IP Library Granted Patent US 12,632,514
Granted Patent B2
US 12,632,514 · App. 19/306,727 · Granted May 19, 2026

Ai-generated music derivative works

Inventors: Christopher Horton (Santa Monica, CA); Jeremy Uzan (Santa Monica, CA); Sion Elliott (Santa Monica, CA); Daniel A. Drolet (Charleston, SC)
Assignee: Music IP Holdings, Inc.
G06F21/106G06F16/632G06F21/1084G06F40/205G06F2221/2137
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,632,514
App. No.
19/306,727
Granted
May 19, 2026
Kind
B2
Abstract

A system and method for creating AI-generated derivative works from predetermined content with copyright compliance and content owner control. In some aspects, the system receives predetermined content and a user-requested transformation theme, then employs generative artificial intelligence to create a derivative work. The system may enable scalable rights management for AI-generated content across music, video, text, and other media formats.

Claims (79)

1 . A system comprising:

a database server;

a content derivation platform comprising:

at least one processor; and

instructions that, when executed by the at least one processor, cause the system to:

receive predetermined content from at least one of the database server or a user upload;

receive a request to transform the predetermined content into a derivative work, wherein the request includes a requested theme directed to a characteristic of the derivative work, and wherein the at least one processor executes transformation operations based on the predetermined content and the requested theme;

evaluate, using an evaluation model, the request against pre-generation preference data;

in response to the request satisfying the pre-generation preference data, create the derivative work as a function of the predetermined content and the requested theme using generative artificial intelligence comprising a generative machine learning model configured to generate content based on the requested theme; and

provide access to the derivative work.

2 . The system of claim 1 , wherein the instructions further cause the system to:

after creating the derivative work, evaluate the derivative work against post-generation preference data; and

provide access to the derivative work in response to the derivative work satisfying the post-generation preference data.

3 . The system of claim 2 , wherein:

the pre-generation preference data comprises at least one of prohibited themes, restricted transformation types, or usage limitations; and

the post-generation preference data comprises at least one of explicit content filters, copyright infringement detection, or quality thresholds.

4 . The system of claim 1 , wherein the instructions further cause the system to:

embed a watermark in the derivative work before providing access, wherein the watermark comprises at least one of an identifier for a generative source, a flag indicating AI content, usage rights, attribution data, or a unique identifier.

5 . The system of claim 4 , wherein:

the watermark includes a time-to-live value; and

the instructions further cause the system to revoke access to the derivative work upon expiration of the time-to-live value.

6 . The system of claim 4 , wherein the instructions further cause the system to:

configure an authorization server to detect usage of the derivative work based on the watermark; and

automatically execute a smart contract to distribute payments to stakeholders upon detecting usage of the derivative work.

7 . The system of claim 4 , wherein to provide access to the derivative work comprises to:

request a verification of the watermark by an authorization server in response to an access attempt; and

maintain an access log that records each verified access instance for royalty calculation.

8 . The system of claim 1 , wherein the instructions further cause the system to:

apply a content approval machine learning model trained on historical approval decisions of a content owner to determine whether to create the derivative work.

9 . The system of claim 8 , wherein:

the content approval machine learning model is configured to identify prohibited elements in at least one of the request or the derivative work based on learned patterns from content owner feedback; and

the instructions further cause the system to reject at least one of creation of or access to the derivative work upon detection of prohibited elements.

10 . The system of claim 8 , wherein the instructions further cause the system to:

update the content approval machine learning model based on new content owner feedback for approved and rejected derivative works; and

adjust approval thresholds as the content approval machine learning model learns content owner preferences over time.

11 . The system of claim 1 , wherein to provide access to the derivative work comprises to:

generate a time-limited access token unique to a requesting user;

deliver the derivative work through a secure streaming protocol that prevents local storage; and

terminate access upon expiration of the time-limited access token.

12 . The system of claim 1 , wherein to provide access to the derivative work comprises to:

detect a geographic location of a user requesting access;

verify the geographic location against geographic restrictions associated with the derivative work; and

at least one of:

selectively enable or disable access based on geographic verification, or

selectively pay rights owners based on geographic verification.

13 . The system of claim 1 , wherein the requested theme directs the at least one processor to execute at least one of tempo modifications, instrumentation changes, style transformations, or genre adaptations.

14 . A system comprising:

a content derivation platform comprising:

at least one processor; and

instructions that, when executed by the at least one processor, cause the system to:

receive content;

receive a request to transform the content into a derivative work, wherein the request includes a requested theme directed to a characteristic of the derivative work, and wherein the at least one processor executes transformation operations based on the content and the requested theme;

evaluate, using an evaluation model, the request against pre-generation preference data;

in response to the request satisfying the pre-generation preference data, create the derivative work as a function of the content and the requested theme using generative artificial intelligence comprising a generative machine learning model configured to generate content based on the requested theme; and

provide access to the derivative work.

15 . A method for creating a derivative work, comprising:

receiving predetermined content from at least one of a database server or a user upload;

receiving a request to transform the predetermined content into a derivative work, wherein the request includes a requested theme directed to a characteristic of the derivative work, and wherein at least one processor executes transformation operations based on the predetermined content and the requested theme;

evaluating, using an evaluation model, the request against pre-generation preference data;

in response to the request satisfying the pre-generation preference data, creating the derivative work as a function of the predetermined content and the requested theme using generative artificial intelligence comprising a generative machine learning model configured to generate content based on the requested theme; and

embodying at least a portion of the derivative work in computer storage media.

16 . The method of claim 15 , further comprising:

conducting an interactive interview with a user through a chatbot interface to determine the requested theme; and

mapping user responses to pre-approved theme parameters stored in a filter database.

17 . The method of claim 15 , further comprising:

embedding a watermark in the derivative work, wherein the watermark includes a cryptographically signed hash that enables verification of authenticity; and

storing metadata about the derivative work in a data store to create an immutable record of the derivative work.

18 . The method of claim 15 , wherein creating the derivative work comprises:

encoding the predetermined content into a shared latent space using an encoder;

applying a diffusion model in the shared latent space to transform the predetermined content according to the requested theme; and

decoding transformed content from the shared latent space to generate the derivative work.

19 . The system of claim 1 , wherein the evaluation model and the generative machine learning model are a same machine learning model.

20 . The system of claim 1 , wherein the evaluation model and the generative machine learning model are different models.

21 . The method of claim 15 , further comprising:

after creating the derivative work, evaluating the derivative work against post-generation preference data; and

providing access to the derivative work in response to the derivative work satisfying the post-generation preference data.

22 . The method of claim 21 , wherein:

the pre-generation preference data comprises at least one of prohibited themes, restricted transformation types, or usage limitations; and

the post-generation preference data comprises at least one of explicit content filters, copyright infringement detection, or quality thresholds.

Continuity (4)
Continuation 19197818 · May 2, 2025
Continuation In Part 18926097 · Oct 24, 2024
Provisional Application 63592741 · Oct 24, 2023
Related Publication 20250371114A1 · Dec 4, 2025
References Cited (101)
US 6700989B1 · Itoh et al. · 2004 [cited by applicant]
US 6810388B1 · Sato · 2004 [cited by applicant]
US 12019982B2 · Veyseh et al. · 2024 [cited by applicant]
US 12080046B2 · Saraee et al. · 2024 [cited by applicant]
US 12086857B2 · Kharbanda et al. · 2024 [cited by applicant]
US 12105729B1 · Haq et al. · 2024 [cited by applicant]
US 12106318B1 · Chiang et al. · 2024 [cited by applicant]
US 12106548B1 · Brudalla et al. · 2024 [cited by applicant]
US 12118325B2 · Gray et al. · 2024 [cited by applicant]
US 12118976B1 · Chen et al. · 2024 [cited by applicant]
US 12165655B1 · Sandrew · 2024 [cited by examiner]
US 12204627B2 · Wexler · 2025 [cited by applicant]
US 20040024588A1 · Watson et al. · 2004 [cited by applicant]
US 20060004669A1 · Ito · 2006 [cited by examiner]
US 20060190970A1 · Hellman · 2006 [cited by applicant]
US 20060271494A1 · Ito · 2006 [cited by examiner]
US 20070140318A1 · Hellman · 2007 [cited by applicant]
US 20070266252A1 · Davis et al. · 2007 [cited by applicant]
US 20210233204A1 · Alattar et al. · 2021 [cited by applicant]
US 20220059063A1 · Balassanian et al. · 2022 [cited by applicant]
US 20220092267A1 · Hou et al. · 2022 [cited by applicant]
US 20220134914A1 · Jung · 2022 [cited by applicant]
US 20230095092A1 · Xiao et al. · 2023 [cited by applicant]
US 20230100289A1 · Kare et al. · 2023 [cited by applicant]
US 20230377099A1 · Kreis et al. · 2023 [cited by applicant]
US 20230377214A1 · Kansy et al. · 2023 [cited by applicant]
US 20240005604A1 · Kreis et al. · 2024 [cited by applicant]
US 20240095987A1 · Piramutha et al. · 2024 [cited by applicant]
US 20240152544A1 · Aykut et al. · 2024 [cited by applicant]
US 20240160902A1 · Padgett et al. · 2024 [cited by applicant]
US 20240185396A1 · Hatamizadeh et al. · 2024 [cited by applicant]
US 20240202795A1 · Kharbanda et al. · 2024 [cited by applicant]
US 20240253217A1 · Vahdat et al. · 2024 [cited by applicant]
US 20240282079A1 · Saraee et al. · 2024 [cited by applicant]
US 20240289407A1 · Rofouei et al. · 2024 [cited by applicant]
US 20240304177A1 · Wu et al. · 2024 [cited by applicant]
US 20240312087A1 · Agrawal et al. · 2024 [cited by applicant]
US 20240346629A1 · Harikumar et al. · 2024 [cited by applicant]
US 20250131928A1 · Drolet · 2025 [cited by applicant]
US 20250139375A1 · Bright et al. · 2025 [cited by applicant]
AU 2004258523A1 · 2006 [cited by examiner]
CA 2065641A1 · 2006 [cited by applicant]
CA 2605641A1 · 2006 [cited by examiner]
CA 2605646A1 · 2006 [cited by examiner]
CN 1525363A · 2004 [cited by examiner]
EP 1146411B2 · 2005 [cited by applicant]
JP 2004013493A · 2004 [cited by examiner]
JP 2004506947A · 2004 [cited by examiner]
JP 2004193843A · 2004 [cited by examiner]
JP 2006244075A · 2006 [cited by examiner]
JP 3990853B2 · 2007 [cited by examiner]
JP 4353651B2 · 2009 [cited by examiner]
JP 4456185B2 · 2010 [cited by examiner]
KR 100865247B1 · 2008 [cited by examiner]
TR 2024005874 · 2024 [cited by applicant]
TR 2024006991A2 · 2024 [cited by applicant]
WO WO2024097380A1 · 2024 [cited by examiner]
WO WO2024158853A1 · 2024 [cited by examiner]
WO WO2024220450A1 · 2024 [cited by examiner]
WO WO2024243183A2 · 2024 [cited by examiner]
Ramponi, Marco, “Recent developments in Generative AI for Audio”, AssemblyAI, retrieved from the internet on Oct. 20, 2024, https://www.assemblyai.com/blog/recent-developments-i n-generative-ai-for-audio/, 34 pages. [cited by applicant]
Weng, Lilian, “What are Diffusion Models?”, GitHub, Jul. 11, 2021, https://liliamweng.github.io/posts/2021-07-11-diffusion-models/#reverse-diffusion-process, 25 pages. [cited by applicant]
O'Connor, Ryan, “Automatic summarization with LLMs in Python”, AssemblyAl, retrieved from the internet on Oct. 20, 2024, https://www.assemblyai.com/blog/automatic-summarization-llms-python/, 12 pages. [cited by applicant]
“Apply LLMs to audio files, Learn how to leverage LLMs for speech using LeMUR”, Assembly AI, retrieved from the internet on Oct. 20, 2024, https://www.assemblyai.com/docs/getting-started/apply-llm-to-audio-files, 5 page… [cited by applicant]
Chen et al., “Buildin In-Video Search”, Medium, Nov. 6, 2023, https://netflixtechblog.com/building-in-videosearch-936766f0017c, 12 pages. [cited by applicant]
Stevens, Ingrid, “Chat with Your Audio Locally: A guide to RAG with Whisper, Ollama, and FAISS”, Medium, Nov. 19, 2023, https://medi um. com/@ingridstevens/chat-with-your-audio-locally-a-gui de-to-rag-with-whisperollama… [cited by applicant]
Anderson, Brian, “Reverse-Time Diffusion Equation Models”, Stochastic Processes and their Applications 12 (1982) 313-326, North-Holland Publishing Company, 14 pages. [cited by applicant]
“Content Moderation”, AssemblyAI, retrieved from the internet on Oct. 20, 2024, https://www.assembyai.com/docs/audio-intelligence/content-moderation, 11 pages. [cited by applicant]
Muthukumar, “Detecting Voiced, Unvoiced and Silent parts of a speech signal”, Medium, Mar. 19, 2024, https://muthuku37.medium.com/detecting-voiced-unvoiced-and-silent-parts-of-a-speech-signal-?4e6fbf5e 75, 26 pages. [cited by applicant]
“Diffusion Models: A Comprehensive High-Level Understanding”, Research Graph, Medium, May 21, 2024, https://medium.com/@researchgraph/diffusion-model-compreshensive-high-level-understanding-55d6ecad2cba, 22 pages. [cited by applicant]
Larcher, Mario, “Diffusion Transformer Explained”, Towards Data Science, Feb. 28, 2024, https://medium.com/towards-data-sciene/diffusion-transforer-explained-e603c4770f7 e, 19 pages. [cited by applicant]
O'Connor, Ryan, “Introduction to Diffusion Models for Machine Learning” AssemblyAI, May 12, 2022, https://www. assemblyai.com/blog/diffusion-models-for-machine-learning-introduction/, 34 pages. [cited by applicant]
Andreas, et al., “DRCap_Zeroshot_Audio-Captioning”, GitHub, retrieved from the internet on Oct. 20, 2024, https://github.com/X-LANCE/SLAM-LLM/blob/main/examples/drcap_zeroshot_aac/README.md, 3 pages. [cited by applicant]
Di Pietro, Mauro, “GenAI with Python: Build Agents from Scratch (Complete Tutorial)”, Towards Data Science, Sep. 29, 2024, https://towardsdatascience.com/genai-with-python-build-agents-from-scratch-com pletetutorial-4fc… [cited by applicant]
Ramesh, et al., “Hierarchical Text-Conditional Image Generation with CLIP Latents”, Cornell Univ., arXiv:2204.06125v1 [cs.CV], Apr. 13, 2022, 27 pages. [cited by applicant]
Ramirez, et al., “Voice Activity Detection. Fundamentals and Speech Recognition System Robustness.” InTech Open Science Open Minds, 2007, 24 pages. [cited by applicant]
Swimberghe, Niels, “How to integrate spoken audio into LangChain.js using AssemblyAI”, AssemblyAI, Aug. 15, 2023, https://www.assemblyai.com/blog/integrate-audio-langchainjs/, 12 pages. [cited by applicant]
“A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching”, F5-TTS, retrieved from the internet on Oct. 20, 2024, https:/swivid.github.io.F5-TTS/, 17 pages. [cited by applicant]
“CLIP: Connecting text and images”, OpenAI, Jan. 5, 2021, https://openai.com/index/clip/, 16 pages. [cited by applicant]
“CLIP”, Hugging Face, retrieved from the internet on Oct. 22, 2024, https://huggingface.co/docs/transformers/model_doc/clip, 49 pages. [cited by applicant]
Rustamy, Fahim, Phd., “CLIP Model and The Importance of Multimodal Embeddings”, Towards Data Science, Dec. 11, 2023, https://towardsdatascience.com/clip-model-and-the-importance-of-multimodalembeddings-1c8f6b13bf72, 20 … [cited by applicant]
“Diffusion Models from Scratch”, Hugging Face Diffusion Course, retrieved from the internet on Oct. 22, 2024, https://huggingface. co/learn/diffusion-course/en/unit1 /3, 31 pages. [cited by applicant]
“MC_MusicCaps”, GitHub, retrieved from the internet on Oct. 20, 2024, https://github.com/L-LANCES/SLAM-LLM/blob/main/examples/mc_musiccaps/README. md, 2 pages. [cited by applicant]
Briggs, James, “Quick-fire Guide to Multi-Modal ML With OpenAI's CLIP”, Towards Data Science, Aug. 11, 2022, https://towardsdatascience. com/quick-tire-guide-to-multi-modal-ml-with-openais-cli p-2dad7 e398ac0, 21 pages. [cited by applicant]
Bouchard, Louis-Francois, “Stable Diffusion for Videos Explained”, Towards AI, Nov. 29, 2023, https://pub.towardsai.net/stable-diffusion-for-videos-explained-fawf0b6af3b0, 15 pages. [cited by applicant]
Erdem, Kemal, “Step by Step visual introduction to Diffusion Models”, published Nov. 1, 2023, https://erdem.pl/2023/11/step-by-step-visual-introduction-to-diffusion-models, 15 pages. [cited by applicant]
“Stable Diffusion: Training Your Own Model in 3 Simple Steps”, run:ai, https://www.run.ai.guides/generative-ai/stablediffusion-training, 10 pages. [cited by applicant]
Stevens, Ingrid, “Uncovering Insights in Audio: An Exploration”, GitHub, retrieved from the internet on Oct. 20, 2024, https://github.com/ingridstevens/whisper-audio-transcriber/tree/main, 6 pages. [cited by applicant]
Palucha, Szymon, “Understanding OpenAI's CLIP model”, Medium, Feb. 24, 2024 https://medium.com/@paluchasz/understanding-openais-cl i p-m odel-6b52bade3fa3, 23 pages. [cited by applicant]
Chen et al., JEN-1 DreamStyler: Customized Musical Concept Learning via Pivotal Parameters Tuning, Cornell Univ., arXiv:2406.12292 [cs.SD], Jun. 18, 2024, 13 pages. [cited by applicant]
GitHub, “SLAM-MC”, retrieved from the internet on Oct. 20, 2024, https://github.com/X-LANCE/SLAM-LLM, 5 pages. [cited by applicant]
GitHub, “SLAM-LLM”, retrieved from the internet on Oct. 20, 2024, https://github.com/X-LANCE/SLAM-LLM, 4 pages. [cited by applicant]
Huggingface.co Blog, “The Annotated Diffusion Model”, retrieved from the internet on Oct. 20, 2024, https://huggingface.co/blog/annotated-diffusion, 38 pages. [cited by applicant]
Huggingface.co Blog, “Train a Diffusion Model”, retrieved from the internet on Oct. 20, 2024, https://huggingface.co/docs/ diffusers/tutorials/basic_training, 12 pages. [cited by applicant]
IBM, “What are Diffusion Models?”, retrieved from the internet on Oct. 22, 2024, 18 pages. [cited by applicant]
Lil 'Log, “What are Diffusion Models?”, retrieved from the internet on Oct. 22, 2024, 25 pages. [cited by applicant]
Assembly AI, “Topic Detection”, retrieved from the internet on Oct. 20, 2024, https://www.assemblyai.com/docs/audiointelligence/topic-detection, 6 pages. [cited by applicant]
Assembly AI, “Key Phrases”, retrieved from the internet on Oct. 20, 2024, https://www.assemblyai.com/docs/audiointelligence/key-phrases, 5 pages. [cited by applicant]
Assembly All, “Sentiment Analysis”, retrieved from the internet Oct. 20, 2024, https://www.assemblyai.com/docs/audio-intelligence/sentiment-analysis, 4 pages. [cited by applicant]
Assembly AI, “Summarization”, retrieved from the internet on Oct. 20, 2024, https://assemblyai.com/docs/audiointelligence/summarization, 6 pages. [cited by applicant]
Atal, Bishnu S.; and Rabiner, Lawrence R.; “A Pattern Recognition Approach to Voiced-Unvoiced-Silence Classification with Applications to Speech Recognition” IEEE Transactions on Acoustics, Speech, and Signal Processing… [cited by applicant]