Multi-level index method for data of audio fingerprint database

Abstract

The invention provides a multi-level index method and device for data of an audio fingerprint database in order to solve the technical problems that in the prior art, the searching performance of massive audio fingerprint databases is low, and the number of calculation times is larger during comparison searching. The method comprises the steps that the value data bit of each audio fingerprint in the data of the fingerprint database is segmented into an index part and a fingerprint data part, and the index part is a multi-level index; all levels in the index parts point to the next level through a pointer; the pointer at the final level of the index parts is pointed to the fingerprint data parts. The device comprises a device body used for segmenting the value data bit of each audio fingerprint in the data of the audio fingerprint database into the index part and the fingerprint data part, a device body used for pointing all the levels in the index parts to the next level through the pointer and a device body used for pointing the pointer at the final level of the index parts to the fingerprint data parts, wherein the index parts are the multi-level indexes.

Description

Technical field

[0001] The invention relates to the technical field of audio fingerprint identification, in particular to a multi-level indexing method for audio fingerprint library data. Background technique

[0002] Audio fingerprint recognition (commonly known as "listening to recognize music") is an emerging music search technology. This technology extracts digital features called "audio fingerprints" by analyzing the frequency spectrum of audio files, and uses digital features to construct "massive audio fingerprints" "Library" is deployed on a network server. The client can send the recorded unknown audio clips to the server. The server uses the same algorithm to extract the "audio fingerprint", and then uses the search algorithm to find similar features in the massive audio fingerprint library, and according to the found Several features of, speculate the target audio file, and return the result to the client to complete the entire recognition process. Audio fingerprint r...

Examples

Embodiment 1

[0033] Without loss of generality, such as figure 1 As shown, the present invention provides a multi-level indexing method for audio fingerprint library data, including:

[0034] S1000: Divide the value data bit of each audio fingerprint in the audio fingerprint library data into an index part and a fingerprint data part, where the index part is a multi-level index;

[0035] S2000: Point all levels in the index part to the next level through pointers; and

[0036] S3000: Point the last level pointer of the index part to the fingerprint data part.

[0037] The present invention divides the value data bit of each audio fingerprint in the audio fingerprint database data into an index part and a fingerprint data part, and provides technical support in terms of data structure for realizing multi-level index; In the value of the level index condition, the value that meets the index condition of the next level is retrieved. Due to the step-by-level comparison, the retrieval volume at the nex...

Embodiment 2

[0039] Preferably, such as figure 2 As shown, the present invention uses the first half of the segmented value as the index part YN, and the second half of the segmented value as the fingerprint data part SJ. Only the first half of the value is used as the index code to retrieve the value. Under the premise of affecting index accuracy, reduce the amount of calculation required for index comparison.

Embodiment 3

[0041] Preferably, such as image 3 As shown, on the basis of Embodiment 1, the present invention further includes the steps of the multi-level index method for audio fingerprint library data:

[0042] S4000: Build an index list, including:

[0043] S4100: Each level in the index list includes multiple index bits;

[0044] S4200: Combine each index bit at the previous level in the index list with

[0045] Corresponding to multiple index bits at one level;

[0046] S4300: All index bits at the last level correspond to the fingerprint data pointed to by them.

[0047] In order to facilitate indexing, the present invention constructs an index list whose data format is consistent with the index part. The construction of the index list provides powerful technical support for further realizing serialized storage of values ​​and efficient audio fingerprint data indexing. When searching for audio fingerprint data, only access to the index list can complete the data retrieval required for the "v...

Technical field

[0001] The invention relates to the technical field of audio fingerprint identification, in particular to a multi-level indexing method for audio fingerprint library data. Background technique

[0002] Audio fingerprint recognition (commonly known as "listening to recognize music") is an emerging music search technology. This technology extracts digital features called "audio fingerprints" by analyzing the frequency spectrum of audio files, and uses digital features to construct "massive audio fingerprints" "Library" is deployed on a network server. The client can send the recorded unknown audio clips to the server. The server uses the same algorithm to extract the "audio fingerprint", and then uses the search algorithm to find similar features in the massive audio fingerprint library, and according to the found Several features of, speculate the target audio file, and return the result to the client to complete the entire recognition process. Audio fingerprint r...

Examples

Embodiment 1

[0033] Without loss of generality, such as figure 1 As shown, the present invention provides a multi-level indexing method for audio fingerprint library data, including:

[0034] S1000: Divide the value data bit of each audio fingerprint in the audio fingerprint library data into an index part and a fingerprint data part, where the index part is a multi-level index;

[0035] S2000: Point all levels in the index part to the next level through pointers; and

[0036] S3000: Point the last level pointer of the index part to the fingerprint data part.

[0037] The present invention divides the value data bit of each audio fingerprint in the audio fingerprint database data into an index part and a fingerprint data part, and provides technical support in terms of data structure for realizing multi-level index; In the value of the level index condition, the value that meets the index condition of the next level is retrieved. Due to the step-by-level comparison, the retrieval volume at the nex...

Embodiment 2

[0039] Preferably, such as figure 2 As shown, the present invention uses the first half of the segmented value as the index part YN, and the second half of the segmented value as the fingerprint data part SJ. Only the first half of the value is used as the index code to retrieve the value. Under the premise of affecting index accuracy, reduce the amount of calculation required for index comparison.

Embodiment 3

[0041] Preferably, such as image 3 As shown, on the basis of Embodiment 1, the present invention further includes the steps of the multi-level index method for audio fingerprint library data:

[0042] S4000: Build an index list, including:

[0043] S4100: Each level in the index list includes multiple index bits;

[0044] S4200: Combine each index bit at the previous level in the index list with

[0045] Corresponding to multiple index bits at one level;

[0046] S4300: All index bits at the last level correspond to the fingerprint data pointed to by them.

[0047] In order to facilitate indexing, the present invention constructs an index list whose data format is consistent with the index part. The construction of the index list provides powerful technical support for further realizing serialized storage of values ​​and efficient audio fingerprint data indexing. When searching for audio fingerprint data, only access to the index list can complete the data retrieval required for the "v...

Images