CN103455642B - A kind of method and apparatus of multimedia document retrieval - Google Patents

Abstract

The present application discloses a method for retrieving multimedia files. The method includes: when receiving a voice command, separating the action keyword and description keyword in the voice command; using the separated description keyword to match the stored embedded description If the multimedia file of the voice data is matched to the multimedia file, according to the separated action keyword, the corresponding action is executed on the multimedia file determined to be retrieved. Based on the same inventive concept, this application also proposes a device that can simply retrieve multimedia files through voice, bringing friendly user physical examination.

Description

Method and device for multimedia file retrieval

technical field

The present application relates to the technical field of voice processing, in particular to a method and device for multimedia file retrieval.

Background technique

At present, multimedia recording modules, such as cameras and microphones, have become the standard configuration of personal daily electronic equipment, such as cameras, mobile phones, tablet computers and so on. People are increasingly using these devices to record life, and also save a large number of multimedia files such as photos, videos, and audios.

These files are generally named with digital serial numbers or shooting time, but it is precisely this serial number naming that makes file retrieval and management quite inconvenient.

For example, if a user wants to retrieve certain photos, he usually needs to browse through a large number of photos and select the desired ones, especially on terminals without general input devices such as mouse and keyboard. For example, when retrieving photos on TV, he can only use simple It is quite inconvenient to use the remote control to browse and retrieve photos.

In existing implementations, some technical solutions have been used to solve the problem of multimedia file retrieval, basically by establishing the mapping relationship between keywords and multimedia files, and storing these mapping relationships in the database in the form of independent file storage or data .

The biggest disadvantage of these solutions is that the relationship between the retrieved files and the mapping relationship files is relatively loose, and a large number of mapping files will be generated. For the user, the files are more messy; Update the mapping file or database; at the same time, when transferring the retrieved file to another storage device or display device, it is necessary to transfer the mapping file or database file at the same time.

Contents of the invention

In view of this, the present application provides a method and device for retrieving multimedia files, which can simply perform retrieval of multimedia files through voice, and bring friendly user physical examination.

In order to solve the problems of the technologies described above, the technical solution of the present invention is achieved in the following way:

A method for multimedia file retrieval, the method comprising:

Storing the multimedia file, wherein, when the multimedia file is captured, the descriptive voice data for the multimedia file is embedded;

When receiving a voice instruction for retrieving a multimedia file, identify and separate the action keyword in the voice instruction and the description keyword of the multimedia file to be retrieved;

Match the stored multimedia files according to the separated description keywords; wherein, when matching each multimedia file, identify the description keywords of the multimedia file by embedding the description voice data of the multimedia file; if the separated description If the keyword matches the identified description keyword, it is determined that the multimedia file is a matching multimedia file;

According to the separated action keyword, a corresponding action is performed on the matched multimedia file.

A device comprising: a storage unit, a receiving unit, an identification unit, a matching unit and a processing unit;

The storage unit is used to store a multimedia file, wherein when the multimedia file is captured, the description voice data for the multimedia file is embedded;

The receiving unit is configured to receive voice instructions for retrieving multimedia files;

The identification unit is used to identify and separate the action keyword in the voice instruction and the description keyword of the multimedia file to be retrieved when the receiving unit receives a voice instruction for retrieving multimedia files;

The matching unit is configured to match the multimedia files stored in the storage unit according to the description keywords separated by the recognition unit; wherein, when matching each multimedia file, by embedding the description voice data of the multimedia file, Identify the description keyword of the multimedia file; if the separated description keyword matches the identified description keyword, then determine that the multimedia file is a matching multimedia file;

The processing unit is configured to perform corresponding actions on the multimedia files matched by the matching unit according to the action keywords separated by the recognition unit.

To sum up, this application separates the action keyword and description keyword in the voice instruction when receiving the voice instruction; uses the separated description keyword to match the stored multimedia file embedded in the description voice data, if it matches According to the separated action keyword, the corresponding action is executed on the multimedia file determined to be retrieved. In the case that there is no need to save and maintain the mapping relationship between keywords and multimedia files, multimedia files can be retrieved through speech recognition.

Description of drawings

FIG. 1 is a schematic flow diagram of a method for retrieving multimedia files in Embodiment 1 of the present invention;

2 is a schematic flow diagram of a method for retrieving multimedia files in Embodiment 2 of the present invention;

3 is a schematic flow diagram of a method for retrieving multimedia files in Embodiment 3 of the present invention;

4 is a schematic flow diagram of a method for retrieving multimedia files in Embodiment 4 of the present invention;

5 is a schematic flow chart of a method for determining copyright of a multimedia file in Embodiment 5 of the present invention;

FIG. 6 is a schematic structural diagram of a device applied to the above technology.

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the solutions of the present invention will be further described in detail below with reference to the accompanying drawings and examples.

In the embodiment of the present invention, a method for multimedia file retrieval is proposed. When a voice command is received, the action keywords and description keywords in the voice command are separated; the separated description keywords are used to match the stored embedded description voice data If the multimedia file is matched to the multimedia file, according to the separated action keyword, a corresponding action is performed on the multimedia file determined to be retrieved. It can simply retrieve multimedia files by voice, bringing a friendly user experience.

In a specific embodiment of the present invention, the device for capturing multimedia files and the device for retrieving multimedia files may be the same device or different devices. If it is not the same device, the multimedia file embedded with voice data needs to be transferred to the device for retrieving the multimedia file.

The device for retrieving the multimedia file stores the multimedia file, wherein when the multimedia file is captured, descriptive voice data for the multimedia file is embedded.

If the device has the function of acquiring multimedia files, the device will directly store the acquired multimedia files; if the device does not have the function of capturing multimedia files, transfer the multimedia files on the device that captures the multimedia files to this device .

Wherein, the descriptive voice data for the multimedia file may be the captured speaker's descriptive voice data for the multimedia file.

For example, when taking a picture, directly capture the speaker's description voice for the picture, and embed it into the picture, and describe the content of the voice data, such as: This is a picture taken in Shanghai. A more detailed description, such as a specific location, can also be provided.

The descriptive voice data for the multimedia file may also be preset default voice data.

The preset default voice is a pre-captured descriptive voice, and it is assumed that the content of the descriptive voice data is also: this is a picture taken in Shanghai. In this way, all the pictures taken in Shanghai can be embedded with this preset default voice data, and it is not necessary to capture and describe the voice data every time.

When the multimedia file is captured, the descriptive voice data for the multimedia file is embedded, including: the descriptive voice data for the multimedia file is embedded in the form of extended data, metadata, digital watermark, or other forms that retain the original format of the voice data to the captured multimedia file.

Among them, embedding in the form of extended data means embedding the description voice data as the extension header of the multimedia file; embedding in the form of metadata means embedding the description voice data as the auxiliary data of the multimedia file; embedding in the form of digital watermark , about to embed the voice data into multimedia files in the form of digital watermark.

The method of describing speech embedding data in the specific embodiment of the present invention does not need to independently save the description keywords and the mapping relationship between the multimedia file and the corresponding description keywords as in the existing implementation. When operations such as dumping or modifying a name of a multimedia file are performed, there is no need to maintain the mapping relationship between the multimedia file and its corresponding description keywords. Retrieval of multimedia files is not affected.

Embodiment one

Referring to FIG. 1 , FIG. 1 is a schematic flowchart of a method for retrieving multimedia files in Embodiment 1 of the present invention. The specific steps are:

Step 101, when the device receives a voice instruction for retrieving multimedia files, it recognizes and separates action keywords in the voice instruction and description keywords of the multimedia files to be retrieved.

When the device receives a voice command, it recognizes the voice command through the voice recognition engine or program of the device, or through a voice recognition engine or program provided by other devices, and separates the recognized keywords into description keywords and action keywords.

Assuming that the content of the received voice instruction is "display the photos of Shanghai", the separated action keyword is: "display"; the description keyword is: "Shanghai".

As for how to identify and separate the keywords in the voice instruction, it can be realized by the existing voice recognition technology, and no specific implementation method will be given here.

Step 102, the device matches the stored multimedia files according to the separated description keywords.

Wherein, when the device matches each multimedia file, it recognizes the description keyword of the multimedia file by embedding the description voice data of the multimedia file; if the separated description keyword matches the identified description keyword, then determine The multimedia file is a matching multimedia file; if the separated description keyword does not match the identified description keyword, it is determined that the multimedia file is not a matching multimedia file.

If the description keyword recognized by the description speech embedded in a certain multimedia file is "Shanghai", then it is determined that the multimedia file is a matching multimedia file; if the description keyword recognized by the description speech embedded in a certain multimedia file is "Beijing", then it is determined that the multimedia file is an unmatched multimedia file.

Step 103, the device executes corresponding actions on the matched multimedia files according to the separated action keywords.

As an example above, when the separated action keyword is "display", the corresponding action to be executed is to display the matched multimedia file.

Embodiment two

Referring to FIG. 2 , FIG. 2 is a schematic flowchart of a method for retrieving multimedia files in Embodiment 2 of the present invention. The specific steps are:

In step 201, when the device receives a voice instruction for retrieving multimedia files, it recognizes and separates action keywords in the voice instruction from description keywords of the multimedia files to be retrieved.

In step 202, the device determines whether the separated description keyword matches a preset description keyword, and if yes, executes step 203; otherwise, executes step 205.

If the content of the received voice command is "show the photos I took", the recognized and separated action keyword is "display"; the description keyword is "the photos I took", of course, the description keywords in the specific implementation It can also be "I", etc., to implement specific settings.

At this time, the separated descriptive keyword is associated with a specific person, and step 203 needs to be executed to match using the voice feature of the speaker.

The preset description keywords are description keywords related to the speaker himself, such as: mine, photos I took, etc.

Step 203, the device performs voice feature recognition on the received voice command, and obtains the voice feature of the speaker of the voice command.

The specific implementation of this step can be based on biometric technology to extract the voice features of the speaker, which will not be described in detail here.

In step 204, the device uses the acquired voice characteristics of the speaker to match the stored multimedia files, and then executes step 206.

When the device matches each multimedia file, it recognizes the voice feature of the description voice data embedded in the multimedia file; if the voice feature of the obtained speaker matches the voice feature of the recognized voice data, then it is determined that the multimedia file is matched. A multimedia file; if the acquired voice feature of the sender does not match the recognized voice feature, then it is determined that the multimedia file is not a matching multimedia file.

Step 205, the device matches the stored multimedia files according to the separated description keywords.

Wherein, when the device matches each multimedia file, it recognizes the description keyword of the multimedia file by embedding the description voice data of the multimedia file; if the separated description keyword matches the identified description keyword, then determine The multimedia file is a matching multimedia file; if the separated description keyword does not match the identified description keyword, it is determined that the multimedia file is not a matching multimedia file.

In step 206, the device executes corresponding actions on the matched multimedia files according to the separated action keywords.

In this embodiment, when the separated description keyword matches the configured description keyword, that is, when the separated description keyword is associated with the speaker itself, the multimedia file is matched according to the voice feature of the received voice command.

Embodiment three

Referring to FIG. 3 , FIG. 3 is a schematic flowchart of a method for retrieving multimedia files in Embodiment 3 of the present invention. The specific steps are:

Step 301, the device receives a voice command for retrieving multimedia files, recognizes and separates the action keywords in the voice command and the description keywords of the multimedia files to be retrieved; and performs voice feature recognition on the received voice command to obtain the voice The speech characteristics of the speaker of the command.

In this embodiment, when a voice command is received, the voice feature recognition can be directly performed on the received voice command to obtain the voice feature of the speaker of the voice command, or the voice feature can be obtained when voice feature recognition is used. , that is, to recognize speech features before performing step 303 .

Step 302, the device matches the stored multimedia files according to the separated description keywords.

Wherein, wherein, when the device matches each multimedia file, it identifies the description keyword of the multimedia file by embedding the description voice data of the multimedia file; if the separated description keyword matches the identified description keyword, Then it is determined that the multimedia file is a matching multimedia file; if the separated description keyword does not match the identified description keyword, it is determined that the multimedia file is not a matching multimedia file. .

In step 303, the device uses the acquired voice features of the speaker to match the multimedia files matched by the separated description keywords again.

After the retrieval using the separated descriptive keywords, some multimedia files matching the descriptive keywords are obtained. In this step, the acquired voice features of the speaker are used to perform secondary matching in the matched multimedia files.

Wherein, when the device matches each multimedia file, it recognizes the voice feature of the description voice data embedded in the multimedia file; if the voice feature of the obtained speaker matches the voice feature of the recognized voice data, then it is determined that the multimedia file is A matching multimedia file; otherwise, it is determined that the multimedia file corresponding to the voice feature of the recognized voice data is not a matching multimedia file.

In step 304, the device performs a corresponding action on the multimedia file matched through the voice feature according to the separated action keyword.

In this embodiment, the retrieval is performed by combining the description keyword and the speech feature, which is equivalent to displaying the multimedia file after the intersection of the two factors retrieval, which can improve the retrieval accuracy. When this embodiment is actually implemented, it is first searched through description keywords, and then searched through speech features.

Embodiment four

Referring to FIG. 4 , FIG. 4 is a schematic flowchart of a method for retrieving multimedia files in Embodiment 4 of the present invention. The specific steps are:

Step 401, when the device receives a voice command for retrieving multimedia files, it recognizes and separates the action keywords in the voice command and the description keywords of the multimedia files to be retrieved; and performs voice feature recognition on the received voice commands to obtain the Voice characteristics of the speaker of the voice command.

Step 402, the device matches the stored multimedia files by using the acquired voice characteristics of the speaker.

Wherein, when the device matches each multimedia file, it recognizes the voice feature of the description voice data embedded in the multimedia file; if the voice feature of the obtained speaker matches the voice feature of the recognized voice data, then it is determined that the multimedia file is A matching multimedia file; otherwise, it is determined that the multimedia file corresponding to the voice feature of the recognized voice data is not a matching multimedia file.

In step 403, the device performs matching again on the multimedia file matched by the acquired voice features of the speaker according to the separated description keywords.

In step 404, the device executes corresponding actions on the multimedia files matched by the separated keywords according to the separated action keywords.

In this embodiment, the search is also performed by combining the description keyword and the speech feature, and finally what is displayed is the multimedia file after the intersection of the two factors, which can improve the accuracy of the search. When this embodiment is actually implemented, it is first searched through speech features, and then searched through description keywords.

Embodiment five

Referring to FIG. 5 , FIG. 5 is a schematic flowchart of a method for determining the copyright of a multimedia file in Embodiment 5 of the present invention. The specific steps are:

Step 501, when the device receives the voice data of the speaker, it performs voice feature recognition to obtain the voice feature of the speaker.

Step 502, the device performs voice feature recognition on the descriptive voice data embedded in the multimedia file whose copyright needs to be determined, and obtains the voice feature of the descriptive voice data.

In step 503, the device determines that the obtained voice feature of the speaker matches the obtained voice feature describing the voice data.

Step 504, if the obtained voice feature of the speaker matches the obtained voice feature describing the voice data, then determine that the speaker is the copyright owner of the multimedia file; otherwise, determine that the speaker is not the copyright owner of the multimedia file By.

Copyright protection of multimedia files using biological consistency.

Based on the same inventive concept, the present application also proposes a device. Referring to FIG. 6 , FIG. 6 is a schematic structural diagram of a device applied to the above technology. The device includes: a storage unit 601 , a receiving unit 602 , an identification unit 603 , a matching unit 604 and a processing unit 605 .

The storage unit 601 is configured to store a multimedia file, wherein when the multimedia file is captured, description voice data for the multimedia file is embedded.

The receiving unit 602 is configured to receive voice instructions for retrieving multimedia files.

The recognition unit 603 is configured to, when the receiving unit 602 receives a voice command for retrieving a multimedia file, identify and separate the action keyword in the voice command and the description keyword of the multimedia file to be retrieved.

The matching unit 604 is configured to match the multimedia files stored in the storage unit 601 according to the descriptive keywords separated by the identifying unit 603; wherein, when matching each multimedia file, identify the multimedia file by embedding the description voice data of the multimedia file A description keyword of the multimedia file; if the separated description keyword matches the identified description keyword, it is determined that the multimedia file is a matched multimedia file.

The processing unit 604 is configured to perform corresponding actions on the multimedia files matched by the matching unit 604 according to the action keywords separated by the recognition unit 603 .

Preferably, the descriptive voice data for the multimedia file is: preset default voice data, or the captured speaker's descriptive voice data for the multimedia file.

Preferably,

The storage unit 601 , when the stored multimedia file is captured, embeds the description voice data for the multimedia file into the captured multimedia file in the form of extended data, metadata, digital watermark, or retaining the original format of the data.

Preferably,

The processing unit 604 is further configured to determine whether the keyword separated by the recognition unit 603 matches a preset keyword.

The recognition unit 603 is further configured to, if the processing unit 604 determines that the separated description keyword matches the preset description keyword, perform voice feature recognition on the received voice command, and obtain the voice feature of the speaker of the voice command.

The matching unit 604 is further used to match the stored multimedia files using the voice features of the speaker obtained by the recognition unit 603; wherein, when matching each multimedia file, identify the voice features embedded in the multimedia file to describe the voice data; If the voice feature of the obtained speaker matches the recognized voice feature, then it is determined that the multimedia file is a matched multimedia file; and the processing unit 604 is triggered to perform corresponding actions on the matched multimedia file according to the separated action keywords ; When the processing unit 604 determines that the description keywords separated by the recognition unit 603 do not match the preset description keywords, then match the stored multimedia files according to the separated description keywords.

Preferably,

The recognition unit 603 is further configured to perform voice feature recognition on the received voice command, and obtain the voice feature of the speaker of the voice command.

The matching unit 604 is further configured to, after matching the stored multimedia files according to the separated description keywords, use the obtained voice characteristics of the speaker to further match the multimedia files matched by the description keywords; wherein, the matching When each multimedia file is identified, the voice feature describing the voice data embedded in the multimedia file is identified; if the voice feature of the obtained speaker matches the voice feature of the recognized voice data, the multimedia file is determined to be a matching multimedia file.

The processing unit 604 is further configured to perform corresponding actions on the multimedia files matched by the matching unit 604 through voice features according to the separated action keywords.

Preferably,

The recognition unit 603 is further configured to perform voice feature recognition on the received voice command, and obtain the voice feature of the speaker of the voice command.

The matching unit 604 is further configured to use the obtained voice features of the speaker to match the stored multimedia files; wherein, when matching each multimedia file, identify the voice features embedded in the multimedia file to describe the voice data; if obtained If the voice feature of the speaker matches the voice feature of the recognized voice data, it is determined that the multimedia file is a matched multimedia file; and according to the separated keywords, matching is performed in the multimedia file matched by the voice feature.

The processing unit 604 is further configured to perform a corresponding action on the multimedia file matched by the matching unit 604 through the description keyword according to the separated action keyword.

Preferably,

The receiving unit 602 is further configured to receive the voice data of the speaker.

Recognition unit 603 is further used for receiving unit 602 to perform speech feature recognition when receiving the voice data of the speaker to obtain the voice feature of the speaker; to perform voice feature recognition on the descriptive voice data embedded in the multimedia file that needs to determine the copyright, Obtain speech features describing the speech data.

The matching unit 604 is further configured to determine that the obtained voice features of the speaker are matched with the obtained voice features describing the voice data.

The processing unit 604 is further configured to determine that the speaker is the copyright owner of the multimedia file if the matching unit 604 determines that the obtained voice feature of the speaker matches the obtained voice feature describing the voice data; otherwise, determine that the speaker is not the copyright owner of this multimedia file.

In summary, in specific embodiments of the present invention, when a voice command is received, the action keywords and description keywords in the voice command are separated; the separated description keywords are used to match the stored multimedia files embedded in the description voice data , if a multimedia file is matched, perform a corresponding action on the multimedia file determined to be retrieved according to the separated action keywords. Without the need to save and maintain the mapping relationship between keywords and multimedia files, multimedia files can be retrieved simply by voice, bringing friendly user checkups.

In the specific embodiment, it also provides the embodiment of searching through the voice features of the speaker, and combining the voice features and description keywords, so that the multimedia files to be retrieved can be retrieved more conveniently and accurately.

At the same time, it also provides an embodiment of determining the copyright owner of the multimedia file through the recognition of the voice feature, which can simply and effectively protect the copyright of the multimedia file.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (14)

1. a kind of method of multimedia document retrieval, it is characterised in that methods described includes：

Storing multimedia, wherein, the multimedia file embedded in the description language for the multimedia file in capture Sound data；

Receive retrieving multimedia file phonetic order when, recognize and separate the phonetic order in action keyword and needs The description keyword of the multimedia file of retrieval；

According to the description keyword that isolates, the multimedia file to storing mates；Wherein, mate each multimedia file When, by the description speech data of the multimedia file is embedded in, identify the description keyword of the multimedia file；If isolating The description keyword and description Keywords matching for identifying, it is determined that the multimedia file is the multimedia file of coupling；Its In, when the multimedia file to storing mates, the phonetic feature according to sounder is mated；

According to the action keyword that isolates, the multimedia file to matching executes corresponding action.

2. method according to claim 1, it is characterised in that

The description speech data for the multimedia file is：Default default speech data, or, the sounder pin of capture Description speech data to the multimedia file.

3. method according to claim 1, it is characterised in that the multimedia file is embedded in for should in capture The description speech data of multimedia file, including：By the description speech data for the multimedia file, with growth data, unit Data, digital watermarking, or the form of retention data unprocessed form, are embedded in the multimedia file for capturing.

4. the method according to claim 1-3 any one, it is characterised in that the action in the separation phonetic order Keyword and the description keyword of the multimedia file for needing to retrieve, afterwards, the description keyword that the basis is isolated, to depositing The multimedia file of storage is mated, and before, methods described is further included：

Determine the description keyword isolated whether with default description Keywords matching, if it is, to the phonetic order for receiving Phonetic feature identification is carried out, obtains the phonetic feature of the sounder of the phonetic order；Using the phonetic feature of the sounder for obtaining, Multimedia file to storing mates；Wherein, when mating each multimedia file, the embedded multimedia file is identified The phonetic feature of description speech data；If the phonetic feature of the sounder for obtaining is mated with the phonetic feature for identifying, it is determined that The multimedia file is the multimedia file of coupling；And according to the action keyword that isolates, to the multimedia file for matching Execute corresponding action；

Otherwise, execute the description keyword that the basis is isolated, the step of the multimedia file to storing mates and after Continuous step.

5. the method according to claim 1-3 any one, it is characterised in that the action in the separation phonetic order During the description keyword of keyword and the multimedia file for needing to retrieve, methods described is further included：To the voice for receiving Instruction carries out phonetic feature identification, obtains the phonetic feature of the sounder of the phonetic order；

The description keyword that the basis is isolated, the multimedia file to storing mate, and afterwards, the basis is isolated Action keyword, corresponding action is executed to the multimedia file that matches, before, methods described is further included：

Using obtain sounder phonetic feature, to by description Keywords matching to multimedia file carry out further Join；Wherein, the phonetic feature of the description speech data of the embedded multimedia file, when mating each multimedia file, is identified； If the phonetic feature of sounder for obtaining is mated with the phonetic feature of the speech data for identifying, it is determined that the multimedia file is The multimedia file of coupling；

The action keyword that the basis is isolated, the multimedia file to matching execute corresponding action, including：According to point The action keyword for separating out, the multimedia file to being matched by phonetic feature execute corresponding action.

6. the method according to claim 1-3 any one, it is characterised in that the action in the separation phonetic order During the description keyword of keyword and the multimedia file for needing to retrieve, methods described is further included：To the voice for receiving Instruction carries out phonetic feature identification, obtains the phonetic feature of the sounder of the phonetic order；

The description keyword that the basis is isolated, the multimedia file to storing mate, and before, methods described is further Including：Using the phonetic feature of the sounder for obtaining, the multimedia file to storing mates；Wherein, mate each many matchmaker During body file, the phonetic feature of the description speech data of the embedded multimedia file is identified；If the voice of the sounder for obtaining Feature is mated with the phonetic feature of the speech data for identifying, it is determined that the multimedia file is the multimedia file of coupling；

Wherein, according to the description keyword that isolates, the multimedia file to storing mates, including：According to isolated Keyword, is mated in the multimedia file matched by phonetic feature.

7. the method according to claim 1-3 any one, it is characterised in that methods described is further included：

Receive sounder speech data when, carry out phonetic feature identification, obtain the phonetic feature of the sounder；

To it needs to be determined that the description speech data being embedded in the multimedia file of copyright carries out phonetic feature identification, acquisition description language The phonetic feature of sound data；

The phonetic feature for determining the sounder for obtaining is mated with the phonetic feature of the description speech data for obtaining；

If the phonetic feature of the sounder for obtaining is mated with the phonetic feature of the description speech data for obtaining, it is determined that the sounder Copyright owner for the multimedia file；Otherwise, it determines sounder copyright owner not for the multimedia file.

8. a kind of device, it is characterised in that the device includes：Memory cell, receiving unit, recognition unit, matching unit and place Reason unit；

The memory cell, for storing multimedia, wherein, the multimedia file is embedded in be directed to and is somebody's turn to do in capture The description speech data of multimedia file；

The receiving unit, for receiving the phonetic order of retrieving multimedia file；

The recognition unit, for receive when the receiving unit retrieving multimedia file phonetic order when, recognize and point Action keyword in the phonetic order and the description keyword of the multimedia file for needing to retrieve；

The matching unit, for the description keyword that is isolated according to the recognition unit, to memory cell storage Multimedia file is mated；Wherein, when mating each multimedia file, by being embedded in the description voice number of the multimedia file According to identifying the description keyword of the multimedia file；If the description keyword that isolates and the description keyword for identifying Join, it is determined that the multimedia file is the multimedia file of coupling；Wherein, when the multimedia file to storing mates, Phonetic feature according to sounder is mated；

The processing unit, for the action keyword that is isolated according to the recognition unit, matches to the matching unit Multimedia file execute corresponding action.

9. device according to claim 8, it is characterised in that the description speech data for the multimedia file For：Default default speech data, or, description speech data of the sounder of capture for the multimedia file.

10. device according to claim 8, it is characterised in that

The memory cell, the multimedia file of storage will be directed to the description speech data of the multimedia file in capture, with Growth data, metadata, digital watermarking, or the form of retention data unprocessed form, are embedded in the multimedia file for capturing.

11. devices according to claim 8-10 any one, it is characterised in that

The processing unit, be further used for determining keyword that the recognition unit isolates whether with predetermined keyword Join；

The recognition unit, if be further used for the processing unit to determine that the description keyword that isolates is crucial with default description Word mates, and the phonetic order to receiving carries out phonetic feature identification, obtains the phonetic feature of the sounder of the phonetic order；

The matching unit, is further used for the phonetic feature of sounder obtained using the recognition unit, to many of storage Media file is mated；Wherein, the description voice number of the embedded multimedia file, when mating each multimedia file, is identified According to phonetic feature；If the phonetic feature of the sounder for obtaining is mated with the phonetic feature for identifying, it is determined that the multimedia text Part is the multimedia file of coupling；And the processing unit is triggered according to the action keyword that isolates, to many matchmakers for matching Body file executes corresponding action；When the processing unit determines that the description keyword that the recognition unit is isolated is retouched with default Crucial word mismatch is stated, then according to the description keyword that isolates, the multimedia file to storing mates.

12. devices according to claim 8-10 any one, it is characterised in that

The recognition unit, being further used for the phonetic order to receiving carries out phonetic feature identification, obtains the phonetic order Sounder phonetic feature；

The matching unit, is further used for according to the description keyword that isolates, and the multimedia file to storing mates Afterwards, using obtain sounder phonetic feature, to by description Keywords matching to multimedia file carry out further Coupling；Wherein, when mating each multimedia file, identify that the voice of the description speech data of the embedded multimedia file is special Levy；If the phonetic feature of the sounder for obtaining is mated with the phonetic feature of the speech data for identifying, it is determined that the multimedia text Part is the multimedia file of coupling；

The processing unit, is further used for, according to the action keyword that isolates, passing through phonetic feature to the matching unit The multimedia file for matching executes corresponding action.

13. devices according to claim 8-10 any one, it is characterised in that

The recognition unit, being further used for the phonetic order to receiving carries out phonetic feature identification, obtains the phonetic order Sounder phonetic feature；

The matching unit, the phonetic feature being further used for using the sounder for obtaining, the multimedia file to storing are carried out After coupling；Wherein, the voice of the description speech data of the embedded multimedia file, when mating each multimedia file, is identified Feature；If the phonetic feature of the sounder for obtaining is mated with the phonetic feature of the speech data for identifying, it is determined that the multimedia File is the multimedia file of coupling；And according to the keyword that isolates, in the multimedia file matched by phonetic feature In mated；

The processing unit, is further used for according to the action keyword that isolates, crucial by describing to the matching unit The multimedia file that word is matched executes corresponding action.

14. devices according to claim 8-10 any one, it is characterised in that

The receiving unit, is further used for receiving the speech data of sounder；

The recognition unit, when being further used for the speech data that the receiving unit receives sounder, carries out phonetic feature Identification, obtains the phonetic feature of the sounder；To it needs to be determined that the description speech data being embedded in the multimedia file of copyright enters Row phonetic feature recognizes that acquisition describes the phonetic feature of speech data；

The matching unit, is further used for the language of the phonetic feature and description speech data for obtaining for determining the sounder for obtaining Sound feature is mated；

The processing unit, if be further used for the matching unit to determine the phonetic feature of the sounder for obtaining and retouching for obtaining State the phonetic feature coupling of speech data, it is determined that the sounder is the copyright owner of the multimedia file；Otherwise, it determines should Sounder copyright owner not for the multimedia file.