US 5867819

I'll search for authoritative information on this patent number, then summarize.

US 5,867,819 — "Audio decoder"

Search scope note: I searched for this exact number (5,867,819) across patent databases and litigation records. I found no 2026 CAFC docket activity for this patent. Its known litigation is from 2005–2007 (below). The patent expired in 2016, so any 2026 CAFC appeal would be unusual; I could not confirm one, and I will not assume one exists. Where sources disagree by a day on dates, I flag it rather than pick silently.

Bibliographic data (as listed in the patent record)

Assignment chain: Nippon Steel Corp (original, 1996) → United Module Corporation (2002-08-21) → MediaTek Incorporation (2004-08-23) and MediaTek, Inc. (2005-01-21).

Classifications: G10L19/00, G10L19/008, G10L19/0204 (audio coding/decoding, multichannel, subband/transform vocoders); G11B20/10; H03M1/00; H03M7/30; H04N19/00.

Examiners of record (per Unified Patents): Daniel Abebe; David R. Hudspeth.

Abstract (as issued)

"An audio decoder which can reduce a memory circuit capacity necessary for performing a series of decoding processes and can perform a down mixing. The audio decoder decodes audio data of a plurality of channels encoded in a frequency domain by using a time base to frequency base conversion. After a down mixing process was performed to the audio data of the frequency domain by frequency domain down mixing circuit, it is converted into audio data of a time domain by frequency base to time base converting circuit, thereby reducing memories by the number corresponding to the reduced number of channels. Further, by executing an inverse quantizing process of each channel and a frequency base to time base converting process of each channel by pipeline processes, a work buffer can be shared in both of the processes."

Plain-language overview of the specification

The patent addresses a mundane but real engineering problem in multi‑channel audio decoding (e.g., 5‑channel movie audio being played on a 2‑channel home device): the conventional decoder needed about 6400 words of buffer memory. The invention's core idea is to reorder the decoding steps so that channel reduction ("down mixing," bringing 5 channels down to 2) happens earlier — either in the frequency domain before the IMDCT, or immediately after windowing and before the output buffer — so fewer channels flow through the memory-hungry stages. The five disclosed embodiments progressively shrink memory, with a summary Table 1: prior art 6400 words (100%) → 2nd embodiment 4608 (72%) → 3rd 3968 (62%) → 4th 2432 (38%) → 5th 2048 (32%).

Specific techniques include:

• Pipeline processing between the demultiplexer/inverse‑quantizer and the IMDCT so one "work buffer" is reused (fundamentally two buffers, plus a third for "coupling" data).
• Cosine‑symmetry trick: because the IMDCT output is symmetric, only the original ~128 sample values need to be stored in the delay buffer; the mirror values are recomputed.
• Down mixing before the output (PCM) buffer; and later, down mixing before the delay buffer as well.

Independent claims — plain language

Claim 1 (frequency‑domain down mixing): An audio decoder that receives frequency‑domain audio data for multiple channels, includes a frequency‑domain down mixer that mixes those channels into a predetermined (smaller) number of channels at a set level ratio, and a frequency‑base‑to‑time‑base converter (IMDCT) that converts the already down‑mixed frequency data to the time domain. Core point: mix in the frequency domain before the inverse transform.

Claim 8 (shared memory / pipelining): A decoder with an inverse quantizer (identifies variable‑length sample data), a frequency‑base‑to‑time‑base converter, and a memory used by both, where the inverse quantization and the inverse transform are executed as pipeline processes per channel — so one buffer is shared rather than two separate buffers.

Claim 11 (cosine symmetry): A decoder whose IMDCT‑type converter produces first and second sets of time‑domain data in a symmetrical relation (from the cosine function); a buffer stores only one of those sets, and the windowing stage reconstructs the other set from the stored one using the symmetry, then windows. This is the memory‑saving symmetry trick.

Claim 13 (down mix into the output buffer): A decoder with an inverse transform, a windowing stage, and a down mixer that operates on windowed time‑domain data and writes the mixed result into a buffer memory, where the down mixer adds data already held in that buffer to the windowing output to perform the level‑ratio mixing. I.e., the PCM/output buffer holds only the reduced channel count.

Claim 16 (down mix into the delay buffer): Similar to claim 13 but the buffer is the delay buffer used by the windowing stage; the down mixer adds buffered (delayed) data to the windowing output and stores the result back into that same buffer.

Dependent claims of note: Claim 4 (block‑length matching before down mixing when channel block lengths differ); Claim 5 (block‑length matching implemented by an extra inverse‑transform + re‑transform); Claim 6 (drop channels whose block length differs); Claim 7 (conversion uses MDCT); Claim 9 (buffer area for shared "coupling" data); Claim 17 (distributing step separating delayed vs. non‑delayed portions).

Litigation history (known; no confirmed 2026 CAFC activity)

• Sanyo Electric Co., Ltd. v. MediaTek Inc., No. 2:06-cv-01225, C.D. Cal. (2006) — listed in the Stanford NPE Litigation Database with this patent.
• Case No. 6:05-cv-00323, E.D. Tex. — the Unified Patents litigation link for this family. A 2007 claim‑construction order (Document 344, filed 03/26/2007) in that case construes several means‑plus‑function terms of the '819 patent, including "down mixing means," "separating means," and the IMDCT limitations, holding the corresponding structure to be "an IMDCT circuit implementing equation (2), (5), or (6); and equivalents," and a "demultiplexer or demultiplexing circuit" for the separating means. (Source: govinfo.gov PDF for 6:05-cv-00323.)

Uncertainty flags:

1. I found no 2026 CAFC docket entry for 5,867,819; I cannot confirm any 2026 appellate activity, and given expiry and the age of the suits, none is expected.
2. Priority/filing/issue dates differ by one day between the patent record (09‑29 / 09‑27 / 02‑02) and Unified Patents (09‑28 / 09‑26 / 02‑01). I have not resolved which is the authoritative legal date.
3. The full claim text supplied to me was truncated at claim 8; I reconstructed claim 8 and the claim dependencies from the freepatentsonline full‑text listing, so exact claim 8 wording should be verified against the issued patent.

I'll research litigation involving US patent 5867819. Let me start with several searches.

Litigation Involving U.S. Patent No. 5,867,819

Confirmed cases

1. MediaTek, Inc. v. Sanyo Electric Co., Ltd. and Sanyo North America Corporation

• Court / Jurisdiction: U.S. District Court for the Eastern District of Texas (Tyler Division)
• Case No.: 6:05-cv-00323 (also styled 6:05-cv-00323-LED)
• Filed: August 31, 2005 (cause of action: patent infringement; sector: Computer Architecture and Software)
• Plaintiff: MediaTek, Inc. (Taiwan)
• Defendants: Sanyo Electric Co., Ltd. (Japan) and Sanyo North America Corporation (Delaware)
• Patents asserted: U.S. Patent Nos. 5,867,819 ("Audio Decoder" / the '819 patent); 6,118,486 ("Synchronized Multiple Format Video Processing Method and Apparatus"); and 5,751,356 ("Video/Audio Signal Coding System and Method"). The operative pleading was a Second Amended Complaint for Patent Infringement.
• Judge: Leonard E. Davis (Magistrates Keith F. Giblin, Christopher H. Steger, Judith K. Guthrie)
• Key proceeding: The court issued a claim construction order on March 26, 2007, construing the means-plus-function limitations of the '819 patent (e.g., construing the "frequency domain down mixing" function and identifying the corresponding structure as a frequency domain down mixing circuit implementing equation (4); construing the IMDCT/IMDCT-related structures; and the "separating means" as a demultiplexer). The court applied WMS Gaming and Harris Corp. v. Ericsson to hold the patent directed to hardware implementations.
• Outcome / status: Settled — dismissed with prejudice. Although the case was pushed toward trial on an expedited schedule, the parties settled shortly before opening arguments; the settlement also resolved the parallel California case. The parties exchanged cross-licenses under the asserted patents. The docket lists the case as Closed. (Settlement terms otherwise undisclosed.)
• Source: Ex Parte case page https://ai-lab.exparte.com/case/dct/txed/6:05-cv-00323/mediatek-inc-v-sanyo-electric-co-ltd ; court opinions at govinfo https://www.govinfo.gov/content/pkg/USCOURTS-txed-6_05-cv-00323/pdf/USCOURTS-txed-6_05-cv-00323-0.pdf ; RPX Insight litigation document https://insight.rpxcorp.com/litigation_documents/[2377627](/patent/2377627) (Second Amended Complaint).

2. Sanyo Electric Co., Ltd. v. MediaTek Inc.

• Court / Jurisdiction: U.S. District Court for the Central District of California
• Case No.: 2:06-cv-01225
• Plaintiff: Sanyo Electric Co., Ltd.
• Defendants: MediaTek Inc., together with customers ASUSTek Computer, Inc. and ASUS Computer International
• Filed: 2006 (per the Stanford NPE Litigation Database, which lists the venue as C.D. Cal.)
• Outcome / status: Settled — dismissed with prejudice as part of the same global settlement resolving the Texas action, with cross-licenses exchanged.
• Sources: Stanford NPE Litigation Database https://npe.law.stanford.edu/patent/5867819 ; settlement report (CDRinfo) https://cdrinfo.com/d7/content/sanyo-and-mediatek-reach-settlement-patent-infringement-litigation

3. In re MediaTek (related appellate/mandamus proceeding)

• Court: U.S. Court of Appeals for the Federal Circuit
• Case No.: 07-M844
• Petitioner: MediaTek, Inc.; Respondent: Sanyo Electric Co., Ltd.
• Filed: February 7, 2007; opinion issued April 3, 2007 (nature of suit: "Other"; handled by counsel Otteson, James C. and Hentschel, Craig N.)
• Relevance: This appears to be a Federal Circuit petition arising out of the Sanyo/MediaTek dispute in which the '819 patent was asserted (the Texas case 6:05-cv-00323 was before Judge Davis at that time).
• Source: https://dockets.justia.com/docket/circuit-courts/cafc/07-M844

Related/adjacent filing (not confirmed to involve the '819 patent)

The RPX litigation document also references MediaTek, Inc. v. Matsushita Electric Industrial Co., Ltd. et al., 2:05-cv-00484 (E.D. Tex.) as a contemporaneous MediaTek enforcement action. I did not confirm that the '819 patent was among the patents asserted in that case, so I do not list it as '819 litigation.

Notes and caveats

• The Google Patents page for US5867819 itself links two U.S. litigation entries: the C.D. Cal. case (2:06-cv-01225) and the E.D. Tex. case (6:05-cv-00323), and notes "Family has litigation," pointing to the Darts-IP global patent litigation dataset (family 26359282).
• The patent expired (anticipated expiration 2016-09-27) and is now Expired – Lifetime. Ownership passed Nippon Steel Corp. → United Media Corporation → MediaTek. MediaTek was the plaintiff/practitioner asserting the patent (and later the assignee), so this was commercial-company-versus-commercial-company litigation rather than standard NPE activity.
• I did not find any other district court or ITC actions specifically asserting U.S. Patent No. 5,867,819. If further cases exist, they are not surfaced by the sources searched (Stanford NPE Database, RPX/Insight, Ex Parte docket mirror, govinfo, and news coverage of the Sanyo–MediaTek settlement).
• Regarding identifiers I did not auto-correct: the case numbers, patent numbers, and docket numbers are reproduced exactly as found. Note that secondary sources list slightly different date conventions (e.g., "filing date" vs. "application date"), which I have preserved.

Bottom line: The only known litigation over U.S. Patent No. 5,867,819 is the 2005–2007 Sanyo–MediaTek dispute — the Texas case (MediaTek v. Sanyo, 6:05-cv-00323) and the California case (Sanyo v. MediaTek, 2:06-cv-01225) — which produced a March 2007 claim construction ruling in Texas and ultimately settled with dismissal with prejudice and cross-licenses.

I'll start by verifying the canonical "no proceedings" baseline against public sources, since the structured block says ODP has no AIA trials indexed.

No AIA trial hits so far. Let me run targeted checks for any AIA proceeding, reexamination, and the litigation/appeal history tied to this patent.

Let me check the PTAB/Unified litigation tracker and any ex parte reexamination history, plus the CAFC appeal chain.

Proceedings overview

There are zero AIA trial proceedings on file for US 5,867,819. The structured "PTAB proceedings on file" block (USPTO Open Data Portal) returns no IPR, PGR, or CBM for this patent, and my independent web checks surfaced none either — no petition, no institution decision, no Final Written Decision, and no PTAB appeal. Bottom line for a defendant today: the patent is completely untested at the PTAB — all 18 claims are live-in-the-record and un-cancelled, so there is no claims-canceled shortcut here. Any challenge you want to make, you will have to file; you get the benefit of a blank estoppel slate (see below).

Because the answer is an absence rather than a set of decisions, I'm substituting a "no-proceedings" entry plus the relevant non-AIA litigation backdrop (which is what the searches actually returned) in place of the per-proceeding template.

(none) — No AIA trial proceeding ever filed against US 5,867,819

• Type: N/A — no IPR / PGR / CBM petition on file.
• Filed: N/A.
• Status: No PTAB proceeding exists (canonical ODP result: none).
• Judge panel: N/A.
• Petition grounds: N/A.
• Institution decision: N/A.
• Final Written Decision: N/A — no claim of 5,867,819 has ever been adjudicated at the PTAB.
• Settlement / termination: N/A.
• Appeal: N/A at the PTAB. (The CAFC activity below is a district-court appeal, not a PTAB appeal.)
• Defensive value: Neutral-to-adverse for a defendant relying on an IPR head start. There is no FWD to lean on, no canceled claim to cite, and no § 315(e)(2) estoppel binding anyone. If you intend to invalidate, the burden of building and filing the petition is entirely yours, and the clock is against you — the patent expired 2016-09-27, so the damages tail is limited to pre-expiration conduct, which materially reduces the ROI of an IPR.

Timing context for why no AIA trial exists

AIA trials only became available 2012-09-16. This patent's enforcement history predates that:

• MediaTek, Inc. v. Sanyo Electric Co., Ltd., No. 6:05-cv-00323 (E.D. Tex., filed 2005; Judge Leonard Davis) — MediaTek asserted the '819 patent (with U.S. 6,118,486 and a '356 patent) against Sanyo's DVD recorders, DVD players, digital TVs, set-top boxes and the Model DRW-1000, alleging literal infringement / DOE and willfulness. Source: Second Amended Complaint, 6:05-cv-00323.
• Sanyo Electric Co. Ltd. v. MediaTek Inc., No. 2:06-cv-01225 (C.D. Cal.; marked CV 05-2580 RSWL in a related '801 ruling) — asserted U.S. 5,818,801 (a different patent); the court denied summary judgment of non-infringement on 2006-08-17. Source: Markman/order PDF.
• Claim construction: The E.D. Tex. court issued a Markman order on 2007-03-26 construing means-plus-function terms (e.g., "video encoding means" limited to structure using an MPEG standard; "data number information" = a value indicative of bit length of a picture). Source: govinfo PDF, 6:05-cv-00323.
• Federal Circuit: The claim-construction dispute was appealed; the CAFC opinion (addressing at least the '356 patent terms, § 112 ¶ 6) is posted as MediaTek, Inc. v. Sanyo Electric Co. Ltd. Source: CourtListener. I could not confirm the precise CAFC docket number or disposition date from the available results, so I am not stating one — treat that as unverified.

None of the above are AIA trials; they are district-court and appellate proceedings from the 2005–2008 window, before IPR/PGR/CBM existed.

Strategic summary

Claim status: all 18 claims are UNTESTED at the PTAB. There is no claim of 5,867,819 that has been canceled, held unpatentable, or confirmed in any AIA proceeding. Based on the printed claims, the independent claims are 1, 8, 11, 13 and 16, with dependent claims 2–7, 9–10, 12, 14–15, and 17–18. Every one of these remains in force in the record as issued — the only narrowing that ever occurred came from district-court claim construction (means-plus-function treatment of the "means" limitations), not from the Patent Office. Do not let anyone tell you a claim is "dead" — it isn't.

Estoppel landscape: a clean slate. Because no IPR/PGR/CBM was ever filed and reached an FWD, § 315(e)(2) estoppel binds no one. A defendant today is free to run any § 102/§ 103 ground, any art, and also § 101 and § 112 defenses that an IPR could never have carried. There is no prior petitioner whose grounds are foreclosed and no "reasonably could have raised" shadow to worry about. The flip side: there is also no petitioner IPR record, no PTAB claim construction, and no institutional finding to use as leverage for a district-court stay.

Pattern signals. No repeat-petitioner pattern exists because there is no petitioner at all. The patent's asserted-life story is a pre-AIA two-way litigation between MediaTek (patent owner by assignment from Nippon Steel → United Module → MediaTek) and Sanyo, not a PTAB story. There is no defensive-aggregator IPR in the chain — the Unified Patents and Darts-ip links in the structured block are litigation-database citations, not evidence that Unified Patents filed an AIA challenge. And the ownership trail (Nippon Steel → United Module Corp. 2002-08-21 → MediaTek 2004-08-23/2005-01-21) placed the patent in the hands of a product company (MediaTek), not a classic NPE — so the lack of IPR is unsurprising.

Recommended next steps

• If you are a defendant and you expected canceled claims: there are none. Nothing in the PTAB record helps you. The patent expired 2016-09-27 (per the authoritative ODP data; note the Unified Patents portal lists 2016-09-26 and 1995-09-28 — a one-day discrepancy, so verify the exact expiration and priority dates against the USPTO PatentCenter before relying on them for a damages cutoff).
• Because the patent is expired, calibrate whether to file at all. An IPR would still be paper-possible against an expired patent, but with no injunctive exposure and a shrinking pre-expiration damages window, the economics rarely justify it. The more efficient posture is usually straight district-court invalidity (all grounds open, no estoppel) plus non-infringement on the construed "means" limitations, which are the terms the E.D. Tex./CAFC litigation actually litigated.
• If you are a patent owner or licensing plaintiff: the clean PTAB record cuts both ways. You can truthfully say no claim has ever been canceled — but you also have no defensive IPR win to point to, and a defendant remains free to file the first IPR or to run the full invalidity case in court. Budget for a first-ever challenge rather than assuming the patent is "hardened."
• Verify before filing. Re-confirm the zero-proceeding status at: USPTO Patent Trial and Appeal Board E2E and the PTAB Decisions page, and pull the CAFC docket for MediaTek v. Sanyo at CourtListener to pin the exact appeal number and disposition before you cite it.

Caveat on completeness: the initial searches surfaced no AIA trial, and the ODP block confirms none, but my verification was cut short before I could exhaustively query the PTAB API and every reexamination record. If you need a belt-and-suspenders confirmation — including whether any ex parte/inter partes reexamination was ever requested (a pre-AIA avenue that would not appear in an AIA-trial query) — run a PatentCenter "Transaction History" and a reexamination-certificate check on 5,867,819 before you rely on the "no proceedings" conclusion in a filing.

I'll research the assignment record for this patent, starting with several targeted searches.

Let me dig into the specific assignment records, correspondents, and entity details.

Assignment / Ownership Record — US 5,867,819 ("Audio decoder")

Data-source caveat (read first). The recorded event data below is grounded in the reassignment entries reproduced on the Google Patents legal‑events record for US 5,867,819 (https://patents.google.com/patent/US5867819/en) and corroborated by the Unified Patents and Stanford NPE records. Two fields the task specifically asks for — the reel/frame number and the recorded correspondent of record — were NOT retrievable within my search budget. I will not invent them. The USPTO Assignment Center search returned no exposed reel/frame or correspondent strings for this patent in the material I could reach, so those cells are marked [not retrieved] and flagged wherever they would have mattered. This is a genuine gap, not a "no records" finding — four conveyances are on record.

Inventors

Unusual‑pattern check. Both inventors are named jointly on the same first assignment to the original corporate assignee, executed on the filing date (1996‑09‑27). That is the ordinary "employees assign to employer at filing" pattern and does not show the "all inventors depart the assignee within 12 months" fire‑sale tell. I found no evidence of either inventor's departure, nor any later assignment executed by an inventor (as opposed to by a corporate assignee). Could not determine whether either inventor left Nippon Steel; insufficient data.

Original assignee

• Entity on the issued patent: Nippon Steel Corporation (Japan).
• Primary line of business: integrated steel manufacturing. The '819 subject matter (a multi‑channel MDCT audio decoder) is non‑core to steel production — it is the product of Nippon Steel's electronics/advanced‑technology research activity, not its steel line. This is a classic "non‑core patent held by a non‑electronics operating company" fact pattern, which is why the asset later moved.
• Did the original assignee ship a product embodying the claims? Unclear. Nothing in the record shows Nippon Steel selling an audio decoder or MDCT codec chip. No product evidence either way in the sources reached.
• Current status: Operating. Nippon Steel Corporation remains an active company (via the Nippon Steel & Sumitomo Metal merger lineage; renamed back to Nippon Steel Corporation in 2019). It did not enter bankruptcy and there is no Kodak/Nortel‑style patent fire‑sale proceeding attached to this asset.

Assignment timeline

Four recorded conveyances. All are "ASSIGNMENT OF ASSIGNORS' INTEREST" (i.e., outright assignment, not security interest, license, merger, or change of name). Reel/frame and correspondent are [not retrieved] for every entry; execution dates are the recording/event dates shown by Google Patents, and I flag the one‑day date drift consistent with the earlier sections.

• 1996‑09‑27 (executed) / recorded 1996‑09‑27 — Reel [not retrieved]

  • Conveyance: Assignment of assignors' interest
  • Assignor: Hiroyuki Fukuchi; Hirofumi Sato (individual inventors)
  • Assignee: Nippon Steel Corporation
  • Correspondent: [not retrieved] — cannot run the repeat‑correspondent check
  • Context: Ordinary inventor‑to‑employer assignment executed at filing; original prosecution/ownership record.

• 2002‑08‑21 (executed) / recorded 2002‑08‑21 — Reel [not retrieved]

  • Conveyance: Assignment of assignors' interest
  • Assignor: Nippon Steel Corporation
  • Assignee: United Module Corporation
  • Correspondent: [not retrieved] — this is the entry where a recurring NPE‑recording attorney would be the key tell, and it is the one I could not capture
  • Context: Divestiture of a non‑core asset by a steel company to a patent‑acquisition/holding entity. This is the single most analytically significant link (see signals 1 and 4).

• 2004‑08‑23 (executed) / recorded 2004‑08‑23 — Reel [not retrieved]

  • Conveyance: Assignment of assignors' interest
  • Assignor: United Module Corporation
  • Assignee: MediaTek Incorporation
  • Correspondent: [not retrieved]
  • Context: Transfer to operating acquirer (MediaTek) — first of two records naming a MediaTek entity as assignee from the same assignor.

• 2005‑01‑21 (executed) / recorded 2005‑01‑21 — Reel [not retrieved]

  • Conveyance: Assignment of assignors' interest
  • Assignor: United Module Corporation
  • Assignee: MediaTek, Inc.
  • Correspondent: [not retrieved]
  • Context: Second United‑Module‑to‑MediaTek record, directed to the slightly different legal name "MediaTek, Inc." (vs. "MediaTek Incorporation" in 2004). This duplicate‑looking pair is the likely source of the chain‑of‑title challenge discussed below.

Note on the missing reel/frame. Because the two 2004/2005 entries run from the same assignor to two differently styled MediaTek names, the reel/frame and correspondent fields would normally be the deciding evidence of whether (a) these are two records of one transaction, (b) a corrective/confirmatory recording, or (c) genuinely separate conveyances. I could not retrieve them, so I decline to characterize the relationship beyond what the names show.

Timeline diagram

timeline
    title Ownership of US 5867819
    1995 : Priority date 1995-09-29
    1996 : Filed by Nippon Steel Corp
         : Inventors assign to Nippon Steel
    1999 : Patent issued 1999-02-02
    2002 : Nippon Steel to United Module Corp
    2004 : United Module to MediaTek Incorporation
    2005 : United Module to MediaTek Inc
         : MediaTek sues Sanyo in E D Tex
    2006 : Sanyo sues MediaTek in C D Cal
    2007 : Texas case settled
         : Standing challenge denied
    2016 : Patent expired

NPE / troll-pattern signals

1. Shell‑entity transfer — Unclear. The 2002‑08‑21 transfer moved the patent from an operating steel company (Nippon Steel) to United Module Corporation, an entity whose name suggests a holding/module business rather than a disclosed product line. That is the shape of a shell transfer. But per the task's own rule, naming alone is not proof: I have no evidence of United Module's product absence, registered‑agent address, or single‑member LLC status, and the recording correspondent that would normally confirm this is [not retrieved]. Flag as unclear, not established.

2. Known asserter in the chain — Not present. No assignee in the chain matches the enumerated NPE lists (Acacia, Marathon, IV, IPNav, Wi‑LAN, Conversant/Mosaid, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, DGC, Spangenberg entities). The end‑point asserter, MediaTek, is a product company — Stanford's NPE Litigation Database categorizes the '819 asserter as "8 Product company" (https://npe.law.stanford.edu/patent/[5867819](/patent/5867819)). No NPE name appears anywhere in the four‑link chain.

3. Repeat correspondent across the chain — Unclear / unverified. This is the signal I most wanted to run and could not: the correspondent of record is [not retrieved] on all four entries. There is no basis to claim a recurring recording attorney, and I will not assert one. If you can pull the correspondent strings from the Assignment Center, this is the check to run — a single attorney appearing on both the 2002 Nippon‑Steel‑to‑United‑Module recording and the 2004/2005 United‑Module‑to‑MediaTek recordings would be the highest‑value finding available on this patent.

4. Cascading transfers — Partially present, but not through chained LLCs. Three transfers occur in roughly three years (2002‑08‑21 → 2004‑08‑23 → 2005‑01‑21), two of them from the same assignor (United Module) to two differently named MediaTek entities within ~17 months. That is a cluster. However, it is not the classic "chained unrelated LLCs sharing a correspondent" pattern — the 2004 and 2005 assignees are the same corporate family (MediaTek), which reads more like a corrective/confirmatory recording or a chain‑of‑title cleanup than a cascading shell relay. Mixed: date‑cluster present; shell‑relay character not established.

5. Pre‑litigation transfer — Unclear / borderline. The 2005‑01‑21 recording to "MediaTek, Inc." precedes MediaTek's first suit naming the '819 patent (MediaTek v. Sanyo, 6:05‑cv‑00323, E.D. Tex., filed 2005‑08‑31) by ~7.3 months — just outside the 6‑month window the template uses. I do not know the execution date of the 2005 assignment (only the 2005‑01‑21 date as recorded); if it was executed later in 2004 or in 2005, the interval tightens or loosens accordingly. Cannot confirm a within‑6‑month strategic transfer. Notably, though, the chain‑of‑title did matter: in the related MediaTek enforcement campaign, Matsushita moved to dismiss for lack of standing, and the court denied the motion, affirming MediaTek held the patents "through an assignment agreement with United Module Corp." (Law360, 2007‑06‑13, https://www.law360.com/texas/articles/26830/judge-denies-matsushita-s-motion-to-dismiss). That confirms the assignment record was load‑bearing for standing — but it does not establish a pre‑litigation shell transfer.

6. Bankruptcy fire‑sale — Not present. Nippon Steel (the original assignee) did not file Chapter 7/11, and I found no Kodak/Nortel/Polaroid‑style sale proceeding involving this asset. The 2002 transfer to United Module reads as a negotiated divestiture of non‑core IP, not a bankruptcy liquidation.

7. Privateering — Not present. The patent was ultimately asserted by MediaTek itself, an operating company, against actual competitors (Sanyo; Matsushita). That is direct operating‑company assertion, not an operating company transferring to an NPE to assert on its behalf. The Nippon‑Steel‑to‑United‑Module step is a corporate divestiture, not the "operating company feeds an NPE against rivals" privateering pattern. No SEC/Patent Progress/EFF privateering coverage surfaced.

8. Defensive aggregator (anti‑NPE) — Not present. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. It terminates at an asserting operating company (MediaTek). The Unified Patents and Darts‑IP links on the Google Patents page are litigation‑database citations, not evidence of a defensive aggregation purchase.

Cross‑reference note (not a recorded assignment): Unified Patents lists the parent company for this asset as "Mediatek / HFI Innovation" (https://portal.unifiedpatents.com/patents/patent/US-5867819-A). "HFI Innovation" is the style of a MediaTek‑affiliated IP‑holding entity. I could not confirm a separate recorded assignment to HFI Innovation in the reel/frame data, so I treat this as an ownership‑structure cross‑reference, not a fifth assignment link. Verify before relying on it.

Verdict

Operating‑company assertion.

The chain runs Nippon Steel Corporation (2002‑08‑21) → United Module Corporation (2004‑08‑23 / 2005‑01‑21) → MediaTek, and the entity that asserted the patent is MediaTek — a fabless semiconductor product company that ships chips and decoders and sued actual competitors (Sanyo; Matsushita), categorized by Stanford as a "Product company" asserter. None of the enumerated NPE assignees appears, the patent did not end at a defensive aggregator, and there is no evidence of a bankruptcy fire‑sale or SEC‑documented privateering. The only residual NPE‑flavored features are (i) the intermediate United Module Corporation, whose status is unclear because I could not retrieve its products, address, or the recording correspondent, and (ii) the 2002→2004→2005 transfer cluster, which looks more like a MediaTek chain‑of‑title cleanup (one that was actually litigated via a denied standing motion) than a shell relay. On the evidence available, this is a product‑company‑versus‑product‑company enforcement chain, not a troll pattern.

Verify at:

• USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search "5867819")
• USPTO Patent Assignment Search (legacy) — https://assignment.uspto.gov/patent/index.html
• Google Patents legal events — https://patents.google.com/patent/US5867819/en

Residual gap to close: the reel/frame numbers and the correspondent of record for the four recordings. If you supply those, run signal 3 (repeat correspondent) first — it is the one live NPE tell on this patent and the only one I could not evaluate.

I'll research the prior art cited against US 5,867,819, starting with the patent's own citation records.

Good — Unified Patents lists 8 prior-art references and 2 NPL items. Let me pull the specific citation lists.

I have the citation counts but need the actual reference list. Let me target the citation tables directly.

Important limitation up front

I was unable to retrieve the actual front-page "References Cited" list for US 5,867,819. The USPTO/Google Patents full text supplied to me is truncated at claim 8 and stops before the citation tables. My searches confirmed that the patent has citations, but not which references they are:

• Unified Patents' record for US-5867819-A shows: "Patent Art (8)" (8 patent references) and "Non-Patent Literature (2)", plus "Referenced By (74)." — https://portal.unifiedpatents.com/patents/patent/US-5867819-A
• Google Patents shows the same document with cited/citing tables, but the fetched HTML does not include the reference rows.

Per your operating rules, I will not fabricate the identity of those 8 patent references or 2 NPL items. Below I separate (A) what is verified, (B) the only references I can ground in the patent's own text, and (C) an analytic prior-art assessment with explicit confidence flags. To close the gap you would need the USPTO PatentCenter/Patent Public Search "References Cited" (front-page) list or the EP/WO equivalent of the same family (family 26359282).

A. Verified identifiers (no auto-correction)

B. References I can ground in the patent's own text

These are the only two prior-art items I can name with direct textual support, and both are cited in the specification prose (which is likely why Unified Patents counts 2 NPL entries):

1. Non-patent literature — book chapter

• Citation: Advances in Speech Signal Processing, edited by Furui & Sondhi, Marcel Dekker, Inc. (New York), pages 109–140, 1991.
• Relevance: The spec cites it for "outlines of the sub-band filter encoding method and MDCT encoding method." It is background to the time-base-to-frequency-base conversion (MDCT) that underlies claims 1 and 7. It is general background, not an anticipation reference for the claimed decoder architecture.

2. Non-patent literature — standard

• Citation: ISO/IEC 11172-3 (MPEG Audio Layer III), referenced by the spec as the example of "an MDCT encoding method of variable block length."
• Relevance: Cited for the variable conversion block length concept that claim 3 and dependent claims 4–6 build on. Again background/§ 103-type art, not a decoder anticipation reference on its own.

Flag: I cannot confirm these two are among the "2 NPL" counted by Unified Patents — they are simply the only references the patent text itself names.

C. Most relevant prior-art assessment (analytic, with confidence flags)

Because I could not obtain the cited-reference list, the following identifies the technical classes of prior art most relevant to each independent claim, plus the concrete documents I can point to. Treat § 102 attribution as hypothesis to verify, not established fact.

Against Claim 1 (frequency-domain down mixer + IMDCT)

The novel point is doing channel down mixing in the frequency domain before the inverse transform. The most relevant art class is frequency-domain (spectral) downmixing in multi-channel audio decoders. A highly relevant document I can cite directly in this space is:

• US 6,205,430 B1, "Audio decoder with an adaptive frequency domain downmixer." The patent's own background section (retrieved via EveryPatent, http://www.everypatent.com/comp/pat6205430.html) states that frequency-domain downmixing was previously disclosed in U.S. Pat. No. 5,400,433 and in Steve Vernon, "Design and Implementation of AC-3 Coders," IEEE Trans. Consumer Electronics, vol. 41, no. 3, Aug. 1995, pp. 754–759. That background further states downmixing in the frequency domain was known "but only in the case where block switching is not used."
• § 102 relevance: US 5,400,433 and the Vernon AC-3 paper are strong candidate § 102/§ 103 art for claim 1 if they show frequency-domain downmixing before the inverse transform. The '819 patent's asserted distinction is the combination with variable block length / block-length matching (claims 3–6), which the '430 background expressly acknowledges was not handled by the earlier art. Confidence: high that this is the right art family; unverified that any single one anticipates claim 1.

Against Claim 8 (shared memory / pipelining between inverse quantizer and IMDCT)

Relevant art is pipelined/memory-shared transform decoding in audio codecs — i.e., reusing one work buffer across inverse quantization and the inverse transform. MPEG-1 Audio (ISO/IEC 11172-3) decoder implementations are the natural background. Confidence: the claim is a memory-architecture claim; I could not identify a specific cited reference.

Against Claims 11 and 16 (cosine symmetry in the IMDCT output; delay buffer reduction)

Relevant art is the symmetry/aliasing (TDAC) properties of the MDCT/IMDCT used to reduce storage. The MDCT overlap-add TDAC symmetry is textbook material (the Princon–Bradley–Noll work underlying AC-3 and MPEG). Confidence: the underlying principle is well known; whether the specific "store only one of the two symmetric sets" limitation was previously disclosed requires the cited-reference list.

Against Claims 13 and 16 (down mixing into the PCM output buffer / the delay buffer)

Relevant art is decoder buffer architectures that reduce memory by mixing before the output stage. The patent's own Table 1 quantifies the savings (6400 → 4608 → 3968 → 2432 → 2048 words). Confidence: no specific cited reference identified.

D. Cross-check against the earlier sections (consistency)

• The claim set matches: independent claims are 1, 8, 11, 13, 16, with dependents 4, 5, 6, 7, 9, 17 identified earlier. The full claim text now retrieved from FreePatentsOnline confirms the wording of claims 9–18 (e.g., claim 9 = coupling-data area in memory; claim 13/16 = down mixer adds buffered data to windowing output). This upgrades the earlier caveat that claim 8's exact wording was reconstructed — the FPO text now corroborates it.
• The litigation/claim-construction content in the earlier section stands; it is not prior art and I did not repeat it.
• No contradiction found between the earlier sections and the authoritative patent text, other than the unresolved one-day date discrepancy, which I flag again rather than resolve.

E. What I recommend to complete this task

To produce a defensible reference-by-reference § 102 mapping, you need the front-page citation list, obtainable from any of:

1. USPTO Patent Public Search / PatentCenter for 5,867,819 → "References Cited."
2. The WO/EP family member (family 26359282 per the Darts-IP link) international search report, which typically lists the same X/Y references.
3. The Espacenet "Cited documents" tab for the same family.

Bottom line: I verified that US 5,867,819 carries 8 patent citations and 2 NPL citations, and I identified the two NPL items the specification itself discusses (Furui & Sondhi 1991; ISO/IEC 11172-3) as the likely NPL entries. I was not able to retrieve the 8 patent references within my search budget, so I have declined to name them rather than guess. The analytically most relevant prior-art families are frequency-domain downmixing (e.g., US 5,400,433; Vernon 1995 AC-3 paper) for claim 1, pipelined/shared-memory decoding for claim 8, IMDCT symmetry for claims 11/16, and pre-output-stage mixing for claims 13/16 — each flagged as unverified against the patent's actual citation list.

I'll try to close the biggest gap from the prior-art section — the actual front-page "References Cited" list — and verify the candidate art families before building the § 103 analysis.

The front-page citation list is still not directly retrievable, but I found something more valuable: US 5,400,433 (Dolby) — which expressly discloses frequency-domain downmixing before inverse synthesis. Let me verify its date/scope and look for the EP/WO family search report for the '819 family, plus pre-1995 memory-reduction art.

Obviousness Analysis — U.S. Patent No. 5,867,819 under 35 U.S.C. § 103

Building on the prior-art section. This analysis assumes the verified facts and the "unresolved gap" identified earlier: the patent record shows 8 patent references and 2 NPL items on the front page, but the citation rows themselves were never retrieved. Rather than fabricate that list, I rely on (a) the two NPL items the specification itself names, (b) one reference I newly grounded in this pass — US 5,400,433 (Dolby), whose abstract and claim 1 I retrieved directly — and (c) the art family identified in the earlier section (frequency-domain downmixing, e.g. Vernon's AC-3 paper). I flag below exactly which items are strict § 102 prior art and which are merely corroborative of the state of the art.

1. Governing law and framework

• Statute: Pre-AIA 35 U.S.C. § 103(a) applies. The '819 application was filed 1996-09-27 with a 1995-09-29 priority date — both well before the AIA's 2013-03-16 transition. Obviousness is measured as of the effective filing date (1995) from the perspective of a person having ordinary skill in the art ("PHOSITA").
• Test: Graham v. John Deere Co., 383 U.S. 1 (1966) — scope/content of the prior art; differences between the prior art and the claims; level of ordinary skill; secondary considerations. Combined with KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) — the combination need not be taught expressly; it is enough that the improvement is within the ordinary creativity of a skilled artisan, that the elements are known and combined according to known methods, or that the solution is one of a finite number of identified, predictable options.
• Articulated reasonings available (MPEP § 2143/2144): (A) combining known elements according to known methods yielding predictable results; (C) use of a known technique to improve a similar device in the same way; (F) obvious to try; (G) design incentives / market forces; (H) "obvious to try" in a field of finite solutions.

2. Level of ordinary skill in the art (PHOSITA)

A PHOSITA here is an engineer with a bachelor's degree in electrical engineering (or equivalent) and 2–4 years of experience in perceptual audio coding — specifically MDCT-based codecs (MPEG-1/2 Audio Layers I–III, Dolby AC-3), digital filterbanks, and DSP/memory-architecture design for consumer audio hardware. By 1995 this skill set was mature: ISO/IEC 11172-3 (1993) and ATSC A/52 were published standards, and the TDAC/MDCT design literature dated to the mid-1980s.

3. References relied upon (with prior-art status)

Honesty flag carried forward: I could not retrieve the 8 patent references / 2 NPL items actually listed on the '819 front page, and I could not confirm the priority date of US 6,931,291 ("frequency-domain downmixing with block-switch forcing") — my search budget expired before that result returned. US 6,931,291 and EP 2 360 683 are therefore treated as corroborative only, not as citable art.

4. The key structural insight that drives the whole § 103 case

The independent claims divide into two distinct inventive themes, and they have very different obviousness postures:

1. Reordering (claims 1, 13, 16): move the admitted-prior-art downmix (equation (3); FIG. 4 down mixing circuit 36) earlier in the decoder chain so fewer channels traverse the memory-heavy stages. This is a pure ordering/architecture optimization with a predictable, quantified result (Table 1: 6400 → 4608 → 3968 → 2432 → 2048 words).
2. Storage reduction (claims 8, 11): (a) share one work buffer across dequantization and IMDCT by pipelining; (b) exploit the inherent cosine symmetry of the IMDCT to store only half the samples.

Both themes are textbook design-incentive / known-technique-improving-a-similar-device cases under KSR.

5. Claim-by-claim § 103 analysis

Claim 1 — frequency-domain downmix means + frequency-base-to-time-base converting means

Mapping to R1 (US 5,400,433):

Conclusion: Claim 1 is anticipated by R1 in substance, or at minimum obvious over R1 alone. R1 not only discloses the same ordering (downmix in the frequency domain before synthesis) but states the exact design rationale the '819 patent offers as its advantage — implementation cost scaling with presentation-channel count rather than delivery-channel count is "highly desirable." If narrowed scope is argued (R1's "distribution means" vs. a "down mixing circuit implementing equation (4)"), R4 supplies the additional express teaching of frequency-domain downmixing using linear mixing coefficients, and R6/AC-3 the "predetermined level ratio" coefficients. Motivation: express — reduced implementation cost; reduction of channels before the computationally/memory-expensive synthesis stage. Confidence: high.

Claim 2 — inverse quantizing means for variable-bit-length sample data

R2 (MPEG-1 Layer III) discloses variable-length (Huffman) coded sample data and the requantization ("inverse quantization") stage that recovers each sample's value from side information; R1 discloses deformatting means that "obtains a linear representation using a method inverse to that used by a companion encoder." Combination is compelled: one cannot combine spectral information in the frequency domain (R1) until the channel data has first been reconstructed to linear frequency-domain samples (R2). Motivation: necessity/operability. Confidence: high.

Claim 3 — variable-length data blocks + block-length information + separating means

Directly met by R2: MPEG-1 Layer III transmits block-type/side information indicating the conversion block length (long vs. short), and every decoder necessarily has a demultiplexer that separates side information from the main data. The E.D. Tex. court construed the "separating means" as a "demultiplexer or demultiplexing circuit" — precisely the R2 structure. Confidence: high.

Claims 4, 5, 6 — block-length matching, the real point of novelty

This is the only place the claims arguably depart from R1/R2, and it is where the § 103 case must be made most carefully.

• Claim 4 (make block lengths coincide before down mixing). Converting block lengths to a common basis before summing frequency-domain data is not an optional nicety but a mathematical prerequisite: spectral coefficients from blocks of different lengths are not addable. A PHOSITA downmixing in the frequency domain must either match lengths or exclude mismatched channels. This is a finite set of two identified solutions (KSR). The motivation is therefore inherent in adopting R1's frequency-domain downmixing technique. Confidence: moderate-to-high.
• Claim 5 (match by IMDCT of the short block → re-MDCT to the long block). This is the "use of a known technique to improve a similar device in the same way" rationale. Both halves — the IMDCT and the forward MDCT — are admitted/conventional (R2, R3, R5). Selecting a transform-domain conversion to reconcile differing resolutions is a routine signal-processing maneuver, and the '819 spec does not assert any unexpected result — only that "the conversion block lengths of all channels are equalized." That is the predictable consequence of the chosen technique. Obvious-to-try under KSR. Confidence: moderate (weakest on a "teaching away" or "unexpected result" counter-argument, but no such teaching appears in the record).
• Claim 6 (instead, eliminate channels whose block length differs). The alternative branch of the same two-solution choice, and expressly the simpler one. Dropping a channel from the downmix is the classic "known workaround" with a known trade-off (minor sound-quality loss, admitted in the spec as "enough sound quality … can be obtained"). A predictable trade-off selected for simplicity is obvious. Confidence: high.

Claim 7 — conversion by MDCT

Met verbatim by R2 (MPEG-1 Layer III uses the MDCT) and R1 (TDAC transform), and named in the '819 specification as the exemplary transform. Confidence: high.

Claim 8 — shared memory + pipeline processes between inverse quantizer and IMDCT

Mapping: an inverse quantizer (R2 requantization), an IMDCT (R2/R1 synthesis), and a memory used by both, executing the two processes as pipeline processes per channel.

• Known elements: both processing stages (R2) and the use of a shared work RAM in fixed-function audio decoders are conventional.
• Known technique / design incentive: the '819 specification itself identifies the motivation ("a fairly large memory capacity is necessary" is a "problem"), and the solution — pipelining to reuse a buffer — is the standard DSP time/space trade. Under MPEP § 2144(G) (design incentives/market forces) and KSR, reducing buffer memory in a consumer audio decoder is a recognized design objective.
• Predictable result: the reduction from (256×10) to (256×3) words is a direct arithmetic consequence of channel-serial processing.

Caveat / lower confidence: I identified no specific pre-1995 reference that discloses pipelining the dequantizer with the IMDCT over a shared, channel-serial buffer. R2 shows the two stages but not necessarily their pipelined sharing. This is the claim in the set where a § 103 challenge is weakest, and where the patent's own specification (the "pipeline processes … the same memory is commonly used" passage) reads as the point of intended novelty. Confidence: moderate; flag for a fuller art search (e.g., MPEG decoder VLSI papers 1993–1995, and pre-1995 DSP architecture patents).

Claim 9 — memory area storing shared "coupling" data

Coupling is admitted prior art in the specification itself (Dolby AC-3's coupling method, R6, and the '819 spec's own description of the coupling method as known). Allocating a buffer area for data the standard already defines as shared is a routine design step. Confidence: high.

Claim 11 — IMDCT cosine symmetry; store only one of the symmetric sets; windowing reconstructs the other

This is the cleanest § 103 case on the storage side.

• The symmetry is inherent, not invented. R5 (Princen & Bradley, TDAC, 1986) establishes that the MDCT/IMDCT output is symmetric about the block center — i.e., "a first set and a second set of time domain audio data having a symmetrical relation … derived from said cosine function" is a property of the transform, not an inventive feature. R2's IMDCT exhibits it; R1's decoder cites R5 and uses TDAC.
• Motivation: store half the samples → halve the delay buffer. That is precisely the memory-reduction objective stated throughout the '819 specification.
• KSR rationale: "use of a known technique (exploiting TDAC symmetry) to improve a similar device (an MDCT-based multichannel decoder) in the same way."
• Reasonable expectation of success: absolute — the symmetry is exact, so reconstruction is lossless by definition.

Confidence: high. The only residual argument is claim-drafting scope (whether R5 "stores only one set"), but § 103 does not require a single reference to disclose the storage economy; it requires the claimed combination to have been obvious, which it plainly was.

Claim 13 — downmix after windowing, writing back into the PCM buffer

• The downmix itself is admitted prior art: equation (3) and the FIG. 4/FIG. 6 "down mixing circuit 36" operating on PCM data read from the PCM buffer.
• The claim's addition is placement: perform the mix before storing to the PCM buffer, and add the buffered value to the windowing output ("overwriting") rather than storing all channels first.
• KSR rationale: "use of a known technique [time-domain downmix via equation (3)] to improve a similar device [the FIG. 6 decoder] in the same way" — the device is the same decoder, the technique is the identical admitted downmix equation, and the only change is where in the pipeline it happens. The result (PCM buffer shrinks from (256×10) to (256×4) words) is a predictable arithmetic consequence of channel-count reduction, which the '819 specification itself asserts as the benefit.
• The double-arrows/overwrite behavior is explicitly described in the spec as reading the once-stored value, adding the downmixed value, and re-storing — a routine accumulate-in-place step.

Confidence: high.

Claim 14 — buffer area for output at a predetermined rate

The '819 specification admits this is conventional: the double-buffering for constant-rate output is described as what "generally" happens in audio equipment, and is the stated reason the PCM buffer is (256×10) rather than (256×5). An admitted-conventional limitation cannot supply novelty. Confidence: high.

Claim 16 — downmix into the delay buffer used by windowing

Structurally parallel to claim 13, applied to the delay buffer. Same KSR rationale applies. The specification itself concedes the equivalence: "The same shall also apply to the relation with the delay buffer." The advantage (delay buffer (128×5) → (128×2) words) is the same predictable channel-count arithmetic. Confidence: high.

Claim 17 — distributing means separating delayed vs. non-delayed portions

The partition is dictated by the symmetry already relied on in claim 11: from equations (5)–(7) of the spec, only the k = N/2…3N/4−1 (delayed) values are "original," so only those need to be downmixed and stored; the rest are reconstructed. Once claims 11 and 16 are obvious, claim 17 is their mechanical combination — the finite, identified partition dictated by the transform's own algebra. Confidence: high.

6. Motivation to combine — stated explicitly

For each combination the record supplies an articulated reason (MPEP § 2143):

1. Same field / analogous art. R1 (multichannel subband decoder), R2 (MPEG-1 audio), R4 (AC-3), R5 (TDAC mathematics), and R6 (AC-3 standard) are all in audio coding/decoding using frequency-domain transforms. Combination is between references addressing the same problem (multichannel decode with limited resources), so any "non-analogous art" argument fails.
2. Common, recognized problem. Memory/DSP cost in consumer audio decoders. The '819 specification states this as the problem ("the memory capacity of a total of 6400 words is necessary … therefore, there is a problem such that a fairly large memory capacity is necessary"). Under KSR, identifying a known problem and applying a known solution is not invention.
3. Express teaching in R1. R1 does not merely render the frequency-domain reordering possible — it states it is "highly desirable" because cost becomes proportional to presentation channels rather than delivery channels. That is the strongest possible motivation, and the opposite of a teaching away.
4. Linearity makes the reorder a design choice, not a discovery. R1 explains that because "synthesis is linear … it is … possible to interpose a distributor between the deformatters and the synthesizers" — i.e., frequency-domain and time-domain downmixing are mathematically equivalent. A PHOSITA would therefore view the placement of the mix as a free design variable to be optimized for memory, exactly as the '819 claims do.
5. Finite, predictable solutions. For every architectural question the claims raise, the art presents a small, enumerated set with predictable outcomes: match block lengths or drop channels (claims 4/5/6); store both symmetric sets or one (claim 11); downmix before or after the output buffer (claim 13); before or after the delay buffer (claim 16). KSR holds such choices obvious.
6. No teaching away, no unexpected results. Nothing in R1–R6 discourages pre-buffer or frequency-domain downmixing, and the '819 specification's own Table 1 shows memory declining monotonically with each reordering — an arithmetic, not surprising, benefit. The only asserted quality caveat (dropping mismatched channels in the claim-6 alternative) is expressly described as acceptable, i.e., a predictable trade-off.

7. Secondary considerations (Graham factor 4) — likely unavailable to the patentee

• No evidence of unexpected results. Table 1 tracks the number of retained channels; the savings are exactly what channel-count reduction predicts. That is the antithesis of an unexpected result.
• No evidence of long-felt unmet need, failure of others, or copying appears in the record (and none is described in the specification).
• No nexus for any commercial success. The patent was never shown to be practiced by the original assignee. Its later assertion by MediaTek against Sanyo/Matsushita is litigation activity, not evidence of commercial success attributable to the claimed features. (See the earlier litigation section — do not conflate assertion with secondary-consideration evidence.)
• Skeptical secondary considerations (weigh for validity): the persistence of the block-length-matching problem through the 1990s is mild evidence that the claim 4–6 subject matter was not trivial — MPEG-1/MPEG-2 and AC-3 decoders continued to downmix in the time domain even though frequency-domain downmixing was mathematically available. That fact cuts against obviousness of claim 1's combination with variable block length, and it is the same fact the examiner presumably relied on in allowing claims 3–6. It does not rescue claims 8–18.

8. Bottom line and confidence summary

Overall assessment: Independent claims 1, 11, 13 and 16 are, on the record I could ground, strong § 103 (and for claim 1, potentially § 102) candidates, with claim 11 the most exposed because it rests on a transform property (R5) that is both inherent and four decades old. The patent's genuine center of gravity is the variable-block-length reconciliation of claims 3–6, which the art — as the later Dolby documents' own backgrounds admit — did not previously apply when downmixing in the frequency domain. Claim 8's pipelining limitation is the second relative weak point for a challenger, and it is the one limitation for which my search did not produce a specific pre-1995 reference.

Recommended verification before relying on this in a filing:

1. Retrieve the actual front-page "References Cited" for 5,867,819 (USPTO Patent Public Search / PatentCenter) to test whether R1 or R5 was already of record — if so, the § 103 case is far easier to plead; if not, that supports an inference the examiner never considered them.
2. Pin the priority date of US 6,931,291 and the publication date of EP 2 360 683 (currently treated as corroborative only).
3. Run a focused pre-1995 search on pipelined dequantizer/IMDCT decoder VLSI to shore up or abandon the claim-8 theory.

Two caveats restated from the earlier sections and not resolved here: the one-day bibliographic discrepancies (priority/filing/issue) between the patent record and Unified Patents, and the unretrieved front-page citation list. Neither changes the analysis above, but both should be pinned down before this is filed or relied upon.