DCT

5:25-cv-00680

Fraunhofer Gesellschaft zur Forderung der angewandten Forschung Ev v. Lenovo Group Ltd

Log In
Key Events
Complaint

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 5:25-cv-00680, E.D.N.C., 10/27/2025
  • Venue Allegations: Plaintiff alleges venue is proper in the Eastern District of North Carolina because Defendants Lenovo US and Motorola Mobility maintain offices in the district, have employees in the state, and have previously asserted in other litigation that the district is a proper venue for patent disputes involving them.
  • Core Dispute: Plaintiff alleges that Defendant’s mobile devices capable of utilizing the Enhanced Voice Services (EVS) standard infringe six U.S. patents related to audio and speech codec technology.
  • Technical Context: The technology concerns methods for efficiently encoding, decoding, and processing audio signals for voice calls, which are fundamental to standards like EVS used for Voice over LTE (VoLTE) and 5G voice services on modern wireless networks.
  • Key Procedural History: The complaint states that the asserted patents have been declared essential to the 3GPP EVS standard and are subject to Fair, Reasonable, and Non-Discriminatory (FRAND) licensing obligations. Plaintiff alleges it initiated licensing discussions by sending a FRAND package to Lenovo on October 13, 2023, but that communications from Lenovo ceased after March 22, 2024, leading to this lawsuit.

Case Timeline

Date Event
2011-02-14 Earliest Priority Date for ’263 Patent
2013-01-29 Earliest Priority Date for ’631 Patent
2013-06-21 Earliest Priority Date for ’624 and ’455 Patents
2014-07-28 Earliest Priority Date for ’993 and ’317 Patents
2014-01-01 3GPP adopts the EVS Standard
2016-01-01 T-Mobile upgrades its U.S. network to support EVS
2017-03-14 ’263 Patent Issued
2018-10-02 ’993 Patent Issued
2019-03-19 Lenovo allegedly becomes aware of ’455 Patent family during prosecution of its own patent
2019-04-02 ’317 Patent Issued
2019-11-12 ’455 Patent Issued
2020-05-05 ’624 Patent Issued
2022-12-06 ’631 Patent Issued
2023-10-13 Plaintiff sends FRAND package to Defendant
2025-10-27 Complaint Filed

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

U.S. Patent No. 11,521,631 - “Apparatus and Method for Selecting One of a First Encoding Algorithm and a Second Encoding Algorithm”

The Invention Explained

  • Problem Addressed: Switched audio codecs must decide which encoding algorithm to use for different parts of an audio signal (e.g., one for speech, another for music). The patent describes that making this decision via a "closed-loop" approach—which involves fully encoding and decoding the signal with each algorithm to compare quality—is computationally complex and expensive, while simpler "open-loop" methods are less reliable (Compl. ¶¶58-59; ’631 Patent, col. 1:40-60).
  • The Patented Solution: The invention claims a more efficient "open-loop" selection method. It proposes estimating a quality measure for each encoding algorithm without performing the actual, full encoding and decoding process. This allows the system to make a high-quality selection comparable to a closed-loop method but with significantly reduced computational complexity (Compl. ¶61; ’631 Patent, col. 4:16-37).
  • Technical Importance: This approach enables resource-constrained devices, such as smartphones, to perform high-quality, real-time codec mode selection, improving audio fidelity without demanding excessive processing power or battery life (Compl. ¶¶58, 60).

Key Claims at a Glance

  • The complaint asserts independent claim 1 (Compl. ¶71).
  • Essential elements of claim 1 include:
    • An apparatus for selecting one of a first and second encoding algorithm.
    • A first estimator for estimating a first quality measure associated with the first algorithm, without actually encoding and decoding the audio portion with that algorithm.
    • A second estimator for estimating a second quality measure associated with the second algorithm, without actually encoding and decoding the audio portion with that algorithm.
    • A controller for selecting an algorithm based on a comparison of the first and second quality measures.
    • A specific process for the first estimator, including receiving an input signal, windowing it, transforming it with a Modified Discrete Cosine Transform (MDCT), shaping the spectrum with weighted Linear Prediction Coding (LPC) coefficients, and estimating a global gain.
  • The complaint reserves the right to assert additional claims (Compl. ¶71).

U.S. Patent No. 9,595,263 - “Encoding and Decoding of Pulse Positions of Tracks of an Audio Signal”

The Invention Explained

  • Problem Addressed: In certain audio codecs (like ACELP), audio signals are represented by a series of "pulses" within an algebraic codebook. The patent background indicates that encoding the positions of these pulses traditionally requires a large number of bits, which is inefficient for data compression (Compl. ¶98; ’263 Patent, col. 4:22-28).
  • The Patented Solution: The invention describes a more efficient method for encoding and decoding this pulse information by using a single, consolidated "state number." This state number, when used in conjunction with known parameters like the total number of track positions and the total number of pulses, allows a decoder to reconstruct the precise pulse locations and signs using significantly fewer bits compared to prior art methods (Compl. ¶99; ’263 Patent, col. 4:29-34).
  • Technical Importance: This method improves the data compression efficiency of algebraic codebook-based codecs, which is a critical factor in delivering high-quality voice over bandwidth-limited wireless channels (Compl. ¶98).

Key Claims at a Glance

  • The complaint asserts independent claim 1 (Compl. ¶109).
  • Essential elements of claim 1 include:
    • An apparatus for decoding an encoded audio signal comprising one or more tracks with pulses.
    • A pulse information decoder configured to decode a plurality of pulse positions by only using one track position number, one total pulse number and one state number.
    • A signal decoder for generating a synthesized audio signal using the decoded pulse positions.
    • The pulse information decoder is also adapted to decode pulse signs using the same three inputs (track position number, total pulse number, state number).
    • The signal decoder is adapted to use these pulse signs in generating the synthesized audio signal.
    • At least one of the decoders comprises a hardware implementation.
  • The complaint reserves the right to assert additional claims (Compl. ¶109).

U.S. Patent No. 10,643,624 - “Apparatus and Method for Improved Concealment of the Adaptive Codebook in ACELP-Like Concealment Employing Improved Pulse Resynchronization”

  • Technology Synopsis: The patent addresses the problem of reconstructing audio frames that are lost during transmission (packet loss). The invention provides an improved method for this "concealment" process by more accurately reconstructing the timing of audio pulses (pitch cycles) in the lost frame, taking into account differences between the pitch cycles in available frames and the cycles to be constructed. This results in better sound quality when packets are lost (Compl. ¶138).
  • Asserted Claims: Independent claim 1 (Compl. ¶148).
  • Accused Features: The accused Lenovo EVS Products are alleged to implement the frame loss concealment and pulse resynchronization procedures defined in the EVS Standard (Compl. ¶¶152-153).

U.S. Patent No. 10,089,993 - “Apparatus and Method for Comfort Noise Generation Mode Selection”

  • Technology Synopsis: The patent relates to Comfort Noise Generation (CNG), which creates artificial background noise during silent periods of a call to avoid unnerving silence. The invention describes a method for selecting the best CNG mode from two or more available modes (e.g., a frequency-domain mode for car noise, a linear-prediction mode for office noise) based on the specific characteristics of the actual background noise, thereby improving the quality and realism of the generated comfort noise (Compl. ¶177; ’993 Patent, Abstract).
  • Asserted Claims: Independent claim 1 (Compl. ¶187).
  • Accused Features: The accused EVS Products are alleged to implement the EVS standard's discontinuous transmission (DTX) and CNG functionalities, which include selecting between LP-CNG and FD-CNG modes based on background noise characteristics (Compl. ¶¶192, 195).

U.S. Patent No. 10,475,455 - “Method and Apparatus for Obtaining Spectrum Coefficients for a Replacement Frame of an Audio Signal, Audio Decoder, Audio Receiver, and System for Transmitting Audio Signals”

  • Technology Synopsis: This patent discloses a method for frame-loss concealment at an audio receiver. The invention detects a tonal component in an audio signal by identifying a peak that exists in the spectra of the last two frames preceding the lost frame. It then predicts the spectrum coefficients for that tonal component in the lost frame, improving the reconstruction of tonal sounds like music or certain voice characteristics (Compl. ¶216).
  • Asserted Claims: Apparatus claim 28 (incorporating method claim 1) (Compl. ¶226).
  • Accused Features: The accused EVS Products are alleged to practice the EVS standard's packet loss concealment methods, which include tonal concealment using phase prediction based on detecting peaks in the power spectrum of previous frames (Compl. ¶¶232-233).

U.S. Patent No. 10,249,317 - “Estimating Noise of an Audio Signal in a LOG2-Domain”

  • Technology Synopsis: The patent describes a more computationally efficient method for noise estimation in audio signals. Traditional methods that operate on linear energy values require high-bit-depth processing (e.g., 40-bit), which is complex for mobile devices. The invention converts energy values into a logarithmic (log-2) domain first, allowing the noise estimation algorithm to run directly on this data with lower precision (e.g., 16-bit), thus reducing computational overhead (Compl. ¶253).
  • Asserted Claims: Non-transitory digital storage medium claim 9 (Compl. ¶263).
  • Accused Features: The accused EVS Products are alleged to perform noise estimation for functions like CNG by determining energy values, converting them into the log-2 domain, and estimating the noise level directly in that domain, as required by the EVS standard (Compl. ¶¶269-271).

III. The Accused Instrumentality

Product Identification

  • The complaint identifies the accused instrumentalities as "Lenovo’s EVS Products" (Compl. ¶9). This category is defined to include all Lenovo 4G and 5G products with Enhanced Voice Services (EVS) capability, with the Motorola Moto G Power cited as a specific example (Compl. ¶9).

Functionality and Market Context

  • The accused products are mobile devices, such as smartphones, that incorporate hardware and software to comply with the 3GPP EVS Standard (Compl. ¶¶9, 67, 104). This standard enables high-definition voice services, often marketed as "Enhanced HD Voice," "Ultra HD Voice," or "HD Voice+," over modern wireless networks like 4G VoLTE and 5G (Compl. ¶¶9, 28, 40). The EVS codec is designed to provide high compression and flexibility by combining different coding strategies, such as time-domain prediction-based coding (ACELP) for speech and frequency-domain transform-based coding (MDCT/TCX) for other signal types like music (Compl. ¶32). The complaint alleges that EVS has become the dominant voice service codec in LTE and 5G networks (Compl. ¶¶28, 38-39).

IV. Analysis of Infringement Allegations

’631 Patent Infringement Allegations

Claim Element (from Independent Claim 1) Alleged Infringing Functionality Complaint Citation Patent Citation
An apparatus for selecting one of a first encoding algorithm comprising a first characteristic and a second encoding algorithm comprising a second characteristic... Lenovo’s EVS Products implement the EVS standard, which employs multiple coding technologies, including a linear prediction-based (LP-based) approach and a transform-domain (MDCT) approach, and selects between them for each frame. ¶¶75, 76 col. 1:18-25
a first estimator for estimating a first quality measure ... without actually encoding and decoding ...; a second estimator for estimating a second quality measure ... without actually encoding and decoding... The EVS standard specifies a "technology selector" that bases its decision on two estimates of the segmental SNR—one corresponding to the transform-based technology (MDCT) and another to the predictive technology (ACELP)—without performing a full encode/decode loop. ¶76 col. 4:19-23
a controller for selecting the first encoding algorithm or the second encoding algorithm based on a comparison between the first quality measure and the second quality measure, The EVS standard's technology selector makes a decision to use either ACELP or MDCT-based technology based on a comparison of the two SNR estimates and a hysteresis mechanism. ¶78 col. 4:24-27
wherein, in estimating the first quality measure, the first estimator is configured to receive an input signal, window the input signal, transform the windowed input signal using a MDCT... shape the obtained spectrum with weighted LPC... and estimate a global gain... The EVS standard's procedure for "Segmental SNR estimation of the MDCT-based technology" involves filtering the input signal, windowing and transforming with an MDCT, shaping the resulting spectrum with weighted LPC, and deriving the segmental SNR from the global gain. The complaint includes a screenshot from 3GPP TS 26.445 detailing this process. ¶79 col. 14:2-8
  • Identified Points of Contention:
    • Scope Question: A primary point of contention may be the meaning of "without actually encoding and decoding." The Defendant could argue that the EVS standard's SNR estimation process, which involves steps like MDCT transformation and LPC shaping, constitutes a form of partial encoding, and thus does not meet this negative limitation. The case may turn on whether the court construes "actually encoding" to mean only the full process of generating a final bitstream.

’263 Patent Infringement Allegations

Claim Element (from Independent Claim 1) Alleged Infringing Functionality Complaint Citation Patent Citation
a pulse information decoder for decoding a plurality of pulse positions ... by only using one track position number, one total pulse number and one state number... The EVS standard's algebraic codebook decoding process enumerates possible pulse constellations in a track of length L containing p pulses. The state s(c) is decoded by an algorithm using these parameters, which correspond to the claimed numbers, to determine the pulse positions. ¶115 col. 17:53-61
a signal decoder for decoding the encoded audio signal by generating a synthesized audio signal using the plurality of pulse positions and a plurality of predictive filter coefficients... The EVS standard specifies a "General LP-based decoding" process where LSF parameters are decoded from the bitstream, converted to LP coefficients, and used to synthesize the output signal from the excitation. ¶116 col. 18:1-5
wherein the pulse information decoder is furthermore adapted to decode a plurality of pulse signs by only using the track position number, the total pulse number and the state number... The EVS standard's "Decoding the algebraic codebook vector" process is used to extract the positions and amplitudes (signs) of the excitation pulses. The state is decoded via an algorithm that recursively determines the pulse signs. A screenshot from 3GPP TS 26.445 shows the algorithm for decoding the state. ¶118 col. 18:11-15
  • Identified Points of Contention:
    • Technical Question: A potential dispute could arise over whether the parameters used in the EVS standard's decoding algorithm—such as track length L, number of pulses p, and the overall state s(c)—are functionally and definitionally equivalent to the claim terms "one track position number," "one total pulse number," and "one state number." The defense may argue for a technical mismatch between the standard's implementation and the specific terms recited in the claim.

V. Key Claim Terms for Construction

  • For the ’631 Patent:

    • The Term: "without actually encoding and decoding"
    • Context and Importance: This negative limitation is central to the patent's claimed advance over computationally expensive "closed-loop" systems. The infringement analysis for claim 1 will depend entirely on whether the EVS standard's SNR estimation procedure is found to fall "without" this claimed activity. Practitioners may focus on this term because it creates a potential factual dispute over how the accused standard operates relative to the claim's exclusionary language.
    • Intrinsic Evidence for Interpretation:
      • Evidence for a Broader Interpretation: The specification contrasts the invention with closed-loop selection and states, "The quality measures are estimated, i.e. the audio signal is not actually encoded and decoded to obtain the quality measures" (’631 Patent, col. 4:21-23). This language may support a construction where "actually encoding and decoding" refers to the complete, circular process of creating a compressed bitstream and then reconstructing the signal from it.
      • Evidence for a Narrower Interpretation: The detailed description of the invention's own estimation process involves steps like MDCT transformation and LPC shaping (’631 Patent, col. 14:2-8), which are also fundamental steps in an actual TCX encoder. A defendant may argue that since the "estimation" performs some of the same steps as an "actual" encoder, it does not fall "without" that activity, suggesting a narrower scope for the exclusion.

  • For the ’263 Patent:

    • The Term: "state number"
    • Context and Importance: The infringement theory relies on mapping this claim term to the "state" variable used in the EVS standard's algebraic codebook specification. The viability of the infringement claim depends on whether the standard's implementation of a "state" falls within the patent's definition of a "state number."
    • Intrinsic Evidence for Interpretation:
      • Evidence for a Broader Interpretation: The patent's summary and background describe the invention in terms of efficiently encoding pulse information into a single numerical value representing the configuration, stating the pulse positions "can be decoded only based on the state number" (’263 Patent, col. 3:10-12). This may support a broad definition covering any numerical index that uniquely identifies a specific pulse configuration.
      • Evidence for a Narrower Interpretation: Claim 1 recites decoding "by only using one track position number, one total pulse number and one state number." This suggests the "state number" is one of three distinct inputs. If the EVS standard's "state" variable inherently combines or relies on other information in a way that differs from this three-part structure, a defendant could argue for a narrower construction that excludes the accused implementation.

VI. Other Allegations

  • Indirect Infringement: The complaint alleges inducement of infringement for all asserted patents. The factual basis is that Lenovo provides its customers with products (Lenovo EVS Products) and encourages their infringing use by distributing user manuals and other literature that instruct on how to make a phone call (Compl. ¶¶82, 85-87, 122, 125-127, etc.). The complaint provides a screenshot from a "moto g power - 2025" user guide showing instructions to "Make a call" as evidence of this inducement (Compl. ¶87). Using the device as instructed on a supported network allegedly results in direct infringement by the end-user (Compl. ¶88).
  • Willful Infringement: The complaint alleges willful infringement for all asserted patents. The primary basis is alleged actual knowledge of the patents as of October 13, 2023, when Plaintiff sent Lenovo a FRAND package that included a list of the patents (Compl. ¶¶93, 133, 172, etc.). For the ’455 Patent specifically, the complaint also alleges knowledge as early as March 19, 2019, based on a related patent application being cited during the prosecution of one of Lenovo's own Chinese patents (Compl. ¶215).

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

  • A central issue will be one of claim scope versus standard specification: can the specific terminology of the patent claims, such as "without actually encoding and decoding" (’631 Patent) and the three-part "track position number, total pulse number and one state number" input (’263 Patent), be construed to read on the technical implementation details of the 3GPP EVS standard? Resolution of this issue will likely require expert testimony on both the meaning of the patent language and the precise operation of the accused standard.
  • A second core issue will involve FRAND obligations and pre-suit conduct. The case arises from stalled licensing negotiations for patents declared essential to a standard. A key question for the court will be whether the parties negotiated in good faith. This analysis will be critical not only for determining a potential FRAND royalty rate but also for assessing the merits of Plaintiff's willful infringement claim, which is predicated on Defendant’s alleged knowledge and continued infringement after licensing discussions began.