TWI220206B - System and method for searching a single word in accordance with speech - Google Patents

Abstract

A system and method searching a single word in accordance with speech firstly stores audio data. The audio acquisition module acquires the single word to be searched in audio data. The comparison module searches the pronunciation of the single word in word information database to locate matching word. Further forward the corresponding annotation information of single word to output display module for display so as to implement search of annotation for single word through single word pronunciation.

Description

1220206 V. Description of the invention α) [Technology to which the invention belongs The present invention is a kind of pronunciation through words [prior art] When you want to query the content of a word in foreign language learning, when you encounter an electronic dictionary and words that are easy to understand without knowing the dictionary, It can be very easy to hear that π is an externally important skill. A sentence can be composed of words, if the entire sentence is understood. Get out the word obstacle. When we meet all through the dictionary and do not know the word, the user must know that the user can only substitute the word into the sentence based on the word. Many words are only because there are many word fields.] A word query system and method, especially The system and method for querying corresponding word definitions. As well as the meaning and the word used, the computer's soft and easy language search should be carried out immediately without knowing the pronunciation of the spelling words in the native language used, and the query of all the words. Information is often used as a convenience. When inquiries and listening are used to understand the meaning of one of the foreign languages, if you do not understand the meaning, check the spelling of the word to verify the sound, and it is an irregular sound for communication. For example, it is the case that the words in the foreign language of the electronic word are also the same when the word stream is used, but the code requires a method to spell it. This guessing is difficult to regularize. The flow process is as follows: In reading situation, these currently appear code or computer related information. When communicating, ask for meaning, the meaning of the sentence ’. Also, when you need to be able to be true, usually you only need to enter a single query. . Find out whether the phase test is correct. Therefore, often, foreign language words are often checked. There are a variety of types. The process of entering a single entry. If a non-learner hears all the subordinates, it will affect the solution to timely reflection and communication. The spelling of Dao word is sometimes the correct word, but if it is spelled, it will waste a lot

Page 4 1220206 V. Description of the invention (2) It is still possible to guess and verify the word in time, but the meaning of the corresponding word is still not found in the end. [Summary of the Invention] In view of this, the present invention proposes a system for querying words according to speech and a method thereof in order to solve the above-mentioned problems, and uses speech recognition technology to perform word query, which breaks through the traditional dictionary and can only be searched by inputting word spelling. bottleneck. The present invention discloses a system for querying words according to speech, including a receiving module for receiving audio data and a playback module for playing the received audio data, and a word information database for Store more than one word of information, an audio capture module to capture the sound segments of the words in the audio data, as a selected word, a comparison module that compares the selected word with the sound information in the word information, A comparison result is obtained, and an output display module is provided to display and output the annotation information of the corresponding word according to the comparison result. The invention further discloses a method for querying words according to voice. Firstly, an audio data is received, the audio data is played, and a sound segment of a word is selected as a selected word, and then a word matching the selected word in the word information database is searched. Voice information, and then output corresponding word annotation information based on the comparison result. According to the disclosed system and method for querying words in accordance with speech, the bottleneck of traditional word querying can only be performed by one-way spelling of words, especially for listening practice. Solve the new words that appear in listening, use comparative

Page 5 1220206 V. Description of the invention (3) In a short time, great learning results have been obtained. The detailed content and technology of the present invention are described below with reference to the drawings: [Embodiment] The present invention will be described in detail below with reference to the drawings. According to the system and method for querying words according to speech disclosed by the present invention, the present invention will be described in detail by taking a listening practice of a learner as an example. Please refer to "Figure 1". "Figure 1" illustrates the system of the present invention. The figure is a module architecture diagram of the system and method for querying words according to speech according to the present invention, including: a receiving module 11 0, used to receive audio data. Learners can download audio data through 1 network, and can also record listening practice content. At the same time, the receiving module 11 0 also has a storage function, which can receive the data. Audio data is stored in chronological order. The playback module 1 2 0 is responsible for playing the received audio data. During playback, the playback speed can also be adjusted. In order to accurately retrieve the words to be queried, the playback speed can be slowed down. The audio capture module 130 is mainly responsible for capturing the audio clips in the audio data. The pronunciation of a word is selected as the selected word in the audio data being played. The audio capture button can be used to select the audio data being played. Retrieve the selected words for the desired query. The word information database 1 40 is mainly used to store word information, and the word information includes sound information and corresponding words such as spelling, interpretation, phonetic transcription, part of speech, usage, and other annotation information. There is also a comparison table stored,

Page 6 1220206 V. Description of the Invention (4)-The correspondence table describes the correspondence between the sound information of the words and the annotation information. The comparison module 1 50 is responsible for comparing the captured selected word with the sound information in the word information database to obtain the comparison result. If the matching module finds a matching sound information, it checks the The corresponding relationship finds the corresponding word annotation information from the word information database 1 40; if the matching sound information is not found after comparison, it is concluded that there is no such word. The output display module 16 0 is mainly responsible for outputting the comparison result in the comparison module 150. If one sound information corresponds to multiple word annotation information, all annotation information is displayed. ¥ Please refer to "Figure 2", which is a general flowchart of the method for querying words by voice according to the present invention. First receive an audio data (step 2 1 0), play the audio data, and simultaneously capture a word sound segment as the selected word (step 2 2 0), and then search the word information database for words that match the selected word Audio information (step 2 3 0), finally 4 according to the comparison result, output the corresponding word annotation information (step 2 4 0). In the following, the operation flow of the present invention will be described in detail with "Figure 3", which is an operation flow chart of the embodiment of the system and method for word query according to speech provided by the present invention. First, word information is stored in the word information database. The word information includes word sound information and word comment information. The word comment information includes word spelling, interpretation, phonetic transcription, part of speech, usage, and other information. In addition, a comparison table is stored in the word information database. The comparison table associates sound information with annotation information. An audio message in the lookup table may correspond to one or more annotation messages. Learners practice listening

Page 7 1220206 5. In the description of the invention (5), a piece of audio data must be received as the listening practice content (step 310). The method of obtaining can be downloaded from the Internet or recorded by recording equipment. Store by time. Then the recorded audio data is played (step 3 2 0). During playback, the user can adjust the playback speed according to the needs. The learner can slow down the playback speed in order to facilitate the extraction of the word to be queried. Then start the information retrieval module to retrieve the required selected words (step 3 3 0), that is, to extract the pronunciation of the word, and then search the word information database for sound information that matches the selected word (step 3 4 0), when performing comparisons, only similar comparisons are performed, and complete comparisons are not required, because not every language has unique pronunciation rules, and sometimes due to regional differences or original data The speaker's pronunciation causes the pronunciation of the word to be different from the sound information stored in the word information database, so only an approximate comparison is made here. The approximate comparison is based on a table of similar phonetic transcription rules. Taking English as an example, please refer to the table below for a list of similar phonetic rules: Similar vowels P Similar consonants ρ Similar vowels and consonants # i: i ^ p be ij ^ eei ae ^ tdpuw ^ ο: ο ^ fvw ^ B 3 : 1 rp 3: 3 · P sf ^ α: λ ai # Θ 3 ^ υ: υ ο υ ^ z 3 ο j υ: j υ ^ kh ^ Q Π 4 s Hong j 1 Γ 4

Page 8 丄 "U2〇6

(6) V. Description of the invention According to the above table, the use of speech recognition technology to approximate similar pronunciations is used to search for voice information. Then determine whether there is a corresponding sound ^ 2 = (step 3 5 0), if there is corresponding sound information, then through the matching information 1 to the corresponding annotation information (step 3 6 〇), and then the found annotations like library display Come out (step 3 7 0). If the pronunciation has the same or similar pronunciation in the word information data f, the word is directly displayed, and σ is related to the word, part of speech, phonetic symbol, example sentence, definition and other related information. If there are multiple pronunciations in the pronunciation database that are the same as or similar to the pronunciation, then the words corresponding to multiple pronunciations are displayed, and the word, word。:! Standard, example sentence, definition, and other related information tfl can be obtained. If no corresponding information is found, the word does not exist (step 38). Next, please refer to "Fig. 4a" to "Fig. 4b", which are schematic diagrams of the interface of the preferred embodiment of the system and method for querying words according to speech according to the present invention. "Picture 4a" is the pronunciation interface. Learners record the pronunciation through the recording button. The disagreement among the family members healed over with time. Extracted from the sentence, $ is an accurate command to send and send. You can slow down the playback speed by adjusting the speed alone. Retrieve the pronunciation by the Extract pronunciation button. After the pronunciation is finished, 'press the query button to find the same or Ϊί =;,: If you find the same or similar pronunciation words, the "" page shows early spelling." As shown in Figure 4b ". Three early words with similar pronunciation are displayed in the page. You can find the related word information by directly clicking the word.

1220206 V. Description of the invention (7). For example, if you click the word n h e a 1 ", the interface as shown in "Figure 4c" will be displayed. Learners can judge the words corresponding to the pronunciation according to the interpretation of the words. If no corresponding word is found, the system will display a prompt box as shown in "Figure 4d", prompting the learner whether to retrieve the pronunciation again. If the learner selects "n Yes", the system will return to "4a "Picture" capture pronunciation interface. If the learner chooses πNon, the pronunciation query ends and the program exits. Although the present invention is disclosed in the foregoing preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art of similarity can make some modifications and retouching without departing from the spirit and scope of the present invention. The patent protection scope of the invention shall be determined by the scope of the patent application scope attached to this specification.

1220206 on page 10 for a brief explanation. Figure 1 is a block diagram of the system for querying words by speech according to the present invention. Figure 2 is a general flowchart of the method for querying words by speech according to the present invention. FIG. 3 is a flowchart of an embodiment of a system and method for performing a word query according to speech according to the present invention; and FIGS. 4a to 4d are schematic diagrams of an embodiment of a system and method for performing a word query according to speech according to the present invention. . [Illustration of graphical symbols] 110 receiving module 12 0 playing module 130 audio capturing module 140 word information database 150 comparison module 16 0 output display module step 2 1 0 receiving an audio data step 2 2 0 playing The audio data is simultaneously extracted with a word sound segment as the selected word. Step 2 3 0 searches the word information database for the word sound information that matches the selected word. Step 2 4 0 outputs the corresponding word annotation information based on the comparison result. Step 3 1 0 Receive and store audio data Step 3 2 0 Play audio data Step 3 3 0 Retrieve the selected word you need

Page 1212206 Schematic description of step 3 4 0 Compare the corresponding sound information in the word information database Step 3 5 0 Whether there is a corresponding sound information Step 3 6 0 Look up the table to find the comment information corresponding to the selected word Step 3 7 0 Display note information Step 3 8 0 Display word does not exist

Page 12

Claims (1)

1220206 6. Scope of patent application 1. A system for querying words according to voice includes: a receiving module for receiving an audio data; a playback module for playing the received audio data; a An audio capture module is used to capture the word sound segment in the audio data as the selected word during playback; a word information database is used to store more than one word of information; a comparison module that The selected word is compared with the sound information in the word information and a comparison result is obtained; and an output display module displays and outputs the comment information of the corresponding word according to the comparison result. 2. The system for word search according to speech as described in item 1 of the scope of patent application, wherein the receiving module further includes a storage function for storing the audio lean material. 3. The system for querying words by voice as described in item 1 of the scope of patent application, wherein the audio playback module can adjust the playback speed. 4. The system for querying words by voice as described in item 1 of the scope of patent application, wherein the word information includes sound information and corresponding annotation information. 5. The system for querying words by voice as described in item 1 of the scope of patent application, wherein the annotation information can be selected from any combination of word spelling, definition, phonetic transcription, part of speech, usage. 6. Word search by voice as described in item 1 of the scope of patent application Page 13 1220206 6. A patent-applicable system, wherein the word information database further stores a correspondence table of single-word sound information and annotation information. 7. A method for querying words according to speech, comprising the steps of: receiving an audio tribute, playing the audio data, and simultaneously capturing a word sound segment as a selected word; searching a word information database for a word corresponding to the selected word Match word sound information; and output corresponding early word annotation education based on the comparison result ^. 8. The method for querying words by voice as described in item 7 of the scope of patent application, wherein the step of receiving an audio data is to receive and store an audio data. 9. The method for querying words by speech as described in item 7 of the scope of patent application, wherein the search comparison is an approximate comparison based on similar phonetic symbols. 10. The method for querying words by voice as described in item 7 of the scope of patent application, wherein the step of outputting corresponding word annotation information according to the comparison result, if there is no matching word information, prompt the word does not exist. 11 1. The method for querying words by voice as described in item 7 of the scope of patent application, wherein the annotation information may be selected from any combination of word spelling, interpretation, phonetic transcription, part of speech, and usage. Page 14