CN108597493B - The audio exchange method and audio exchange system of language semantic - Google Patents
The audio exchange method and audio exchange system of language semantic Download PDFInfo
- Publication number
- CN108597493B CN108597493B CN201810264460.3A CN201810264460A CN108597493B CN 108597493 B CN108597493 B CN 108597493B CN 201810264460 A CN201810264460 A CN 201810264460A CN 108597493 B CN108597493 B CN 108597493B
- Authority
- CN
- China
- Prior art keywords
- language
- voice
- semantic
- phoneme
- coding
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000013507 mapping Methods 0.000 claims abstract description 84
- 238000006243 chemical reaction Methods 0.000 claims abstract description 28
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 11
- 108091026890 Coding region Proteins 0.000 claims description 31
- 238000013519 translation Methods 0.000 abstract description 7
- 230000008859 change Effects 0.000 abstract description 6
- 239000002131 composite material Substances 0.000 abstract description 3
- 239000012634 fragment Substances 0.000 abstract description 3
- 230000006978 adaptation Effects 0.000 description 40
- 238000010586 diagram Methods 0.000 description 14
- 230000008569 process Effects 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000004044 response Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000003068 static effect Effects 0.000 description 4
- 150000001875 compounds Chemical class 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000005236 sound signal Effects 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 108010076504 Protein Sorting Signals Proteins 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 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
- G10L15/005—Language recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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
-
- 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
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
-
- 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
- G10L15/04—Segmentation; Word boundary detection
-
- 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
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/1822—Parsing for meaning understanding
-
- 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
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
- G10L2015/025—Phonemes, fenemes or fenones being the recognition units
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Acoustics & Sound (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Machine Translation (AREA)
Abstract
Audio exchange method, system and the audio coding figure of language semantic of the invention, because semantic complexity causes data to respond the technical problem to go wrong with real-time difference when solving linguistic intertranslation in the prior art.Method forms the voice mapping structure of each language including the use of minimum aligned phoneme sequence, completes to convert between semantic language by each voice mapping structure.Using language constitute in formed audio minimum short section minimum phoneme as the master data crosspoint of semantic conversion between each language, using minimum phoneme as the basis of coding of data exchange, change the foundation structure of speech recognition, optimize the codec complexity and accuracy rate of language sound intermediate frequency content, so that avoiding the complex audio feature for being coupled the formation of the composite informations such as language fragments medium pitch, scale, range in the cataloged procedure of language audio, phonetic recognization rate ensure that.The mapping structure of the voice coding and literal code that are formed using minimum phoneme is improved data exchange efficiency when language translation.
Description
Technical field
The present invention relates to information exchange fields, and in particular to a kind of audio exchange method of language semantic and audio exchange system
System.
Background technique
Current language translation mainly synthesizes several parts by speech recognition, semantic analysis and sentence and forms, and speech recognition is adopted
With high sensor, audio corresponding with text in sentence is extracted from the frequency domain or time domain speech signal stream of opriginal language
Signal set, semantic analysis utilize the models pair such as hidden Markov model (HMM), self learning model, artificial neural network (ANN)
Word sequence and semantic meaning in audio signal collection are identified and are quantified to determine expression content, sentence synthesis as far as possible
The audio signal collection or word sequence set of object language are formed according to the identification of expression content and quantized data.In this mistake
The computing resource for being influenced by semantic analysis model complexity to need magnanimity in journey, for the application of mobile terminal need using point
The computing architecture of cloth, using the computing resource at the guaranteed bandwidth access service end of internet, therefore the real-time and standard translated
True property is restricted.
In patent document CN104637482B, disclose it is a kind of utilize digital coding realize dress of the voice to text conversion
It sets, wherein storing first language phoneme characteristic using phoneme storage unit;Using phoneme conversion unit by received phoneme
Signal sequence is converted to first language phoneme by first language phoneme characteristic;It is first language using digital coding unit
Phoneme carries out unique encodings, forms first language phoneme encoding sequence;The first language is formed using first language phoneme encoding sequence
The word pronunciation code sequence and vocabulary pronunciation code sequence of speech;Using the word storing unit storage word of first language, vocabulary or
Figure and corresponding coded sequence;First language is generated according to the corresponding relationship of coded sequence using words converting unit
Word, vocabulary, figure and/or combination thereof.There are the bases of coding mapping between the device description words and voice.How coding is utilized
The resource consumption of the picture and text audio conversion of identical semanteme needs inventive improvements between mapping basis reduction language.
Summary of the invention
In view of this, the embodiment of the present invention is dedicated to providing audio exchange method and the audio exchange system of a kind of language semantic
System, semanteme complexity leads to the technical problem of data response and real-time difference when solving linguistic intertranslation in the prior art.
The audio exchange method of the language semantic of the embodiment of the present invention, forms the voice of each language using minimum aligned phoneme sequence
Mapping structure is completed to convert between semantic language by each voice mapping structure.
The audio exchange system of the language semantic of the embodiment of the present invention characterized by comprising
Memory, the program code of the audio exchange method for storing above-mentioned language semantic;
Processor, for running said program code.
The audio exchange system of the language semantic of the embodiment of the present invention, for forming each language using minimum aligned phoneme sequence
Voice mapping structure is completed to convert between semantic language by each voice mapping structure.
The basic voice coding figure of the embodiment of the present invention, for the graphic software platform of language phoneme, including basic framework,
The basic framework includes the first adaptation column, the second adaptation column and adapter rod arranged side by side, the first adaptation column and described second
Adaptation hyte is respectively set in adaptation column, and the adaptation hyte includes several adaptation positions, and the both ends of the adapter rod respectively connect one
One adaptation position of a adaptation column.
The audio exchange method and audio exchange system of the language semantic of the embodiment of the present invention, coded graphics utilize language structure
Master data crosspoint at the middle minimum phoneme for forming audio minimum short section as semantic conversion between each language, utilizes minimum
Basis of coding of the phoneme as data exchange, changes the foundation structure of speech recognition, simplifies the volume of language sound intermediate frequency content
Code length and code efficiency, so that data exchange efficiency when language translation is optimized, to reduction remote data real-time response
Time delay, improving the memory capacity of basic data structure and basic data in local mobile terminal has positive influence.
Detailed description of the invention
Fig. 1 show the data handling procedure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.
Fig. 2 show the cataloged procedure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.
Fig. 3 show the voice mapping structure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.
Fig. 4 show the voice mapping structure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.
The audio exchange method that Fig. 5 show one embodiment of the invention language semantic carries out the schematic diagram of language conversion.
Fig. 6 show the configuration diagram of the audio exchange system of one embodiment of the invention language semantic.
Fig. 7 show a kind of figure of basic voice coding figure in the audio exchange method of language semantic of the embodiment of the present invention
Shape structural schematic diagram.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that the described embodiment is only a part of the embodiment of the present invention, instead of all the embodiments.Based on this
Embodiment in invention, every other reality obtained by those of ordinary skill in the art without making creative efforts
Example is applied, shall fall within the protection scope of the present invention.
The audio exchange method of the language semantic of the embodiment of the present invention, comprising:
The voice mapping structure that each language is formed using minimum aligned phoneme sequence completes semanteme by each voice mapping structure
It is converted between language.
There are essence difference on picture and text and pronunciation, semantic conversion refers to identical semanteme for the expression of identical semanteme between language
The conversion of different picture and text and pronunciation expression-form.
The pronunciation of the semantic text (one kind as graphical symbol) of regional all-purpose language expression has certainty, word
The pronunciation law of remittance and sentence can be summarized as the various combination of syllable.And one group of basic minimum phoneme is used to constitute each sound
The low signal Load Characteristics that section can use minimum phoneme exclude audio redundant signals and interference information, mention for complex data exchange
For the basis of coding more simplified, code length is reduced.
It is compared according to statistics of the those skilled in the art to each regional all-purpose language, most as pronunciation fundamental
It was determined that quantity is less than 1000, the world 7000 or so, which is planted, amounts to 800 left sides in language for small phoneme quantity and its audio frequency characteristics
Right unduplicated minimum phoneme, wherein every kind of western language about uses 40 or so minimum phonemes, Chinese is no more than 150 left sides
Right minimum phoneme, can be established completely using the block code of hundreds value range or kilobit numberical range index be, for example, ten into
Make three digits or four figures, e.g. 0 digit of binary one or 20 digits.
The audio exchange method of the language semantic of the embodiment of the present invention utilizes formation audio minimum short section in language composition
Master data crosspoint of the minimum phoneme as semantic conversion between each language, using minimum phoneme as the coding of data exchange
Basis changes the foundation structure of speech recognition, the code length and code efficiency of language sound intermediate frequency content is simplified, so that language
Say the complicated sound for avoiding being coupled the formation of the composite informations such as language fragments medium pitch, scale, range in the cataloged procedure of audio
Frequency feature, ensure that phonetic recognization rate, and the mapping structure of the voice coding and literal code that are formed using minimum phoneme makes language
Data exchange efficiency when speech translation is optimized.To reduce remote data real-time response time delay, improve basic data structure and
Memory capacity of the basic data in local mobile terminal has positive influence.
Fig. 1 is the data handling procedure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.Such as Fig. 1 institute
Show, comprising:
Step 100: all minimum phonemes of serializing.
Serialization process may include the identification to syllable, phoneme, scale, intonation in language, syllable, sound to identification
The quantitative mathematical description of element, scale, intonation, such as time domain or the audio characteristic data of frequency domain, the knot of data is described to quantitative mathematical
Structureization storage, such as coding forms index one by one.
Step 200: being formed by the subset of all minimum phonemes between the text-to-speech of each language and map data.
The pronunciation basis of every kind of language is determined by the subset of all minimum phonemes, passes through the group of phoneme minimum in subset
The voice identifier for forming word pronunciation in a kind of language is closed, and then forms using voice identifier that text is corresponding between voice identifier to be tied
The mapping data of structure, mapping data include the data structure of storing data.Mapping data may include reflecting between text and voice
Penetrate the mapping data between data and voice.
Step 300: being formed by language semantic between the voice of each language and map data.
The mapping data that correspond to the voice of meaning are established between language using semantic objectivity, and mapping data include storage number
According to data structure.It also may include the mapping data between text and voice.
Step 400: forming semantic language turn using mapping data between data and text-to-speech are mapped between corresponding voice
It changes.
The audio exchange method of the language semantic of the embodiment of the present invention ensure that one kind by mapping data between text-to-speech
The continuity and correctness of the Text-to-Speech of language map the combination that data are mapped between data and text-to-speech between voice
The conversion diversity between language is allowed to realize higher language in conversion process while guaranteeing the conversion quality between language
Say basic data interactive efficiency.It simultaneously can be with shape by mapping the mapping variation of mapping data between data and text-to-speech between voice
At further cipher round results.
Fig. 2 is the cataloged procedure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.As shown in Fig. 2,
On that basi of the above embodiments, step 100 includes:
Step 110: the minimum phoneme of each all-purpose language is acquired by speech recognition.
Based on human physiological's feature and language evolution, the voice of language can be decomposed into be pronounced to word pronunciation extremely by sentence
Word syllable constitutes the STRUCTURE DECOMPOSITION of phoneme to syllable.It will be appreciated by those skilled in the art that carrying out audio using computer technology
Acquisition and time domain or the frequency domain character analysis of audio fragment can determine the audio frequency characteristics of word, word, phrase, determine including
Minimum phoneme feature.
Step 120: minimum phoneme is formed into unified aligned phoneme sequence.
It will be appreciated by those skilled in the art that by speech recognition technology, in conjunction with the speech analysis and statistics of necessary data amount
The minimum phoneme audio frequency characteristics used in each language can be identified and be determined.By the audio frequency characteristics of each of determining minimum phoneme
Unified mark coding, forms the unified aligned phoneme sequence of all minimum phonemes.Unified aligned phoneme sequence allows the voice of language quasi-
The determining combination to be formed by least one minimum phoneme really is deconstructed, determines that combination can pass through unified aligned phoneme sequence acquisition pair
The coded sequence answered.
Such as: syllable is formed using initial consonant and simple or compound vowel of a Chinese syllable in Chinese, initial consonant is by single minimum phoneme or several single minimum sounds
Element is formed, and simple or compound vowel of a Chinese syllable is formed by one or several minimum phonemes, forms syllable using vowel and consonant in similar English, vowel by
Single minimum phoneme or several single minimum phonemes are formed, and consonant is formed by one or several minimum phonemes, the unified sound of formation
It the part of prime sequences can be as shown in the table:
Single minimum phoneme in table in unified aligned phoneme sequence has unique encodings in unified aligned phoneme sequence.For being less than
1000 minimum phonemes can form unique encodings using 10bit (bit) length.
The audio exchange method of language semantic of the embodiment of the present invention forms unified aligned phoneme sequence as same or similar semanteme
The essential information carrier of text or voice conversion between different language, avoids other kinds of composite audio carrier (such as sound
Section) entrained by excessive redundancy formed information interference, be conducive to optimize speech recognition accuracy and recognition efficiency.Most
Small phoneme further can be updated unified aligned phoneme sequence with language evolution using unified aligned phoneme sequence, keep to each language
The synchronous variation of speech sound.
As shown in Fig. 2, step 200 includes: in the audio exchange method of one embodiment of the invention language semantic
Step 210: forming the pronunciation with individual character or word in first language using a part of phoneme in unified aligned phoneme sequence
Corresponding first basic voice coding sequence.
A part of phoneme includes a kind of all minimum phonemes of language pronouncing, can form syllable using this part of phoneme
And then form the pronunciation of the language word or word.Coding based on minimum phoneme in unified aligned phoneme sequence, forms the first language
The basic voice coding of each individual character or word is called the turn, and then forms the basic voice coder of all (or main) individual characters or word
Code sequence.
Such as: " mother " word in Chinese, phonetic are " ma ", including phoneme " m " and " a ", and " m " is in unified aligned phoneme sequence
Be encoded to 120, " a " is encoded to 010 in unified aligned phoneme sequence, then " mother " word is in the basic voice coding sequence of Chinese
Be encoded to 120010.
It in an embodiment of the present invention can also be using other coding compress modes, such as the phoneme for including by " mother " word
Coding adds up, and formation is encoded to 130.Or use the patterned mode of basic voice coding.
It will be appreciated by those skilled in the art that the coding form in basic voice coding sequence in citing is there are redundancy, by
Minimum phoneme encoding effect length can use compression coding technology using the basic voice coding sequence of standard byte and keep compiling
Code uniqueness and compared with lower Item length.
It will be appreciated by those skilled in the art that it is identical basic that there are the different individual characters of same pronunciation or word can have
The different pronunciations of voice coding, individual character or word can make same individual character or word have different basic voice codings.
Step 220: being formed using the first basic voice coding sequence corresponding with phrase in first language or sentence pronunciation
First voice mapping structure.
In the basic voice coding sequence basis that individual character or word determine, the voice mapping structure of phrase or sentence can be with
Form the voice mapping structure that phrase or sentence are formed based on basic voice coding sequence extension.
Voice mapping structure can be using with address feature and addressable data structure, such as team either statically or dynamically
The single form or combining form of column, array, heap, storehouse, chained list, tree or figure etc., can use either statically or dynamically pointer can be with
Realize different data structure type address arithmetic, each data structure involved in voice mapping structure may exist comprising or
Side by side.
In an embodiment of the present invention, it can be formed to pointer with related semantic meaning using above-mentioned data structure
The mapping structure of voice and semanteme between word, word, language, sentence, by establishing part of speech mapping structure with semantic meaning.
Fig. 3 is the voice mapping structure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.Such as Fig. 3 institute
Show, for Chinese, by taking " hair " word, " bright " word, " wound " word, " making " word as an example, each word utilizes correspondence as minimum semantic primitive
The phoneme of pronunciation establishes corresponding basic voice coding, has discreteness between the basic voice coding of each word.Individual character is with chained list knot
Structure (only as an example) stores single character code (the i.e. phoneme feature) filter efficiency that can guarantee high speed.It is formed with individual character
Each word with semantic meaning such as " invention ", " creation " are with the storage of another list structure, the basic voice coder of each word
Code is formed using the basic voice coding of included individual character, has discreteness between the basic voice coding of each word.With individual character or
Phrase of each of the word formation with semantic meaning is with structure of arrays (only as an example) storage, it is ensured that quickly seeks
The efficiency of location and data topology update variation, has discreteness between the basic voice coding of each phrase.
Word, word, phrase are formed according to the semantic dependency of word, word, phrase using the address pointer in data structure
The mapping structure tree of correlation or mapping structure figure, so that form mapping association between voice and semanteme, this mapping association can be with
It is that static or some movable state updates.
In basic vocoded data structure, (perhaps word or phrase) data cell of each word can be with
Extension, such as it is extended to queue, for storing the different semantic words (perhaps word or phrase) of same pronunciation, voice is reflected
Penetrate structure multi-dimension.
The audio exchange method of language semantic of the embodiment of the present invention is made using the data store organisation of voice mapping text
The major part of voice mapping structure is static structure, and it is excellent can to form structure by the computing capability in server end or cloud
Change, a small amount of dynamic update and supplement can be completed using less computing resource in client.Since phoneme in pronunciation is utilized
The basic voice coding sequence formed greatly reduces complexity and data volume for semantic voice mapping structure, so that
The data storage and data processing of voice mapping structure can complete response in client and server-side under low time delay state.
As shown in Fig. 2, step 200 in the audio exchange method of one embodiment of the invention language semantic further include:
Step 230: forming individual character in second language or pronunciation of words using another part phoneme in unified aligned phoneme sequence
Second basic voice coding sequence.
Another part phoneme may include the identical phoneme in part compared with phoneme a part of in above-mentioned steps 130, or
The identical phoneme of person is with the word or symbol logo in different language.
Such as: " and " its phonetic symbol is in EnglishIncluding phoneme" n " and " d "," n " and
" d " is encoded to 018,220 and 200 in unified aligned phoneme sequence, then " and " word is in the basic voice coding sequence of English
Be encoded to 018220200.
It will be appreciated by those skilled in the art that citing in basic voice coding sequence in coding form there are redundancies, can
To keep the uniqueness of coding using compression coding technology and compared with lower Item length.
It will be appreciated by those skilled in the art that it is identical basic that there are the different individual characters of same pronunciation or word can have
The different pronunciations of voice coding, individual character or word can make same individual character or word have different basic voice codings.
Step 240: being formed using the second basic voice coding sequence corresponding with phrase in second language or sentence pronunciation
Second voice mapping structure.
The text (or symbol) of identical semanteme has the possibility of same pronunciation, the difference of identical semanteme in different language
The same pronunciation of text generates encoding variability with the formation of macaronic voice mapping structure.
Fig. 4 is the voice mapping structure schematic diagram of the audio exchange method of one embodiment of the invention language semantic.Such as Fig. 4 institute
Show, for English, by taking " invention ", " creation " as an example, each word is sent out as minimum semantic primitive using corresponding
The phoneme of sound establishes corresponding basic voice coding, has discreteness between the basic voice coding of each word.Word is with database
Form structure (only as an example) storage can guarantee high speed word coding (i.e. phoneme feature) filter efficiency.With list
Morphology at each of the phrase with semantic meaning stored with the form structure (only as an example) of database, it is ensured that
The efficiency of immediate addressing and data topology update variation, has discreteness between the basic voice coding of each phrase.
Word, phrase correlation are formed according to the semantic dependency of word, phrase using the address pointer in data structure
Mapping structure tree or mapping structure figure so that forming mapping association between voice and semanteme, this mapping association can be static state
Or some movable state update.
In basic vocoded data structure, the data cell of each word or phrase can be extended to queue,
It is for storing same pronunciation different semantic words or phrase, voice mapping structure is multidimensional.
The audio exchange method of language semantic of the embodiment of the present invention is made using the data store organisation of voice mapping text
The major part of voice mapping structure is static structure, and it is excellent can to form structure by the computing capability in server end or cloud
Change, a small amount of dynamic update and supplement can be completed using less computing resource in client.Since phoneme in pronunciation is utilized
Basic voice coding sequence, the complexity and data volume for semantic voice mapping structure are greatly reduced, so that voice
The data storage and data processing of mapping structure can complete response in client and server-side under low time delay state.
As shown in Fig. 2, step 300 in the audio exchange method of one embodiment of the invention language semantic further include:
Step 310: pass through each first language and second language using same or similar semantic information (i.e. first and the
Two) voice mapping structure forms the voice primary transformational structure between corresponding language.
Using macaronic voice mapping structure based on same or similar semantic information between needing the language translated
The voice primary transformational structure between the individual character or word of same or similar meaning is formed, macaronic individual character, word, short is stored
Basic voice coding between language or sentence, voice primary transformational structure can be stored using the structure of " key: key assignments ", big to respond
Measure the filter efficiency of concurrent request.
For example, by using
It is semantic: the basic voice coding of English: the basic voice coding of Chinese
Innovation and creation: 092072069:710169555614
The basic voice coding of English and the basic voice coding of Chinese can key and key assignments each other, be used for two-way translation.
As shown in Fig. 2, step 300 in the audio exchange method of one embodiment of the invention language semantic further include:
Step 320: forming corresponding (i.e. first and second) voice using the syntax rule of first language and second language and reflect
Penetrate the advanced transformational structure of interstructural voice.
The syntax rule of each language includes the language between the root according to individual character or word and the individual character or word of part of speech foundation
Pitch class transformational structure.According to voice primary transformational structure, the advanced transformational structure of voice can be deposited using the structure of " key: key assignments "
Storage, to respond the filter efficiency of a large amount of concurrent requests.
For example, by using
It is semantic: grammer: the basic voice coding of English
English " creates (noun) " 0001:092072069;
English " creates (verb) " 0002:092072069;
English " creates (adverbial word) " 0003:092072069;
It is semantic: grammer: the basic voice coding of Chinese
Chinese " creates (noun) " 0001:710169555614;
Chinese " creates (verb) " 0002:710169555614;
Chinese " creates (adverbial word) " 0003:710169555614;
The basic voice coder of the individual character, word or vocabulary with similar semantic will be formed in bilingual according to different grammers
Code opposite can be assembled, and coding dependency improves, and improve filter efficiency and machine translation efficiency of algorithm in translation process.
Fig. 5 is that the audio exchange method of one embodiment of the invention language semantic carries out the schematic diagram of language conversion.Such as Fig. 5 institute
Show, step 400 includes:
Step 410: the sequence set of phonemes of the audio input segment of first language is obtained using speech recognition;
Step 420: determining the first basic of sequence set of phonemes using the first basic voice coding sequence of first language
Voice coding;
Step 430: determining sequence using the first voice mapping structure of first language and the first basic voice coding sequence
The continuous speech of set of phonemes encodes;
Step 440: the second basic voice coder of second language is obtained using the voice primary transformational structure between corresponding language
Code;
Step 450: using between corresponding language the advanced transformational structure of voice and the second basic voice coding sequence obtain the
The continuous speech of two language encodes;
Step 460: encoding to form sound pronunciation according to the continuous speech of second language.
The audio exchange method of language semantic of the embodiment of the present invention carries out utilizing the aligned phoneme sequence-base formed when language conversion
The transformational structure that this voice coding sequence-is formed between voice mapping structure and language can between voice and text between completing bilingual
Inverse conversion is conducive to voice and converts accurately or relatively accurately obtain corresponding alternative spelling words intellectual.Data and data structure
Sizes of memory is limited, and retrieval difficulty is lower, suitable for being locally stored and handling, reality that whole process responds server-side request of data
When property and bandwidth requirement be not high.Fig. 6 is the configuration diagram of the audio exchange system of one embodiment of the invention language semantic.Such as figure
Shown in 6, the audio exchange system of the embodiment of the present invention, the voice for forming each language using minimum aligned phoneme sequence maps knot
Structure is completed to convert between semantic language by each voice mapping structure.
As shown in fig. 6, the audio exchange system of the embodiment of the present invention includes:
Device 1100 is serialized, for serializing all minimum phonemes.
Phoneme maps to form device 1200 in language, for forming the text of each language by the subset of all minimum phonemes
Data are mapped between voice.
Phoneme maps to form device 1300 between language, maps number between the voice for forming each language by language semantic
According to.
Language converting device 1400, for using mapping data are formed between mapping data and text-to-speech between corresponding voice
Semantic language conversion.
As shown in fig. 6, serializing device 1100 includes: in the audio exchange system of the embodiment of the present invention
Phoneme recognition module 1110, for acquiring the minimum phoneme of each all-purpose language by speech recognition.
Phoneme encoding module 1120, for minimum phoneme to be formed unified aligned phoneme sequence.
As shown in fig. 6, phoneme maps to form device 1200 and include: in language in the audio exchange system of the embodiment of the present invention
First voice coding establishes module 1210, for being formed and the first language using a part of phoneme in unified aligned phoneme sequence
Call the turn the corresponding first basic voice coding sequence of pronunciation of individual character or word.
Module 1220 is established in the mapping of first voice, for using in the first basic voice coding sequence formation and first language
Phrase or the corresponding first voice mapping structure of sentence pronunciation.
Second voice coding establishes module 1230, for forming the second language using another part phoneme in unified aligned phoneme sequence
Call the turn the second basic voice coding sequence of individual character or pronunciation of words.
Module 1240 is established in the mapping of second voice, for using in the second basic voice coding sequence formation and second language
Phrase or the corresponding second voice mapping structure of sentence pronunciation.
As shown in fig. 6, phoneme maps to form device 1300 and include: between language in the audio exchange system of the embodiment of the present invention
Language construction primary conversion module 1310, for using same or similar semantic information by each first language and
(i.e. first and second) the voice mapping structure of second language forms the voice primary transformational structure between corresponding language.
The advanced conversion module 1320 of language construction, for being formed accordingly using the syntax rule of first language and second language
The advanced transformational structure of voice between (i.e. first and second) voice mapping structure.
As shown in fig. 6, language converting device 1400 includes: in the audio exchange system of the embodiment of the present invention
Phoneme recognition module 1410, the sequence phoneme of the audio input segment for obtaining first language using speech recognition
Set;
First basic coding identification module 1420 determines suitable for the first basic voice coding sequence using first language
The basic voice coding of the first of sequence set of phonemes;
First continuous speech coding module 1430, substantially for the first voice mapping structure and first using first language
Voice coding sequence determines the continuous speech coding of sequence set of phonemes;
Second basic coding identification module 1440, for obtaining second using the voice primary transformational structure between corresponding language
The basic voice coding of the second of language;
Second continuous speech coding module 1450, for utilizing the advanced transformational structure of voice and the second base between corresponding language
This voice coding sequence obtains the continuous speech coding of second language;
Continuous programming code conversion module 1460 encodes to form sound pronunciation for the continuous speech according to second language.
Those of ordinary skill in the art may be aware that list described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually
It is implemented in hardware or software, the specific application and design constraint depending on technical solution.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
The scope of the present invention.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided herein, it should be understood that disclosed systems, devices and methods, it can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components
It can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown or
The mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, the indirect coupling of device or unit
It closes or communicates to connect, can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be a
People's computer, server or network equipment etc.) it performs all or part of the steps of the method described in the various embodiments of the present invention.
And storage medium above-mentioned includes: that USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
The various media that can store program ver-ify code such as reservoir (RAM, Random Access Memory), magnetic or disk.
In the audio exchange method of one embodiment of the invention language semantic, for utilizing the part in unified aligned phoneme sequence
Minimum phoneme forms the basic voice coding sequence of individual character or pronunciation of words in a kind of language, and basic voice coding therein can be with
Additional pictorial symbols are formed, it is corresponding with the individual character or word accordingly to pronounce.Graphical using basic voice coding can be with
The pronunciation of individual character or word that phoneme is formed is converted to visual identity, is conducive to Computer Vision Recognition and machine word
The communication of sound identification, so that the voice conversion of identical semanteme can have the basis of Computer Vision Recognition between language.
Fig. 7 show a kind of figure of basic voice coding figure in the audio exchange method of language semantic of the embodiment of the present invention
Shape structural schematic diagram.As shown in the part a of Fig. 7, graphic structure includes the basic framework 01 of a H-shaped, and basic framework includes simultaneously
Column further include both ends point in vertical parallel the first adaptation column 10 (bar paten) and the second adaptation column 20 (bar paten)
The adapter rod 30 (bar paten) not connect with the first adaptation column and the second adaptation column.
The first adaptation hyte 11 is provided on first adaptation column (being left side in figure), second is adapted on column (being right side in figure)
It is provided with the second adaptation hyte 21, third adaptation hyte 31 is provided in adapter rod 30, the end of adapter rod 30 is connected to correspondence
Side is adapted on the adaptation position of column, is adapted in the adaptation hyte of column and is included at least three adaptation positions (what is provided in attached drawing is 5).
Adjacent adaptation position is used to adjust the length of adaptation column in same adaptation hyte, is overlapped to form adaptation by being adapted to position
The specific adjusted of column, so that accordingly the length of adaptation column forms corresponding change, the adaptation position that can be overlapped includes at least two.
The end of adapter rod 30 can connect on the coincidence adaptation position of corresponding side adaptation column.
In practical applications, the syllable that the phoneme encoding or phoneme that can will form the pronunciation syllable of individual character or word are formed
Coding is reflected in the first adaptation column, second accommodates in the connection change in shape of adapter rod, utilizes the fixation position of adaptation position and suitable
The coincidence variation of coordination forms the encoded content of enough permutation and combination reflection syllables.
As shown in the part b of Fig. 7 and the part c, it can also include in an embodiment of the present invention and be adapted to the auxiliary of position connection
Help adaptation symbol 40, auxiliary adaptation symbol 40 include have direction vector vector line segment 41 and not direction vector standard symbol
Numbers 42.Vector line segment 41 can be line segment or minor arc, and standard symbol 42, which can be round or annular, vector line segment, can one
Or it is multiple, standard symbol can have one or more.
In practical applications additional vector line segment and standard symbol to can will be relevant with syllable after the connection of adaptation position
The additional audio features such as intonation, the tone increase the information load of syllable coding in conjunction with syllable coding.
In practical application, such as Chinese, as shown in the part b of Fig. 7 and the part c, the part b is individual character " rear " and " time "
The correspondence figure of voice coding, the part c are the correspondence figure of individual character " mouth " and " bandit " voice coding, the pronunciation of above-mentioned each individual character
The breeder mother of syllable shows the fit structure of the length variation and vector line segment 41 of the first adaptation column on the left of basic framework, simple or compound vowel of a Chinese syllable
Show the fit structure of the length variation and vector line segment 41 and standard symbol 42 of the second adaptation column on the right side of basic framework.Base
This frame is adapted to the smoothed processing of symbol with auxiliary and can not only keep figure beautiful but also can guarantee Computer Vision Recognition quality.
Adaptation position and adapter rod 30 and the link position that is adapted to position, basic subrack as shown in the part d of Fig. 7, using coincidence
Frame 01 can be converted to n shape from H-shaped, as shown in the part e of Fig. 7, using coincidence adaptation position and adapter rod 30 be adapted to position
Link position, basic framework 01 can be converted to U-shaped from H-shaped.
As shown in the part d of Fig. 7, the first, second adaptation column around basic framework (H-shaped, n shape or U-shaped) is directly marked
The coding of minimum phoneme, encodes digital number and the adaptation position of corresponding adaptation column is corresponding.Utilize sound minimum in a kind of linguistic syllables
The direct coding of element is shown, phonographic alphabet-phoneme encoding-voice of language is directly done visual expression, so that macaronic
Computer vision conversion may be implemented in basic voice coding figure, while voice conversion, is identified and is protected using computer graphical
Demonstrate,prove the discrimination of language identification.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.
Claims (7)
1. a kind of audio exchange method of language semantic, which is characterized in that form the voice of each language using minimum aligned phoneme sequence
Mapping structure is completed to convert between semantic language by each voice mapping structure;Wherein
The voice mapping structure for forming each language using minimum aligned phoneme sequence includes: all minimum phonemes of serializing;Pass through
The subset of all minimum phonemes maps data between forming the text-to-speech of each language;It is formed by language semantic described
Data are mapped between the voice of each language;And
Conversion includes: to map number using between the corresponding voice between the language for completing semanteme by each voice mapping structure
Semantic language conversion is formed according to data are mapped between the text-to-speech;Wherein
Described using mapping between data and text-to-speech that mapping data form semantic language conversion between corresponding voice includes: benefit
The sequence set of phonemes of the audio input segment of first language is obtained with speech recognition;Utilize the first basic voice of first language
Coded sequence determines the first basic voice coding of sequence set of phonemes;Utilize the first voice mapping structure of first language and
One basic voice coding sequence determines the continuous speech coding of sequence set of phonemes;It is converted using the voice primary between corresponding language
Second basic voice coding of structure acquisition second language;Substantially using the advanced transformational structure of voice and second between corresponding language
Voice coding sequence obtains the continuous speech coding of second language;It is encoded to form voice hair according to the continuous speech of second language
Sound.
2. the audio exchange method of language semantic according to claim 1, which is characterized in that all minimums of serializing
Phoneme includes:
The minimum phoneme of each all-purpose language is acquired by speech recognition;
The minimum phoneme is formed into unified aligned phoneme sequence.
3. the audio exchange method of language semantic according to claim 2, which is characterized in that it is described by it is described it is all most
Mapping data include: between the subset of small phoneme forms the text-to-speech of each language
Corresponding with the pronunciation of individual character in first language or word the is formed using a part of phoneme in the unified aligned phoneme sequence
One basic voice coding sequence;
The first voice corresponding with phrase in first language or sentence pronunciation is formed using the described first basic voice coding sequence
Mapping structure;
It is basic using the second of individual character in another part phoneme formation second language in the unified aligned phoneme sequence or pronunciation of words
Voice coding sequence;
The second voice corresponding with phrase in second language or sentence pronunciation is formed using the described second basic voice coding sequence
Mapping structure.
4. the audio exchange method of language semantic according to claim 3, which is characterized in that described to pass through language semantic shape
Include: at data are mapped between the voice of each language
It is formed using same or similar semantic information by the voice mapping structure of the first language and the second language
Voice primary transformational structure between corresponding language;
The voice between the first language and the voice mapping structure of the second language is formed using the syntax rule of each language
Advanced transformational structure.
5. the audio exchange method of language semantic according to claim 1, which is characterized in that the minimum aligned phoneme sequence is adopted
It is established and is indexed with the block code of hundreds value range or kilobit numberical range.
6. a kind of audio exchange system of language semantic characterized by comprising
Memory, the program code of the audio exchange method for storing language semantic as claimed in claim 1 to 5;
Processor, for running said program code.
7. a kind of audio exchange system of language semantic, the voice for forming each language using minimum aligned phoneme sequence maps knot
Structure is completed to convert between semantic language by each voice mapping structure, comprising:
Device is serialized, for serializing all minimum phonemes;
Phoneme maps to form device in language, reflects between the text-to-speech for forming each language by the subset of all minimum phonemes
Penetrate data;
Phoneme maps to form device between language, maps data between the voice for forming each language by language semantic;
Language converting device, for using mapping data form semantic language between mapping data and text-to-speech between corresponding voice
Speech conversion;Wherein
The language converting device includes:
Phoneme recognition module, the sequence set of phonemes of the audio input segment for obtaining first language using speech recognition;
First basic coding identification module, for determining sequence phone set using the first basic voice coding sequence of first language
The the first basic voice coding closed;
First continuous speech coding module, for the first voice mapping structure and the first basic voice coding using first language
Sequence determines the continuous speech coding of sequence set of phonemes;
Second basic coding identification module, for obtaining the of second language using the voice primary transformational structure between corresponding language
Two basic voice codings;
Second continuous speech coding module, for utilizing the advanced transformational structure of voice and the second basic voice coder between corresponding language
Code sequence obtains the continuous speech coding of second language;
Continuous programming code conversion module encodes to form sound pronunciation for the continuous speech according to second language.
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910143693.2A CN109754780B (en) | 2018-03-28 | 2018-03-28 | Basic speech coding graphics and audio exchange method |
CN201810264460.3A CN108597493B (en) | 2018-03-28 | 2018-03-28 | The audio exchange method and audio exchange system of language semantic |
PCT/CN2019/079834 WO2019184942A1 (en) | 2018-03-28 | 2019-03-27 | Audio exchanging method and system employing linguistic semantics, and coding graph |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810264460.3A CN108597493B (en) | 2018-03-28 | 2018-03-28 | The audio exchange method and audio exchange system of language semantic |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910143693.2A Division CN109754780B (en) | 2018-03-28 | 2018-03-28 | Basic speech coding graphics and audio exchange method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108597493A CN108597493A (en) | 2018-09-28 |
CN108597493B true CN108597493B (en) | 2019-04-12 |
Family
ID=63624812
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910143693.2A Active CN109754780B (en) | 2018-03-28 | 2018-03-28 | Basic speech coding graphics and audio exchange method |
CN201810264460.3A Active CN108597493B (en) | 2018-03-28 | 2018-03-28 | The audio exchange method and audio exchange system of language semantic |
Family Applications Before (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910143693.2A Active CN109754780B (en) | 2018-03-28 | 2018-03-28 | Basic speech coding graphics and audio exchange method |
Country Status (2)
Country | Link |
---|---|
CN (2) | CN109754780B (en) |
WO (1) | WO2019184942A1 (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109754780B (en) * | 2018-03-28 | 2020-08-04 | 孔繁泽 | Basic speech coding graphics and audio exchange method |
CN110991148B (en) * | 2019-12-03 | 2024-02-09 | 孔繁泽 | Information processing method and device, information interaction method and device |
CN114171013A (en) * | 2021-12-31 | 2022-03-11 | 西安讯飞超脑信息科技有限公司 | Voice recognition method, device, equipment and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102063899A (en) * | 2010-10-27 | 2011-05-18 | 南京邮电大学 | Method for voice conversion under unparallel text condition |
CN104637482A (en) * | 2015-01-19 | 2015-05-20 | 孔繁泽 | Voice recognition method, device, system and language switching system |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8219391B2 (en) * | 2005-02-15 | 2012-07-10 | Raytheon Bbn Technologies Corp. | Speech analyzing system with speech codebook |
US7840399B2 (en) * | 2005-04-07 | 2010-11-23 | Nokia Corporation | Method, device, and computer program product for multi-lingual speech recognition |
US7912716B2 (en) * | 2005-10-06 | 2011-03-22 | Sony Online Entertainment Llc | Generating words and names using N-grams of phonemes |
CN101131689B (en) * | 2006-08-22 | 2010-08-18 | 苗玉水 | Bidirectional mechanical translation method for sentence pattern conversion between Chinese language and foreign language |
KR20080046552A (en) * | 2006-11-22 | 2008-05-27 | 가구모토 주니치 | Print having speech code, method and device for reappearing record, and commerce mode |
WO2012061588A2 (en) * | 2010-11-04 | 2012-05-10 | Legendum Pro Vita, Llc | Methods and systems for transcribing or transliterating to an iconophonological orthography |
US10269353B2 (en) * | 2016-08-30 | 2019-04-23 | Tata Consultancy Services Limited | System and method for transcription of spoken words using multilingual mismatched crowd unfamiliar with a spoken language |
CN109754780B (en) * | 2018-03-28 | 2020-08-04 | 孔繁泽 | Basic speech coding graphics and audio exchange method |
-
2018
- 2018-03-28 CN CN201910143693.2A patent/CN109754780B/en active Active
- 2018-03-28 CN CN201810264460.3A patent/CN108597493B/en active Active
-
2019
- 2019-03-27 WO PCT/CN2019/079834 patent/WO2019184942A1/en active Application Filing
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102063899A (en) * | 2010-10-27 | 2011-05-18 | 南京邮电大学 | Method for voice conversion under unparallel text condition |
CN104637482A (en) * | 2015-01-19 | 2015-05-20 | 孔繁泽 | Voice recognition method, device, system and language switching system |
Also Published As
Publication number | Publication date |
---|---|
CN108597493A (en) | 2018-09-28 |
CN109754780A (en) | 2019-05-14 |
CN109754780B (en) | 2020-08-04 |
WO2019184942A1 (en) | 2019-10-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108447486B (en) | Voice translation method and device | |
CN111276120B (en) | Speech synthesis method, apparatus and computer-readable storage medium | |
EP2958105B1 (en) | Method and apparatus for speech synthesis based on large corpus | |
WO2020215551A1 (en) | Chinese speech synthesizing method, apparatus and device, storage medium | |
US12033612B2 (en) | Speech synthesis method and apparatus, and readable storage medium | |
KR100391243B1 (en) | System and method for generating and using context dependent sub-syllable models to recognize a tonal language | |
US11488577B2 (en) | Training method and apparatus for a speech synthesis model, and storage medium | |
CN108597493B (en) | The audio exchange method and audio exchange system of language semantic | |
KR102625184B1 (en) | Speech synthesis training to create unique speech sounds | |
US6188984B1 (en) | Method and system for syllable parsing | |
EP4029010B1 (en) | Neural text-to-speech synthesis with multi-level context features | |
WO2022134164A1 (en) | Translation method, apparatus and device, and storage medium | |
CN110335608A (en) | Voice print verification method, apparatus, equipment and storage medium | |
KR102639322B1 (en) | Voice synthesis system and method capable of duplicating tone and prosody styles in real time | |
CN114255737B (en) | Voice generation method and device and electronic equipment | |
CN116386594A (en) | Speech synthesis method, speech synthesis device, electronic device, and storage medium | |
CN113450758B (en) | Speech synthesis method, apparatus, device and medium | |
CN115206284B (en) | Model training method, device, server and medium | |
CN105895075B (en) | Improve the method and system of synthesis phonetic-rhythm naturalness | |
CN113160849B (en) | Singing voice synthesizing method, singing voice synthesizing device, electronic equipment and computer readable storage medium | |
CN113488010B (en) | Music data generation method, device, equipment and storage medium | |
CN114373445B (en) | Voice generation method and device, electronic equipment and storage medium | |
CN116469371A (en) | Speech synthesis method and device, electronic equipment and storage medium | |
Sehgal et al. | Speech Processing | |
CN115620702A (en) | Speech synthesis method, speech synthesis device, electronic apparatus, and storage medium |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1261697 Country of ref document: HK |