word segmentation in the transcribed text of the source language voice, and the word level voice evaluation result can be the sum of the translation accuracy of the word segmentation in the transcribed text of the source language voice.
According to the method provided by the embodiment of the invention, the translation accuracy of the words in the source language standard text to the source language voice is determined based on the first technical field, so that the credibility of the word-level voice evaluation result is further improved.
Based on any of the above embodiments, fig. 4 is a schematic flow chart of step 130 in the translation quality assessment method provided by the present invention, and as shown in fig. 4, step 130 specifically includes:
step 131, determining a word level text evaluation result based on the translation accuracy rate of the partial words in the target language text to the standard text in the source language;
and/or step 132, performing source language standardized translation on the target language text to obtain a translation standard text, and determining a sentence level text evaluation result based on the semantic accuracy and/or expression fluency of the clauses in the translation standard text;
step 133 determines a text evaluation result based on the word-level text evaluation result and/or sentence-level text evaluation result.
Specifically, for the target language text, the quality of the target language text may also be evaluated from different granularities.
The text evaluation result can be based on the semantics of the target language text and the source language standard text participle, and the translation accuracy of the participle in the target language text to the source language standard text is evaluated by taking the participle as a unit, so that a word level text evaluation result is obtained. The translation accuracy can represent the possibility that when the participles in the source language standard text are translated in sequence, the standard text participles are translated into the target language text participles at the corresponding positions. For example, for the chinese language "who is you", the corresponding english expression is "who are you", the word-level text evaluation result may be determined according to the translation accuracy of "you" corresponding to "you", the translation accuracy of "is" corresponding to "are", and the translation accuracy of "who" corresponds to "who".
The method can also be used for carrying out source language standardized translation on the target language text to obtain a translation standard text, and evaluating the semantic accuracy and/or expression fluency of clauses in the translation standard text by taking the clauses as units from the overall semantics of clauses of the translation standard text, thereby obtaining a sentence-level evaluation result of the text.
The text evaluation result may be determined according to the word-level text evaluation result, or the text evaluation result may be determined according to the sentence-level text evaluation result; or fusing the word-level text evaluation result and the sentence-level text evaluation result to determine a text evaluation result.
According to the method provided by the embodiment of the invention, the text evaluation result is determined based on the word level text evaluation result and/or sentence level text evaluation result, and the text quality is evaluated from different dimensions, so that the translation quality evaluation reliability of the target language text is improved.
Based on any of the above embodiments, fig. 5 is a second flowchart of the method for determining translation accuracy according to the present invention, as shown in fig. 5, the accuracy of translation of a term in a target language text to a standard text in a source language in step 131 is determined based on the following steps:
And step 520, determining the translation accuracy rate of the words in the target language text to the standard text in the source language based on the second technical field.
Specifically, in the process of determining the translation accuracy of the participles in the target language text to the standard text in the source language, the technical field of participles in the target language text is also considered, and the second technical field is determined according to the technical field of participles in the target language text.
It should be noted that, because the judgment criteria of the first technical field and the second technical field are different, the obtained second technical field may be the same as or different from the first technical field.
And then, in the second technical field, translating and decoding the target language text based on the coding features of the participles in the target language text to obtain the translation accuracy rate of the participles in the target language text to the standard text of the source language. The word-level text evaluation result can be the sum of the translation accuracy rates of the sub-words in the target language text to the standard text in the source language.
According to the method provided by the embodiment of the invention, based on the second technical field, the translation accuracy rate of the branch words in the target language text to the standard text in the source language is determined, and the reliability of the target language text evaluation result is further improved.
Based on any of the above embodiments, fig. 6 is a schematic flow chart of the method for determining semantic accuracy of a clause provided by the present invention, and as shown in fig. 6, the semantic accuracy of a clause is determined based on the following steps:
and step 620, based on the target standard language sequence, performing language sequence adjustment on clauses in the source language standard text or the translation standard text, and acquiring semantic accuracy of the clauses in the source language standard text or the translation standard text after the language sequence adjustment.
Specifically, the first technical field is a technical field determined by a technical field of word segmentation in a transcribed text based on a source language speech. Under the technical field, the standard language order of the source language is determined. The standard word order refers to a word order in the transcription text of the source language speech, which conforms to the expression habit of the source language, and is determined according to, for example, a subject, a predicate, an object, a shape of chinese.
The second technical field is the technical field determined based on the technical field of word segmentation in the target language text. Under the technical field, the standard language order of the target language is determined. For example, the target language is english, and the standard language order of the target language is determined according to the grammar and expression habit of english.
Based on the target standard language order, the language order of each clause in the source language standard text obtained by source language voice translation is adjusted, or the language order of each clause in the translation standard text obtained by target language text translation is adjusted, and the semantic accuracy of the clauses in the source language standard text or the translation standard text is determined according to the source language standard text or the translation standard text after the language order is adjusted.
Based on any embodiment, the expression fluency of a clause is determined based on the following steps: and determining the expression fluency of the clauses in the source language standard text or the translation standard text based on the relevance of the clauses in the source language standard text or the translation standard text clause.
Specifically, the relevance of the clauses in the source language standard text or the translation standard text clause is determined, so that the expression fluency of the clauses in the source language standard text or the translation standard text is determined based on the relevance between the clauses. The higher the relevance of the participles in the clause is, the better the semantic consistency of the participles in the clause is shown, and the higher the expression fluency of the clause is.
Furthermore, the determination of the expression fluency can be realized through a neural network model, firstly, word vectors of the participles in the clauses in the source language standard text or the translation standard text are determined, and the relevance of the participles in the clauses in the source language standard text or the translation standard text is determined by utilizing an attention mechanism, so that semantic coding vectors of the participles in the clauses in the source language standard text or the translation standard text are obtained through coding. And then, scoring the text semantic vectors obtained after the semantic coding vectors of the participles are fused by utilizing a scoring network, such as a softmax layer, so as to obtain the expression fluency of the clauses.
Based on any of the above embodiments, taking the source language as mandarin as an example, the present invention provides a translation quality evaluation method, which includes:
(1) determining original voice and target language text;
(2) carrying out standard translation on the original pronunciation to obtain standard text of the Mandarin; carrying out standard translation on the target language text to obtain a translation standard text;
(3) determining a first technical field based on the technical field of each word segmentation in the transcribed text of the original voice; in the first technical field, the translation accuracy of each participle in the standard mandarin text to the original speech is determined as a word-level speech evaluation result and is marked as ALanguage (1)；
(4) Determining a second technical field based on the technical field of each participle in the target language text; in the second technical field, the translation accuracy of each participle in the target language text to the standard mandarin standard text is determined as a word-level text evaluation result and is marked as ATranslation of characters；
(5) In the first technical field, determining the semantic accuracy of each clause in the standard text of Mandarin Chinese, and marking as B1(ii) a Determining the expression fluency of each clause in the standard text of the Mandarin Chinese, and marking as B 2(ii) a The sentence-level speech assessment result can be recorded as: b isLanguage (1)＝(B1+B2)/2；
(6) Under the second technical field, determining the semantic accuracy of each clause in the translation standard text, and marking as B3(ii) a Determining the expression fluency of each clause in the translation standard text, and marking as B4(ii) a The sentence-level text evaluation result can be recorded as: b isTranslation of characters＝(B3+B4)/2；
Further, the word-level evaluation result of the target language text may be denoted as (a)Language (1)+ATranslation of characters) 2; the word level evaluation result of the target language text can be recorded as (B)Language (1)+BTranslation of characters) 2; the translation quality of the target language text can be recorded as ((A)Language (1)+ATranslation of characters)/2+(BLanguage (1)+BTranslation of characters)/2)/2。
The following describes a translation quality evaluation apparatus provided by the present invention, and the translation quality evaluation apparatus described below and the translation quality evaluation method described above may be referred to in correspondence with each other.
Fig. 7 is a schematic structural diagram of a translation quality evaluation apparatus provided by the present invention, and as shown in fig. 7, the apparatus includes:
a speech and text determining unit 710 for determining a speech of a source language and a text of a target language;
the voice evaluation unit 720 is configured to perform source language standard translation on the source language voice to obtain a source language standard text, and determine a voice evaluation result based on the source language standard text;
The text evaluation unit 730 is used for determining a text evaluation result based on the target language text;
a translation quality determination unit 740 configured to determine the translation quality of the target language text based on the speech evaluation result and the text evaluation result.
The translation quality evaluation device provided by the embodiment of the invention obtains the source language standard text by performing source language standard translation on the source language speech, and determines the speech evaluation result based on the source language standard text; determining a text evaluation result based on the target language text; and then, the translation quality of the target language text is determined based on the voice evaluation result and the text evaluation result, so that the accuracy of the translation quality evaluation is improved.
Based on any of the above embodiments, the speech evaluation unit 720 is further configured to:
determining a word level voice evaluation result based on the translation accuracy of the segmented words in the source language standard text to the source language voice;
and/or determining sentence level voice evaluation results based on semantic accuracy and/or expression fluency of clauses in the source language standard text;
determining a speech evaluation result based on the word-level speech evaluation result and/or sentence-level speech evaluation result.
Based on any of the above embodiments, the translation quality evaluation apparatus further includes a translation accuracy determining unit configured to:
determining a first technical field based on the technical field of word segmentation in the transcribed text of the source language voice;
and determining the translation accuracy of the segmented words in the source language standard text to the source language voice based on the first technical field.
Based on any of the above embodiments, the text evaluation unit 730 is further configured to:
determining a word level text evaluation result based on the translation accuracy of the participles in the target language text to the source language standard text;
and/or performing source language standardized translation on the target language text to obtain a translation standard text, and determining a sentence level text evaluation result based on the semantic accuracy and/or expression fluency of the clauses in the translation standard text;
determining a text evaluation result based on the word-level text evaluation result and/or sentence-level text evaluation result.
Based on any of the above embodiments, the translation accuracy determining unit is further configured to:
determining a second technical field based on the technical field of word segmentation in the target language text;
and determining the translation accuracy of the participles in the target language text to the source language standard text based on the second technical field.
Based on any of the above embodiments, the translation quality evaluation apparatus further includes a semantic accuracy determination unit configured to:
determining a target standard language order, wherein the target standard language order is based on a standard language order of a source language in the first technical field, and/or the standard language order of a target language in the second technical field;
and based on the target standard language sequence, performing language sequence adjustment on clauses in the source language standard text or the translation standard text, and acquiring semantic accuracy of the clauses in the source language standard text or the translation standard text after the language sequence adjustment.
Based on any of the above embodiments, the translation quality evaluation apparatus further includes an expression fluency determination unit configured to:
and determining the expression fluency of the clauses in the source language standard text or the translation standard text based on the relevance of the clauses in the source language standard text or the translation standard text clause.
Fig. 8 illustrates a physical structure diagram of an electronic device, and as shown in fig. 8, the electronic device may include: a processor (processor)810, a communication Interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication Interface 820 and the memory 830 communicate with each other via the communication bus 840. Processor 810 may invoke logic instructions in memory 830 to perform a translation quality assessment method comprising: determining source language voice and target language text; performing source language standard translation on the source language voice to obtain a source language standard text, and determining a voice evaluation result based on the source language standard text; determining a text evaluation result based on the target language text; and determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result.
In addition, the logic instructions in the memory 830 can be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, and various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product includes a computer program, the computer program can be stored on a non-transitory computer readable storage medium, when the computer program is executed by a processor, a computer can execute the translation quality evaluation method provided by the above methods, the method includes: determining source language voice and target language text; performing source language standard translation on the source language voice to obtain a source language standard text, and determining a voice evaluation result based on the source language standard text; determining a text evaluation result based on the target language text; and determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result.
In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, the computer program being implemented by a processor to perform the translation quality assessment method provided by the above methods, the method including: determining source language voice and target language text; performing source language standard translation on the source language voice to obtain a source language standard text, and determining a voice evaluation result based on the source language standard text; determining a text evaluation result based on the target language text; and determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment may be implemented by software plus a necessary general hardware platform, and may also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A translation quality evaluation method, comprising:
determining source language voice and target language text;
performing source language standard translation on the source language voice to obtain a source language standard text, and determining a voice evaluation result based on the source language standard text;
determining a text evaluation result based on the target language text;
and determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result.
2. The translation quality evaluation method according to claim 1, wherein the determining a speech evaluation result based on the source language standard text comprises:
determining a word level voice evaluation result based on the translation accuracy of the segmented words in the source language standard text to the source language voice;
and/or determining sentence level voice evaluation results based on semantic accuracy and/or expression fluency of clauses in the source language standard text;
determining a speech evaluation result based on the word-level speech evaluation result and/or sentence-level speech evaluation result.
3. The translation quality evaluation method according to claim 2, wherein the translation accuracy of the segmented word in the source language standard text for the source language speech is determined based on the following steps:
Determining a first technical field based on the technical field of word segmentation in the transcribed text of the source language voice;
and determining the translation accuracy of the segmented words in the source language standard text to the source language voice based on the first technical field.
4. The translation quality evaluation method according to claim 1, wherein said determining a text evaluation result based on said target language text comprises:
determining a word level text evaluation result based on the translation accuracy of the participles in the target language text to the source language standard text;
and/or performing source language standardized translation on the target language text to obtain a translation standard text, and determining a sentence level text evaluation result based on the semantic accuracy and/or expression fluency of the clauses in the translation standard text;
determining a text evaluation result based on the word-level text evaluation result and/or sentence-level text evaluation result.
5. The translation quality evaluation method according to claim 4, wherein the translation accuracy of the participle in the target language text to the standard text in the source language is determined based on the following steps:
determining a second technical field based on the technical field of word segmentation in the target language text;
And determining the translation accuracy of the participles in the target language text to the source language standard text based on the second technical field.
6. The translation quality evaluation method according to any one of claims 2 to 5, wherein the semantic accuracy of the clause is determined based on the steps of:
determining a target standard language order, wherein the target standard language order is based on a standard language order of a source language in the first technical field, and/or the standard language order of a target language in the second technical field;
and based on the target standard language sequence, performing language sequence adjustment on clauses in the source language standard text or the translation standard text, and acquiring semantic accuracy of the clauses in the source language standard text or the translation standard text after the language sequence adjustment.
7. The translation quality evaluation method according to any one of claims 2 to 5, wherein the fluency of expression of the clauses is determined based on:
and determining the expression fluency of the clauses in the source language standard text or the translation standard text based on the relevance of the clauses in the clauses of the source language standard text or the translation standard text.
8. A translation quality evaluation apparatus, comprising:
The voice and text determining unit is used for determining source language voice and target language text;
the voice evaluation unit is used for performing source language standard translation on the source language voice to obtain a source language standard text and determining a voice evaluation result based on the source language standard text;
the text evaluation unit is used for determining a text evaluation result based on the target language text;
and the translation quality determining unit is used for determining the translation quality of the target language text based on the voice evaluation result and the text evaluation result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the translation quality assessment method according to any one of claims 1 to 7 when executing the program.
10. A non-transitory computer-readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the translation quality assessment method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210145951.2A CN114519358A (en) | 2022-02-17 | 2022-02-17 | Translation quality evaluation method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210145951.2A CN114519358A (en) | 2022-02-17 | 2022-02-17 | Translation quality evaluation method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114519358A true CN114519358A (en) | 2022-05-20 |
Family
ID=81599940
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210145951.2A Pending CN114519358A (en) | 2022-02-17 | 2022-02-17 | Translation quality evaluation method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114519358A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4290509A1 (en) * | 2022-06-07 | 2023-12-13 | Interprefy AG | Computer implemented method for evaluating one or more speech recognition systems |
-
2022
- 2022-02-17 CN CN202210145951.2A patent/CN114519358A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4290509A1 (en) * | 2022-06-07 | 2023-12-13 | Interprefy AG | Computer implemented method for evaluating one or more speech recognition systems |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220284198A1 (en) | Facilitating communications with automated assistants in multiple languages | |
| CN108766414B (en) | Method, apparatus, device and computer-readable storage medium for speech translation | |
| CN108447486B (en) | Voice translation method and device | |
| US8924210B2 (en) | Text processing using natural language understanding | |
| CN109635270B (en) | Bidirectional probabilistic natural language rewrite and selection | |
| CN111177324B (en) | Method and device for carrying out intention classification based on voice recognition result | |
| US11043213B2 (en) | System and method for detection and correction of incorrectly pronounced words | |
| CN112599128B (en) | Voice recognition method, device, equipment and storage medium | |
| CN112992125B (en) | Voice recognition method and device, electronic equipment and readable storage medium | |
| CN109448704A (en) | Construction method, device, server and the storage medium of tone decoding figure | |
| CN112016271A (en) | Language style conversion model training method, text processing method and device | |
| CN113707125A (en) | Training method and device for multi-language voice synthesis model | |
| CN113327574A (en) | Speech synthesis method, device, computer equipment and storage medium | |
| US10614170B2 (en) | Method of translating speech signal and electronic device employing the same | |
| CN115455946A (en) | Voice recognition error correction method and device, electronic equipment and storage medium | |
| CN114171002A (en) | Voice recognition method and device, electronic equipment and storage medium | |
| Granell et al. | Multimodality, interactivity, and crowdsourcing for document transcription | |
| CN114519358A (en) | Translation quality evaluation method and device, electronic equipment and storage medium | |
| Tündik et al. | Assessing the Semantic Space Bias Caused by ASR Error Propagation and its Effect on Spoken Document Summarization. | |
| CN116933806A (en) | Concurrent translation system and concurrent translation terminal | |
| CN114783405B (en) | Speech synthesis method, device, electronic equipment and storage medium | |
| CN114254649A (en) | Language model training method and device, storage medium and equipment | |
| CN113096667A (en) | Wrongly-written character recognition detection method and system | |
| CN112686060A (en) | Text translation method and device, electronic equipment and storage medium | |
| Peng | [Retracted] An English Teaching Pronunciation Detection and Recognition Algorithm Based on Cluster Analysis and Improved SSD |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |