DCT

1:24-cv-01166

Muvox LLC v. Spotify USA Inc

Key Events
Complaint
complaint

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 1:24-cv-01166, D. Del., 10/18/2024
  • Venue Allegations: Venue is alleged to be proper in the District of Delaware because Defendant is incorporated in Delaware and has an established place of business in the District.
  • Core Dispute: Plaintiff alleges that Defendant’s music streaming service infringes a patent related to the automatic categorization of music tracks based on acoustic attributes to generate and stream personalized playlists.
  • Technical Context: The technology concerns systems for analyzing the acoustic properties of music to classify tracks by mood or other categories, enabling the creation of customized streaming experiences for end-users.
  • Key Procedural History: The patent-in-suit claims priority from a long chain of predecessor applications, including a provisional application filed in 2014, indicating a lengthy prosecution history. The complaint does not mention any prior litigation or post-grant proceedings involving the patent family.

Case Timeline

Date Event
2014-03-27 U.S. Provisional Application 61/971,490 Priority Date
2024-02-13 U.S. Patent No. 11,899,713 Issues
2024-10-18 Complaint Filed

II. Technology and Patent(s)-in-Suit Analysis

  • Patent Identification: U.S. Patent No. 11,899,713, "Music streaming, playlist creation and streaming architecture", issued February 13, 2024 (’713 Patent).

U.S. Patent No. 11,899,713 - "Music streaming, playlist creation and streaming architecture"

The Invention Explained

  • Problem Addressed: The patent describes a market where various music publishing entities (e.g., radio stations, independent labels) have catalogs of music but lack a practical way to offer customized streaming services to end-users that also attribute the content to the specific publisher ( Compl. ¶9; ’713 Patent, col. 1:49-61).
  • The Patented Solution: The invention is a system that analyzes music to derive "rhythm, texture and pitch (RTP) scores" for each track. These objective scores, derived from low-level acoustic data, are stored in a universal database. End-user applications, sponsored by a music publisher, access this database via an API to create personalized playlists. This architecture allows a publisher to offer a customized streaming experience using only tracks it is licensed to stream, without building the entire infrastructure from the ground up (’713 Patent, Abstract; col. 2:5-34).
  • Technical Importance: The described architecture aims to provide a platform for smaller or specialized music publishers to compete in the streaming market by leveraging a common technical framework for music analysis while maintaining their own branding and user relationships (’713 Patent, col. 2:28-34).

Key Claims at a Glance

  • The complaint alleges infringement of "exemplary method claims" without specifying claim numbers (Compl. ¶11). Independent claim 1 is a method claim.
  • Independent Claim 1 recites a method for selecting a song, which includes the following essential elements:
    • Selecting a song based on a computer-derived comparison between a representation of the song and known similarities in representations of other songs.
    • The known similarities are based at least in part on a "human-trained machine."
    • The song's representation is based on isolating and identifying its "frequency characteristics."
    • The representations of the "other songs" are based on a "human listening" to identify their frequency characteristics.
    • The frequency characteristics of the song and the other songs correspond to "one or more moods."
    • The selection is based on the similarity between the moods of the song and the moods of the other songs.
  • The complaint’s general allegations suggest it may assert other claims, including dependent claims, as the case proceeds (Compl. ¶11).

III. The Accused Instrumentality

Product Identification

The complaint does not specifically identify any accused products by name, referring only to "Exemplary Defendant Products" and "Defendant products identified in the charts" (Compl. ¶11). It further states that these products and the infringement allegations are detailed in an Exhibit 2, which was not publicly filed with the complaint (Compl. ¶13-14). Accordingly, the complaint does not provide sufficient detail for analysis of the accused instrumentality's specific functionality or market context.

IV. Analysis of Infringement Allegations

The complaint references claim-chart exhibits that were not provided. The following summarizes the narrative infringement theory.

The complaint alleges that Defendant directly infringes "exemplary method claims" of the ’713 Patent through the actions of making, using, selling, and offering for sale its "Exemplary Defendant Products" (Compl. ¶11). The core of the infringement theory, as stated in the complaint, is that these products "practice the technology claimed by the '713 Patent" and that they "satisfy all elements of the Exemplary '713 Patent Claims" (Compl. ¶13). The complaint also asserts direct infringement based on Defendant's employees internally testing and using the accused products (Compl. ¶12). Without the specific claim charts from Exhibit 2, a detailed, element-by-element comparison is not possible from the face of the complaint.

No probative visual evidence provided in complaint.

  • Identified Points of Contention:
    • Scope Questions: A central dispute may concern the meaning of a "human-trained machine" as required by the claim. The question for the court could be whether a system trained on vast user-generated data (e.g., user-created playlists with mood-based titles) meets this limitation, or if the claim requires a more direct human process of scoring and labeling tracks for the express purpose of training the machine, as described in the patent's specification (’713 Patent, cl. 1; col. 4:35-40).
    • Technical Questions: A key evidentiary question will be whether the accused Spotify service actually performs the claimed step of selecting songs "based on isolating and identifying frequency characteristics" that "correspond to one or more moods." Plaintiff may need to show that Spotify's recommendation engine relies on this specific technical pathway, as opposed to or in addition to other methods like collaborative filtering, user listening history, or manual curation, which may not map directly onto the claim language (’713 Patent, cl. 1).

V. Key Claim Terms for Construction

  • The Term: "human-trained machine" (’713 Patent, cl. 1).
  • Context and Importance: This term is critical for distinguishing the claimed invention from a purely autonomous or different type of machine learning system. The viability of the infringement claim may depend on whether Spotify's algorithm, whatever its nature, can be characterized as a "human-trained machine" within the patent's meaning.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The specification discusses training a neural network on a "large-scale collection of human-labeled sound clips," which could support an argument that any system trained on data that was originally labeled by humans meets the definition (’713 Patent, col. 4:56-59).
    • Evidence for a Narrower Interpretation: The patent also describes a specific embodiment where "tracks sufficient to create a sample set may be listened to by humans to develop RTP scores that correspond to each track," which are then used for training (’713 Patent, col. 4:35-40). This could support a narrower construction requiring that the training process itself be based on this specific human-scoring activity.
  • The Term: "frequency characteristics...correspond to one or more moods" (’713 Patent, cl. 1).
  • Context and Importance: This limitation links an objective technical feature ("frequency characteristics") to a subjective quality ("moods"). Infringement will depend on proving that the accused system establishes and uses this specific causal link, rather than arriving at mood-based playlists through other means.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The specification states that the identified "moods are just examples of categories" and that categories could be "colors, locations, etc., including completely arbitrary or made-up words," suggesting the term is not limited to conventional emotional states (’713 Patent, col. 6:27-32).
    • Evidence for a Narrower Interpretation: The patent repeatedly uses specific mood examples such as "happy, excited, manic, peaceful, sad and cautious" (’713 Patent, col. 6:18-22). An argument could be made that the term, in the context of the invention, implies a system that categorizes music along recognized affective or emotional lines.

VI. Other Allegations

  • Indirect Infringement: The complaint pleads only one count for "Direct Infringement" and does not allege facts to support a claim for either induced or contributory infringement (Compl. p. 2).
  • Willful Infringement: The complaint does not contain an explicit allegation of willful infringement. However, the prayer for relief asks the court to declare the case "exceptional" and award attorneys' fees pursuant to 35 U.S.C. § 285, which may indicate Plaintiff's intent to develop a record of willfulness or other sanctionable conduct during the litigation (Compl. ¶E(i)).

VII. Analyst’s Conclusion: Key Questions for the Case

  • A central issue will be one of evidentiary proof: Given the proprietary and complex nature of modern recommendation algorithms, a key challenge for the plaintiff will be to present evidence demonstrating that Spotify's system operates according to the specific technical steps of the asserted claims—namely, that it selects music by comparing frequency characteristics that have been mapped to moods via a "human-trained machine."
  • The case will also likely turn on a question of claim construction: Can the term "human-trained machine," as defined by the patent's specification, be construed to read on a system that learns from user behavior and other large-scale data inputs, or is it limited to the more explicit human-scoring and training process described in the patent's preferred embodiments?