Music and digital rights management systems and methods
The system employs transformer model-based AI for metadata processing to address the challenge of royalty identification and distribution, ensuring accurate and efficient royalty payment management for music rights holders.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-03-05
- Publication Date
- 2026-03-25
AI Technical Summary
Music rights holders face difficulties in accurately identifying and obtaining all royalties due to the disorderly nature of the Internet, necessitating systems and methods to assist in royalty payment management.
A system and method utilizing transformer model-based artificial intelligence for metadata matching and distribution, involving preprocessing, block grouping, AI matching, and post-processing to ensure accurate royalty data transmission to digital service providers.
Enables precise identification and distribution of royalties to music rights holders, reducing manual intervention and enhancing the accuracy and efficiency of royalty payments.
Smart Images
Figure 2026509825000001_ABST
Abstract
Description
Technical Field
[0001] Cross - Reference to Related Applications This application claims priority to U.S. Provisional Patent Application No. 62 / 809,590, filed on February 23, 2019, the disclosure and teachings of which are incorporated herein by reference, and is a continuation - in - part of U.S. Utility Patent Application No. 16 / 799,201, filed on February 24, 2020, and claims priority to currently - pending U.S. Utility Patent Application No. 18 / 117,575, filed on March 6, 2023.
Background Art
[0002] The present invention relates to systems and methods for music and digital rights management (DRM). Specifically, the present invention relates to systems and methods for converting, matching, and distributing metadata of music and digital works to usage data of related sound recordings of digital service providers (DSPs).
Summary of the Invention
Problems to be Solved by the Invention
[0003] Given the nature of the current Internet, which spreads widely and disorderly, it is a difficult task for music rights holders to accurately identify and obtain all the royalties that should be paid to them. Therefore, there is a need for systems and related methods to assist music rights holders and other digital rights holders in finding and obtaining the royalties that should be paid.
Means for Solving the Problems
[0004] In general, in one aspect, the present invention features a method for matching digital work metadata and delivering it to one or more digital service providers, the method comprising: modifying one or more digital work metadata files under the control of one or more processors composed of executable instructions, the modification of one or more digital work metadata files including removing non-essential data or segment error data, or performing language translation; reformatting one or more digital work metadata files to conform to a transformer model-based artificial intelligence matching operation; performing a block grouping operation on one or more digital work metadata files, the data associated with one or more digital work metadata files being grouped into blocks and analyzed for one or more pairs of data records; performing a transformer model-based artificial intelligence matching operation on one or more pairs of data records to determine whether each pair of data records contains a matching pair of data records; and transmitting the output data from the transformer model-based artificial intelligence matching operation to one or more digital service providers.
[0005] Implementations of the present invention may include one or more of the following features: One or more digital work metadata files may be sound recording metadata files. A transformer model-based artificial intelligence matching operation may include matching one or more basic criteria, which may include a work title, a work author, a sound recording title, or a sound recording artist. Output data may be transmitted to one or more digital service providers via one or more third-party application programming interfaces.
[0006] This method may further include sending the output data to a rights management dashboard before sending it to one or more digital service providers. This method may further include converting the output data to a data standard format corresponding to each of the one or more digital service providers before sending it to one or more digital service providers. This method may further include performing a second artificial intelligence matching operation on one or more pairs of data records, this second artificial intelligence matching operation is based on the Dedupe Python library. The transformer model-based artificial intelligence matching operation may be performed on a cloud infrastructure.
[0007] In general, in other embodiments, the present invention features a system configured to match digital work metadata and deliver it to one or more digital service providers, the system comprising one or more processors, one or more non-temporary computer-readable media, and one or more modules maintained on the one or more non-temporary computer-readable media, the non-temporary computer-readable media, when executed by one or more processors, causes the one or more processors to perform an operation, the operation being to modify one or more digital work metadata files, the modification of one or more digital work metadata files including removing non-essential data or segment error data, or performing language translation, and To adapt to a transformer model-based artificial intelligence matching operation, the method includes reformatting one or more digital work metadata files, performing a block grouping operation on one or more digital work metadata files, such that the data associated with one or more digital work metadata files is grouped into blocks and analyzed for one or more pairs of data records, performing a transformer model-based artificial intelligence matching operation on one or more pairs of data records to determine whether each pair of data records contains a matching pair of data records, and transmitting the output data from the transformer model-based artificial intelligence matching operation to one or more digital service providers.
[0008] Implementations of the present invention may include one or more of the following features: One or more digital work metadata files may be sound recording metadata files. A transformer model-based artificial intelligence matching operation may include matching one or more basic criteria, which may include a work title, a work author, a sound recording title, or a sound recording artist. Output data may be transmitted to one or more digital service providers via one or more third-party application programming interfaces.
[0009] This system may further include additional operations to send output data to a rights management dashboard before sending it to one or more digital service providers. This system may further include additional operations to convert output data to a data standard format corresponding to each of the one or more digital service providers before sending it to one or more digital service providers. This system may further include additional operations to perform a second artificial intelligence matching operation on one or more pairs of data records, this second artificial intelligence matching operation based on the Dedupe Python library. The transformer model-based artificial intelligence matching operation may be executed on a cloud infrastructure. [Brief explanation of the drawing]
[0010] [Figure 1] A flowchart of the operation of the system and related methods of one embodiment of the present invention is shown. [Figure 2] A flowchart of the operation of a system and related method of another embodiment of the present invention, which utilizes a job scheduling processing system, is shown. [Figure 3] This shows an operation flowchart of the job scheduling processing system used in the present invention. [Figure 4] A first operational flowchart of a system and related method of another embodiment of the present invention is shown. [Figure 5] Figure 4 shows a second operation flowchart of the system and related methods according to the embodiment. [Modes for carrying out the invention]
[0011] This invention relates to a system and related methods for converting, matching, and distributing metadata of music and digital works to relevant recording usage data for DSPs. The matching mechanism may include a customer data conversion mode and subsequent transmission to an artificial intelligence (AI) entity resolution library. This library may be configured for Python and may be based on open-source libraries such as Dedupe. The system may include a pre-matching analysis mechanism configured to extract positive matches from one or more datasets, such as one-to-one correlations of International Standard Work Codes (ISWC), International Standard Recording Codes (ISRC), and other authoritative data points, and, if there are any mismatches, to send them to an adapted Dedupe library. After AI processing of the data, the system may extract the results into a specific data standard or file format used to distribute ownership information to DSPs available in accordance with DSP submission requirements. The system may take the form of an automated software pipeline.
[0012] The system of the present invention may include several embodiments. The first embodiment is a pre-analysis transformation. This system can process metadata files, such as those provided by customers, and remove unnecessary data and / or split erroneous or potentially erroneous data. The system can then reformat the files, for example, to integrate the functionality of a DSP usage report into a database table.
[0013] The second aspect is pre-analysis matching. This system may utilize a metadata file when performing matching analysis with a DSP usage report table. This matching may be performed based on one or more data points, including ISWC, ISRC, and other authoritative data points. The resulting dataset may be analyzed to determine additional authoritative data points. In a non-limiting embodiment, if only ISRC is available in the metadata file, a new ISWC may be discovered, and if only ISWC is available in the metadata file, a new ISRC may be discovered. The newly discovered authoritative data points may be further matched, and the results are added, for example, to a table of compositions and associated recordings. In one embodiment of such a table, this table may represent a data cluster that is updated through AI matching. Compositions and recordings that do not form part of a data cluster may be subject to AI processing for predictive matching.
[0014] A third aspect is the aforementioned AI matching, particularly via an adapted Dedupe library. The adapted Dedupe library can perform clustering and entity resolution at scale on large datasets, such as those with over 25 million rows. The AI matching mechanism may be run on the entire dataset, including a training predicate set that supports matching on one or more basic criteria, such as work title and author, and recording title and artist. This Dedupe library may consist of matching techniques based on one or more statistical analyses, as well as a training predicate set for generating confidence scores for clustered matches. The generated results may include a table having one or more cluster identifiers, confidence scores, and additional data points added to the original database table. Clusters may be merged with clusters created during the pre-analysis matching aspect, thereby forming an affinity of recordings associated with a single work. Such affinity may be updated, for example, when new work data and DSP usage reports are processed by the system of the present invention.
[0015] A fourth aspect is a post-processing transformation configured to send matching files to the rights management dashboard. Once approved by the rights manager, this data can be transformed into a specific data standard, such as a flat file like a comma-separated values (CSV) file. This transformed data can then be ingested by an ingestion pipeline associated with the system for transforming the relevant metadata into a DSP-specific format.
[0016] Figure 1 shows an operational flowchart of the system and related methods of one embodiment of the present invention. A data file ("Customer Data File") may be uploaded to the rights management dashboard. The data file may be queued for pre-analysis transformation and to request the activation of the AI matching mechanism from Amazon Web Services (AWS), as the AI matching mechanism can only be activated on demand. After the data file is transformed and ingested into a PostgreSQL database, it may be merged into a DSP usage report table. Once the AI matching mechanism is activated, the data is sent to initiate AI matching. Specifically, a Dedupe-based pipeline may be run on the transformed / ingested dataset. The data file may then be transformed into a readable format of predicted matches from the raw AI output and delivered to the rights management dashboard. Once approved by the rights manager, the data may be ingested by the DSP ingestion pipeline. This data is transformed into a specific data standard format that supports transmission to all relevant DSPs and is then sent to the DSPs via a third-party application programming interface (API), etc.
[0017] Figure 2 shows an operational flowchart of a system and related method of another embodiment of the present invention that utilizes a job scheduling processing system, such as the job scheduling processing system embodied in Figure 3, although this is not limited to the embodiment of Figure 1. Figure 2 illustrates these operations, including pre-analysis transformation, pre-analysis matching, AI matching, and post-analysis transformation, as well as DSP ingestion pipeline operations, which are performed on a cloud infrastructure associated with the job scheduling processing system. Outside the cloud infrastructure, there is a rights management dashboard, which may be configured as a graphical user interface (GUI).
[0018] Figure 3 shows an operational flowchart of a job scheduling processing system for use in the present invention. The job scheduling processing system is configured to marshal, monitor, and report on one or more jobs or operational pipelines through the cloud infrastructure. Any type of computing job can be scheduled, picked up, and assigned for cloud computing by the job scheduling processing system. The job scheduling processing system is configured to remain active or available and can scale up the cloud infrastructure as required or needed by the requested computing jobs.
[0019] Figure 3 shows multiple job or operational pipelines, including those related to data ingestion, data distribution, and content discovery, as well as those related to the AI-based system. A pipeline may consist of several computational tasks, indicated by subscripts, that produce a desired output once all related tasks are completed. The job scheduling processing system keeps track of the computational tasks required for a particular pipeline through a database table that defines the tasks and related steps required for the completion of the pipeline. These pipelines illustrated and described are illustrative and not limiting.
[0020] As further illustrated in Figure 3, the modes that can be controlled or operated by the job scheduling processing system include: importing data from the rights management dashboard into the database (e.g., triggered by an end user), transforming data to supply to the AI-based system, transforming and importing result data from the AI-based system into the database, transforming data into a data format acceptable for distribution to all DSPs (including interpretation of rights management regarding the results), and distributing data to DSPs.
[0021] Figures 4-5 show the operation flowcharts of the system and related methods of another embodiment of the present invention. This embodiment is different from the embodiment of FIG. 1 in several aspects. First, before the start of the AI matching mechanism, a blocking mechanism based on the tokenization of specific metadata fields, including but not limited to the work title, work author / author, recording title, and recording artist, is applied to the pipelined data. Next, the blocked data is applied with a candidate pair generation algorithm, which analyzes the blocks and generates a group of candidate pairs of similar data based on the defined threshold. The candidate pairs can be stored in an easily accessible storage system to enable rapid search by the AI mechanism of this system, which can determine whether the candidate pairs actually match (yes / no) without the need for human interpretation or involvement. The AI mechanism can be executed in cooperation with the aforementioned Dedupe library and can utilize either or both sets of results as needed, including depending on specific tasks.
[0022] Another aspect of the embodiment of FIGS. 4-5 is that the AI matching pipeline uses a transformer model-based AI trained specifically for digital asset metadata, such as music metadata like work title, author / author, recording title, artist, etc. Utilizing the transformer model is effective in eliminating the need for human involvement in human interpretation and decision-making regarding the results. For example, the model can output a determination of whether two entities are the same with a 95% confidence level.
[0023] Another aspect of the embodiment of FIGS. 4-5 is to include initial preprocessing in the ingestion pipeline. Before the AI matching aspect of this system, various automated preprocessing pipelines for the ingested data can be utilized, including but not limited to language translation, removal of unnecessary data or null values, and sorting or deduplication of records.
[0024] At the end of the AI matching pipeline, the matched entities can be stored for future use. Specifically, the matched entities can be stored or represented in a database for normalization within each respective work or digital asset to which the matched entities are linked, based on the associated source data (e.g., the data of set A in FIG. 5). By storing the matched entities, in future analysis including additional metadata, for example, the enhancement of related data such as the data of set A illustrated in FIG. 5 becomes possible, and as the amount of data passing through the system of the present invention increases, the blocking and matching functions improve over time.
[0025] FIG. 4 provides an operation flowchart having a structure corresponding to FIG. 1. FIG. 5 provides an alternative operation flowchart of FIG. 4, emphasizing four pipeline aspects, namely, data joining, followed by preprocessing of the joined data set, then blocking for candidate pair generation, and finally, the evaluation of candidate pairs based on the use of, for example, a Dedupe library or a trained natural language processing (NLP) transformer model-based AI as described above.
[0026] The system embodied in FIGS. 4-5 can be used not only to discover relevant matches and distribute them to various digital service providers, but also in conjunction with royalty adjustment, i.e., the adjustment of non-matching revenue data. By using and incorporating relevant revenue data, such as that indicated by set B illustrated in FIG. 5, the system of the present invention including the modified preprocessing pipeline can facilitate matching on internal revenue data to associate revenue with the copyright holders.
[0027] The embodiments and examples described above are illustrative and can be modified in many ways without departing from the spirit and scope of this disclosure or the scope of the invention. For example, elements and / or features of the various descriptive and exemplary embodiments described herein can be combined with and / or substituted for each other within the scope of this disclosure. To further understand the invention, its operational advantages and the specific purposes achieved by its use, please refer to the drawings and descriptions, which illustrate preferred embodiments of the invention.
Claims
1. A method for matching digital copyright metadata and distributing it to one or more digital service providers, Under the control of one or more processors composed of executable instructions, Modifying one or more digital work metadata files, and such modification of one or more digital work metadata files includes removing non-essential data or segment error data, or performing language translation. To adapt to the transformer model-based artificial intelligence matching operation, one or more digital copyright metadata files are reformatted, Performing a block grouping operation on one or more digital copyright metadata files, wherein the data associated with the one or more digital copyright metadata files is grouped into blocks and analyzed for one or more pairs of data records, Perform the transformer model-based artificial intelligence matching operation on one or more pairs of data records to determine whether each of the one or more pairs of data records contains a matching pair of data records, A method comprising transmitting output data from the transformer model-based artificial intelligence matching operation to one or more digital service providers.
2. The method according to claim 1, wherein the one or more digital copyright metadata files are audio metadata files.
3. The method according to claim 1, wherein the transformer model-based artificial intelligence matching operation includes matching one or more basic criteria, the one or more basic criteria including a work title, a work author, a sound recording title, or a sound recording artist.
4. The method according to claim 1, wherein the output data is transmitted to one or more digital service providers via one or more third-party application programming interfaces.
5. The method according to claim 1, further comprising transmitting the output data to a rights management dashboard before transmitting it to one or more digital service providers.
6. The method according to claim 1, further comprising converting the output data into a data standard format corresponding to each of the one or more digital service providers before transmission to the one or more digital service providers.
7. The method according to claim 1, further comprising performing a second artificial intelligence matching operation on one or more pairs of data records, wherein the second artificial intelligence matching operation is based on the Dedup Python library.
8. The method according to claim 1, wherein the transformer model-based artificial intelligence matching operation is performed on a cloud infrastructure.
9. A system configured to match digital copyright metadata and distribute it to one or more digital service providers, One or more processors, One or more non-temporary computer-readable media, The non-temporary computer-readable medium includes one or more modules maintained on one or more non-temporary computer-readable media, and when executed by one or more processors, the non-temporary computer-readable media causes one or more processors to perform an operation. The aforementioned operation involves modifying one or more digital work metadata files, and such modification of one or more digital work metadata files includes removing non-essential data or segment error data, or performing language translation. To adapt to the transformer model-based artificial intelligence matching operation, one or more digital copyright metadata files are reformatted, Performing a block grouping operation on one or more digital copyright metadata files, wherein the data associated with the one or more digital copyright metadata files is grouped into blocks and analyzed for one or more pairs of data records, Perform the transformer model-based artificial intelligence matching operation on one or more pairs of data records to determine whether each of the one or more pairs of data records contains a matching pair of data records, A system comprising transmitting output data from the transformer model-based artificial intelligence matching operation to one or more digital service providers.
10. The system according to claim 9, wherein the one or more digital copyright metadata files are audio metadata files.
11. The system according to claim 9, wherein the transformer model-based artificial intelligence matching operation includes matching one or more basic criteria, the one or more basic criteria including a work title, a work author, a recording title, or a recording artist.
12. The system according to claim 9, wherein the output data is transmitted to one or more digital service providers via one or more third-party application programming interfaces.
13. The system according to claim 9, further comprising the additional operation of sending the output data to a rights management dashboard before sending it to one or more digital service providers.
14. The system according to claim 9, further comprising an additional operation to convert the output data into a data standard format corresponding to each of the one or more digital service providers before transmission to the one or more digital service providers.
15. The system according to claim 9 further includes an additional operation of performing a second artificial intelligence matching operation on one or more pairs of data records, wherein the second artificial intelligence matching operation is based on the Dedup Python library.
16. The system according to claim 9, wherein the transformer model-based artificial intelligence matching operation is performed on a cloud infrastructure.