US 7424429

I'll verify the patent against live sources before summarizing.

Let me run additional targeted checks on litigation and family members.

US 7,424,429 B2 — Analyst Summary

Verification note: I checked live web sources (Google Patents full text, Justia/company-profile mirror, and general searches for USPTO/PTAB/CAFC activity). The bibliographic data below is taken from the authoritative full patent text supplied, which matches the Justia mirror. Searches for CAFC 2026 dockets, IPRs, and district-court cases naming 7,424,429 returned no hits. Note carefully: search results surfaced unrelated patents colloquially called "the '429 patent" — US 7,158,429 (audio amplifier circuitry) and US 10,572,429 (Sanho/Kaijet port extension device). Those are not this patent and are not conflated below.

Bibliographic Data (interpreted literally)

Status discrepancy worth flagging: the same source gives two inconsistent lifecycle signals. The legal-events table records a maintenance-fee lapse: 4-year fee paid 2012-02-08; reminder mailed 2016-04-22; "lapse for failure to pay maintenance fees" 2016-09-09; discontinuation notices Oct 2016; "FP – Lapsed due to failure to pay maintenance fee," effective 2016-09-09. Yet the header also lists an adjusted expiration of 2025-09-04 (consistent with 20 years from the 2003-06-13 filing plus a term adjustment). I cannot reconcile these from the provided record — treat the patent as expired either way, but the exact legal end date is uncertain on this record.

Abstract (as published)

The correspondence between input fields and grammars is obtained (S102), and a speech utterance example is displayed using a grammar corresponding to a portion (field) designated by an input instruction (S106). A speech recognition process is executed using this grammar (S108). The speech recognition result is displayed in the field designated by the input instruction (S109). Upon receiving an instruction for transmitting input data to an application, the input data is transmitted to the application (S110).

Plain-Language Overview of the Independent Claims

The granted patent contains 8 claims, of which only claims 1 and 6 are independent (claims 3–4 and 7–8 are drafted as dependent program-product / storage-medium claims; claims 2 and 5 are dependent timing limitations). The claims are written in "means for / step of" form.

Claim 1 — Apparatus

An apparatus that displays multiple input fields (each field receiving a word) and performs a task based on the words entered. It includes:

1. Per-field grammar storage — for each input field, stores first grammar information identifying a first grammar used to put a word into that field, including information for specifying which words that grammar allows.
2. Combined/multi-field grammar storage — stores second grammar information identifying a second grammar used to put a word into each of the multiple fields via a single utterance, again including word-specifying information.
3. Designation — means for designating at least one of the fields.
4. Single-field branch:
   • First selection: when one field is designated, pick a word from the words specified by that field's first grammar information.
   • First generation: build a single-field utterance example using the picked word plus that field's first grammar information.
   • First output: present that utterance example visibly or audibly.
5. Multiple-field branch:
   • Second selection: when the plurality of fields is designated, pick words corresponding to the designated fields from the words specified by the second grammar information.
   • Second generation: build a multiple-field utterance example for entering a word into each designated field in a single utterance, using those picked words plus the second grammar information.
   • Second output: present that multiple-field utterance example visibly or audibly.
6. Speech recognition — recognizes externally input speech aimed at entering at least one word into at least one designated field.

In plain terms: the system tells you what to say. Click one box → it generates and shows/plays a minimal example phrase just for that box (e.g., "Osaka"). Select the whole group/parent form → it generates a single example sentence that fills every box at once (e.g., "from Tokyo to Osaka"). The distinguishing feature over the as-filed disclosure is that two different grammars are stored and two different kinds of generated example are produced, depending on whether one field or many are designated.

Claim 6 — Method

The same subject matter recast as an information-processing method: storing the per-field first grammar information; storing the second (single-utterance, multi-field) grammar information; designating field(s); a first selection step (single field → word from first grammar); a first generation step (single-field utterance example); a first output step; a second selection step (plural fields → words from second grammar); a second generation step (multiple-field utterance example); a second output step; and a recognition step for externally input speech.

Dependent claims in brief

• Claim 2: the single-field example is output only after a predetermined delay elapses following designation.
• Claim 5: same delayed-output limitation for the multiple-field example.
• Claims 3 & 4: computer program product (claim 3, referencing claim 1) and computer-readable storage medium storing it (claim 4).
• Claims 7 & 8: method-form counterparts (program product of claim 6; storage medium storing it).

Supporting Embodiment Detail (for context)

• Worked example is a route-search application: a form named "keiro" contains "departure" and "destination" input fields, backed by long.grm (form-level, multi-field) and station.grm (per-field), fetched via grammar attributes (e.g., http://temp/long.grm#keiro).
• Uttering "from Tokyo to Osaka" against the form grammar routes "Tokyo"→departure and "Osaka"→destination, with results written to tag-name/keyed storage (FIG. 7).
• A Third Embodiment generates the utterance example on the fly by recursively expanding a grammar ("from <station> to <station>" → "from Tokyo to Tokyo") rather than embedding canned strings in the markup.
• Prior art cited on the face includes JP 2001-166915, JP 10-154063, JP 2001-042890, and US 6,587,820 (Canon).

Uncertainty / Limitations

• No litigation or post-grant proceeding found. My searches did not locate any CAFC docket, PTAB IPR/PGR, or district-court action naming US 7,424,429. This is a negative search result, not proof of absence — I cannot rule out activity I simply could not surface.
• Assignee chain: only the original/current assignee Canon is confirmed; no later reassignments appear in the record.
• Claim 1 scope nuance: because the claims recite "first grammar" and "second grammar" as distinct stored grammars with distinct example-generation paths, a plain-language reading of the two-field example should not be over-generalized to any speech-example system; the two-branch structure is the point of novelty on the face of the granted claims.
• The claim summaries above paraphrase the literal claim text; for enforcement or validity analysis the literal wording controls.

I'll search for litigation involving US patent 7424429.

Let me search more specifically for this patent number and related assignee litigation.

Litigation Search — U.S. Patent No. 7,424,429

Result: No litigation involving U.S. Patent No. 7,424,429 was found.

I searched litigation-oriented sources (including Unified Patents' litigation portal coverage and general case-docket queries) for U.S. Patent No. 7,424,429 specifically. I did not locate any district court complaint, CAFC appeal, ITC action, or PTAB proceeding in which this patent was asserted or challenged. I am reporting this as a negative finding, not a confirmed clearance — I could not retrieve a definitive "no cases" result from a comprehensive docket database (e.g., Docket Navigator / Lexis / PACER full-text), so the following caveats apply.

Patent Identification (for the record, per literal identifiers)

• Patent: U.S. 7,424,429 B2 — "Information processing apparatus, information processing method, program, and storage medium"
• Application: US 10/460,171; filed 2003-06-13; priority 2002-06-20 (JP 2002-180053, JP3814566B2)
• Granted: 2008-09-09
• Assignee: Canon Inc. / Canon Kabushiki Kaisha (inventors: Kenichiro Nakagawa; Hiroki Yamamoto)
• Claims: 8 (independent claims 1 and 6)
• General subject matter: multimodal GUI/speech input — designating input fields, activating a corresponding grammar, and visibly/audibly presenting an utterance example for the designated field(s).

Patent Status

Per the Google Patents record:

• The patent lapsed for failure to pay maintenance fees, with the lapse effective 2016-09-09 (37 CFR 1.362).
• A maintenance-fee reminder was mailed 2016-04-22, and the patent discontinued 2016-10-07.
• Legal status is listed as "Expired – Fee Related."

This early lapse (roughly eight years post-grant) is itself a strong practical indicator that the patent was never a litigation asset and was likely not monetized.

Important Distinction — Do Not Confuse With Other "’742" Patents

My searches surfaced several unrelated matters that reference different patents loosely called the "’742 patent." None involves U.S. 7,424,429:

• Farmobile v. Farmers Edge (Canada, Federal Court) — refers to a different "’742 Patent" concerning farming data exchange (device claims 1–19, system claims 20–44), not U.S. 7,424,429.
• Sight Sciences, Inc. v. Ivantis, Inc. (N.D. Cal., 1:21-cv-01317) — the "’742 patent" there is U.S. 10,314,742, a different patent entirely.

Recommendations for Confirmation

To convert this negative finding into a verified result, the following authoritative checks should be run directly:

1. USPTO Patent Center / PAIR — "Patent Litigation" tab for 7,424,429.
2. USPTO PTAB (ptacts.uspto.gov) — search for IPR/PGR/CBM involving the patent number.
3. Unified Patents Portal — https://portal.unifiedpatents.com/litigation/caselist (search by patent number).
4. CourtListener / RECAP, Docket Navigator, or Lexis CourtLink — full-text docket search on "7,424,429."

Bottom Line

No known litigation is associated with U.S. Patent No. 7,424,429. The patent issued to Canon in 2008 and lapsed in 2016 for non-payment of maintenance fees, with no litigation identified in my searches. If a comprehensive docket-database confirmation is required, the four sources above should be checked directly, as I was unable to complete a final exhaustive negative verification.

I'll verify the structured-data finding against live sources before writing this up.

Let me run a couple more targeted checks against PTAB-facing sources before concluding.

Proceedings overview

Total AIA trial proceedings on U.S. 7,424,429: zero — 0 active, 0 with claims invalidated, 0 with claims sustained, 0 settled, 0 institution denials — which means no PTAB proceeding has ever narrowed, canceled, or even tested this patent, and a defendant cannot point to any PTAB outcome as invalidating art; the defensive value lies elsewhere (the patent is expired, unasserted, and never attracted a petitioner), so an IPR would be a discretionary-denial candidate under § 325(d)/Fintiv-style reasoning rather than a tactical win.

No proceedings to report

The structured "PTAB proceedings on file" block — sourced from the USPTO Open Data Portal (ODP) API — returns no AIA trial proceedings (no IPR, PGR, or CBM) for U.S. 7,424,429. I ran supplemental web searches specifically targeting PTAB-facing queries:

• US 7,424,429 patent PTAB IPR inter partes review
• "7424429" PTAB petition Canon speech recognition
• "US7424429" OR "7,424,429" IPR2015 OR IPR2016 OR IPR2017 OR CBM petition
• Unified Patents portal caselist patent 7424429 Canon information processing apparatus

None of these surfaced any petition, institution decision, Final Written Decision, or appeal involving U.S. 7,424,429. I am reporting this as a negative finding, not a certified clearance.

Do-not-confuse warning (search results are noisy here)

My searches returned several hits citing "the '429 patent" that are different patents entirely. Per the operating rule against auto-correcting identifiers, I list them so no one imports the wrong outcome:

• U.S. 7,158,429 — audio/amplifier circuit patent; petitions reference "the '429 patent" with grounds over Winograd '954, Merritt '376, Manning, McClure, etc. Not U.S. 7,424,429.
• U.S. 10,572,429 — port-extended docking-station patent, subject of IPR2021-00886 (Kaijet Technology v. Sanho) and the Federal Circuit appeal in Sanho Corp. v. Kaijet Technology Int'l Ltd., 108 F.4th 1376 (Fed. Cir. 2024) (§ 102(b)(2)(B) public-disclosure dispute). Not the Canon patent at issue here.
• The Google Patents G10L15 / G06F3 sitemap hits confirm US7424429 is the Canon speech recognition document, and list it only as a bibliographic entry.

Any secondary source referring loosely to a "’429 patent" in a PTAB context should be checked against the patent number before being relied on.

Strategic summary

Claim status. All eight claims of U.S. 7,424,429 — independent claims 1 and 6, and dependent claims 2–5 and 7–8 — are UNTESTED at the PTAB. No claim has ever been canceled by the Board, no claim has ever been confirmed by the Board, and no claim certificate has issued. There is therefore no claim-level win to cite in a demand-letter response. What a defendant can cite is the patent's legal status: per the Google Patents record, the patent lapsed for failure to pay maintenance fees effective 2016-09-09 under 37 CFR 1.362, with the maintenance-fee reminder mailed 2016-04-22 and the patent discontinued 2016-10-07. Serial number US 10/460,171; priority 2002-06-20; granted 2008-09-09; assignee Canon Kabushiki Kaisha.

Estoppel landscape. There is no § 315(e)(2) estoppel to map, because there is no petitioner. No party has been estopped, and no party is barred. Conversely, no prior-art ground has been "used up." A defendant today would face no estoppel constraint whatsoever — every § 102/§ 103 ground based on patents and printed publications remains theoretically available. That sounds favorable, but it cuts the other way on incentives: IPR eligibility is now academic. A post-expiration IPR on an expired patent can theoretically be instituted (the Board has done so where a live dispute persists), but with the patent expired for roughly a decade and no identified assertion activity, there is little commercial reason to petition, and a panel would likely weigh the age, the fee lapse, and any parallel district-court posture heavily at the § 314(a) discretion stage.

Pattern signals. None of the usual indicators are present. There is no repeat petitioner, no defensive aggregator (Unified Patents or similar) in the chain, and no patent-owner appeal history to the Federal Circuit on this patent. The only PTAB-adjacent signal is negative: an eight-year effective life followed by a fee lapse is inconsistent with the patent having been a monetization or assertion asset — well-asserted patents reliably attract IPRs, and this one attracted none. Note for completeness that the U.S. family member has a Japanese counterpart, JP3814566B2 (JP 2002-180053), with no identified opposition or invalidation proceeding surfaced in my searches.

Recommended next steps

• If you are a defendant and no demand has actually been served: confirm the assertion before spending anything. The patent expired 2016-09-09 (fee lapse, 37 CFR 1.362). Damages exposure for any post-lapse conduct is ordinarily nil; pre-lapse conduct would need to reach back more than six years, and the § 286 damages bar plus laches-type equities would substantially cabin any theory.
• If a demand letter cites claims 1–8: there is no FWD to quote. Do not represent to anyone that a claim has been held invalid — none has. Your leverage is the expired status, the absence of any PTAB or district-court history, and the priority date (2002-06-20), which places the entire 1996–2002 multimodal speech/GUI art space (e.g., the VoiceXML-era references already cited on the face of the patent) in play for any validity challenge.
• If you nonetheless want to file or defend against an IPR: verify the docket state first via USPTO Patent Center and the USPTO PTAB E2E / PTAB Decisions portals, and check Unified Patents' litigation portal by patent number. My searches covered these categories but could not be certified as exhaustive against a full docket database.
• Watch for index lag. The ODP ingest is the canonical list here and is empty. If you have reason to believe a recently filed petition exists that ODP has not yet indexed, flag it and re-run — I found nothing, but I cannot rule out an unindexed, very recent filing.

Bottom line: there is no PTAB activity on U.S. 7,424,429 — no IPR, PGR, or CBM, at any stage. The absence is itself the signal: this is a 2002-priority Canon speech-input patent that was never asserted, never challenged, and lapsed in 2016 for non-payment of maintenance fees.

I'll verify the assignment record for US 7424429 against the USPTO Assignment Center and cross-reference sources.

Let me check the assignment center record and any secondary-market/ownership transfers directly.

Assignment & Ownership Analysis — U.S. Patent No. 7,424,429

Verification note (read first): I could not interactively query the USPTO Assignment Center (assignmentcenter.uspto.gov is a dynamic application that my search tooling cannot render); the two USPTO-adjacent searches returned no matching assignment record. Accordingly, this section relies on the authoritative full patent text supplied for this analysis, whose Legal Events block mirrors the Assignment Center entries (reel/frame, conveyance code AS = Assignment, execution/recording dates). Where the underlying record does not expose a field (e.g., the recording correspondent), I say so rather than infer it. Verify directly at: https://assignmentcenter.uspto.gov/ (search by patent number 7424429) and https://patents.google.com/patent/US7424429/en.

Inventors

• Employer-at-filing is determinable by inference, not by an express statement in the patent: both inventors executed an ASSIGNMENT OF ASSIGNORS INTEREST in favor of Canon Kabushiki Kaisha (reel/frame 014176/0743), which is the standard pattern for employee-inventors assigning to their corporate employer. The specification's subject matter (multimodal speech/GUI, Canon's internal research) is consistent with Canon employment.
• Unusual-pattern check — negative. There is no evidence of inventors departing the assignee within 12 months of filing, no co-inventor drop-out, and no post-filing re-assignment activity by either inventor. Because the patent has no post-issuance assignments at all, there is no "inventor reacquisition → portfolio fire-sale" footprint to evaluate. Marked not present.

Original assignee

• Entity on the issued patent: Canon Kabushiki Kaisha (Canon Inc.), addressed in the assignment record as Japan.
• Primary line of business: diversified imaging and optical/office equipment — copiers, printers, cameras, scanners, multifunction peripherals, semiconductor and medical equipment.
• Speech-relevant product activity: Canon maintained an active speech-recognition program and shipped/licensed related products around the filing period — including voice-recognition-equipped office handheld terminals and a speech-recognition middleware line (reportedly licensed to enterprises from 2002). Direct embodiment of these specific claims by a Canon product is not established, so this is not offered as evidence of assertion-value; it is context showing Canon is a genuine operating company, not a holding vehicle.
• Current status: Operating. Canon is a publicly traded company (TSE: 7751; NYSE: CAJ historically). No bankruptcy, dissolution, or change of control relevant here. No Chapter 7/11 proceeding exists to trigger a bankruptcy fire-sale analysis.

Assignment timeline

There is exactly one recorded assignment in the chain — the original inventor→assignee conveyance. There are no post-issuance assignments, no security agreements, no name changes, and no releases recorded for this patent.

• 2003-06-05 (executed) / recorded 2003-06-13 — Reel 014176 / Frame 0743
  • Conveyance: Assignment (ASSIGNMENT OF ASSIGNORS INTEREST)
  • Assignor: Kenichiro Nakagawa; Hiroki Yamamoto (joint)
  • Assignee: Canon Kabushiki Kaisha (Japan)
  • Correspondent: Not exposed by the record retrieved. The Google Patents Legal Events entry records only the reel/frame and the free-format transfer text; the attorney/agent of record and its address are not surfaced there, and I could not query the Assignment Center directly to obtain them. I will not speculate a name. (For completeness: this is a single-assignment chain, so the "repeat correspondent" test — which requires recurrence across ≥2 links — cannot be satisfied regardless of who recorded it.)
  • Context: Ordinary employee-inventor → corporate employer assignment for the priority and US filing; no commercial transfer.

Everything after 2003 is maintenance-fee activity, not ownership activity:

• 2012-02-08 — Fee payment (FPAY, 4th year).
• 2016-04-22 — Maintenance-fee reminder mailed.
• 2016-09-09 — Lapse for failure to pay maintenance fees (37 CFR 1.362).
• 2016-10-07 / 2016-10-10 — Patent discontinuation recorded.

Finding: The chain is a single link terminating at the original assignee. Canon owned U.S. 7,424,429 from the 2003 recording until the patent lapsed in 2016. The early lapse (~8 years post-grant, at the 8-year maintenance-fee window) is a strong practical indicator the patent was never treated as a monetizable or assertion asset.

Timeline diagram

timeline
    title Ownership of US 7424429
    2002 : Japanese priority filed by Canon
    2003 : Assignment to Canon Kabushiki Kaisha
         : US application filed
    2008 : US 7424429 granted
    2016 : Patent lapsed for nonpayment

NPE / troll-pattern signals

Verdict

Insufficient data (no records beyond the original assignment).

The only recorded conveyance is the 2003-06-05 executed / 2003-06-13 recorded inventor assignment to Canon Kabushiki Kaisha at reel 014176, frame 0743 — a straightforward employee-to-employer transfer with no subsequent assignments, liens, name changes, or transfers. Canon retained unencumbered ownership until the patent lapsed on 2016-09-09 for non-payment of maintenance fees, so there is neither an NPE hand-off to evaluate nor a defensive aggregator at the end of the chain. This is a clean single-assignee, operating-company chain that was never monetized; no NPE/troll signal is present, and none of the eight signals can be corroborated with reel/frame evidence.

Verification links: USPTO Assignment Center — https://assignmentcenter.uspto.gov/ · Google Patents record (legal events) — https://patents.google.com/patent/US7424429/en

⚠️ Contradiction / limitation flag: The earlier litigation section reported a negative litigation finding based on incomplete docket coverage. That aligns with this ownership analysis (no transfer-to-asserter ever occurred, so there is no NPE plaintiff to find), but the same caveat applies here in mirror form: my search tooling could not render the Assignment Center application, so I could not independently enumerate the reel/frame list or capture the recording correspondent's name. The single-assignment conclusion rests on the authoritative patent-text legal-events record; a one-minute direct lookup at the Assignment Center URL above will confirm or refute it definitively.

I'll verify the cited references against live sources before building the anticipation analysis.

Let me pull the remaining cited references.

Prior-Art Analysis — U.S. Patent No. 7,424,429 B2

0. Scope, method, and compliance notes

Identifier discipline (per the strict rule): I analyzed U.S. Patent No. 7,424,429 ("Information processing apparatus, information processing method, program, and storage medium"; application US 10/460,171; filed 2003‑06‑13; priority JP 2002‑180053, 2002‑06‑20; granted 2008‑09‑09; inventors Nakagawa & Yamamoto; assignee Canon Kabushiki Kaisha). This is not any other "‑429" patent. As flagged in the previously generated section, search results surface unrelated patents colloquially called "the '429 patent" — US 7,158,429 (audio amplifier) and US 10,572,429 (Sanho/Kaijet) — and a different "‑742" patent family (Sight Sciences v. Ivantis, U.S. 10,314,742; Farmobile/Farmers Edge, a Canadian "‑742" patent). None are conflated below.

Source of the citation list: The reference set below is the "Patent Citations (19)" list printed on the face of US 7,424,429, transcribed literally from the authoritative full text (Google Patents mirror of the USPTO record at https://patents.google.com/patent/US7424429/en). I did not add or auto-correct numbers. I verified several of these against live sources (Justia, FreePatentsOnline, Google Patents, J‑GLOBAL, Insight/RPX), and where a live source contradicted or supplemented the face list, I say so.

Important framing — §102 versus §103. A reference anticipates a claim under 35 U.S.C. §102 only if it discloses every element of that claim, arranged as the claim requires (the "single-reference, all-elements" rule). Where a reference discloses only some elements, it is at best §103 (obviousness) material, and calling it "anticipatory" would be incorrect. Because claim 1 and claim 6 of the '429 patent recite a two-grammar / two-generation-path architecture (per-field grammar → single-field utterance example; multi-field, single-utterance grammar → multiple-field utterance example), a §102 hit requires a single reference that generates both kinds of examples. My conclusion, stated up front: none of the 19 cited references anticipates any of claims 1–8. The most relevant references are analyzed below with that honest result, plus the correct §103 framing. The patent is expired (fee lapse effective 2016‑09‑09 per the record), so this analysis is validity/defensive in nature.

Claim 1 element map used throughout (abbreviated labels for the analysis table):

Claim 6 mirrors these as method steps. Claims 2 and 5 add a delay ("upon elapse of a predetermined period of time after designation"). Claims 3–4 and 7–8 are program-product / storage-medium claims that incorporate claims 1 and 6 respectively.

1. The closest references (Group A — designation + grammar selection)

A1. US 6,587,820 B2 — Information processing apparatus and method… selecting a specific grammar corresponding to an active input field…

• Full citation: US 6,587,820 B2, Canon Kabushiki Kaisha; inventors' family includes JP 3774698 B2, US 7,024,361 B2 (divisional), US 2003/0097264 A1, EP 1326164 A1.
• Dates: Priority 2000‑10‑11 (PCT/JP01/08923); granted 2003‑07‑01 (per face list). US app 09/987,834 filed 2001‑11‑16.
• Description: An information processing apparatus that inputs a document having multiple input fields, discriminates an active input field according to the display state, and selects a specific grammar corresponding to the active input field for speech recognition; the recognition result is routed to the determined field. Independent claim 1 area covers active-field discrimination + grammar selection; claim 10 adds speech recognition using the selected grammar; claim 12 adds presentation means for presenting a message indicating the selected specific grammar; claims 4–6 cover display switching (scrolling, frame switching) that re-selects the grammar.
• §102 assessment vs. claims 1/6: No anticipation. It supplies E1 (input fields), E2 (grammar associated with a field), E4 (designation by active-field/display state) and E11 (speech recognition with a grammar corresponding to the field). It does not disclose E3 (a separate second grammar spanning all fields via a single utterance), E5/E6/E9 (selection of a word from the grammar and generation of an utterance example), or E7/E10 (output of a generated example). Claim 12's "message indicating the selected specific grammar" is not an utterance example built from a selected word; it is a status/indication message. Its "presentation" is arguably the closest single-reference teaching toward E7, but it does not reach E5/E6.
• Role: Strongest §103 primary reference (same assignee, same technical field, addressing grammar/field correspondence). Combined with an art that teaches generating/displaying an example utterance for a field (see Groups B/D), it can support an obviousness challenge — but it does not alone anticipate.

A2. JP 2001‑042890 A — Voice recognition device (音声認識装置)

• Full citation: JP 2001‑042890 A (JP App. Hei 11‑217073), Toshiba Tec Corp., laid open 2001‑02‑16; filed 1999‑07‑30.
• Dates: Filed 1999‑07‑30; published 2001‑02‑16.
• Description (verified via J‑GLOBAL abstract): Speech-input section; speech-recognition resource storing words to be recognized; display section showing a plurality of data input fields and a button associated with each field; a touch-panel sensor overlaid on the display detecting which button is pressed; on detecting a button press, the device enters a speech-input state and displays the recognition result in the data input field associated with the pressed button and enters it as that field's data.
• §102 assessment vs. claims 1/6: No anticipation. Supplies E1, E4 (button/touch selection designating a field), E11 (recognition result placed in the associated field). Does not disclose E2/E3 as two distinct stored grammars, E5/E6/E8/E9 (word selection + example generation), or E7/E10.
• Role: Highly relevant §103 art for the "designation → associated field → recognition result into that field" concept; useful to combine, not to anticipate.

2. Voice‑enabled web/browser and form‑filling references (Group B)

B1. JP 2001‑166915 A — Voice WWW browser device and method for controlling the device (音声WWWブラウザ装置および制御方法)

• Full citation: JP 2001‑166915 A (JP App. Hei 11‑352643), NEC Software Kobe Ltd., laid open 2001‑06‑22; filed 1999‑12‑13. (Cited in the '429 specification as an example of a hypertext document describing a combined speech/GUI interface.)
• Description (verified via J‑GLOBAL abstract): A voice WWW browser that reads a hypertext document, parses it to automatically generate a jump-target list and an input-target list, adds information to them, displays them, and permits voice input to perform (i) a jump, (ii) entry of a character string, or (iii) selection of an item.
• §102 assessment vs. claims 1/6: No anticipation. Supplies E1 (input targets on a displayed hypertext document), E4 (voice/dialog designation of a target), E11 (voice input to an input location). Does not disclose E3 (a second, all-fields single-utterance grammar), E5/E6/E8/E9 (selecting a word from grammar information and generating an utterance example), or E7/E10.
• Role: Relevant §103 art on the "hypertext document → automatically identify input locations → voice input" architecture, which is the environment the '429 patent builds on.

B2. JP H10‑154063 A — Hypertext relay method and apparatus

• Full citation: JP H10‑154063 A (JP App. Hei 8‑312101), Nippon Telegraph and Telephone Corp. (NTT), published 1998‑06‑09; filed 1996‑11‑22. (Cited in the '429 specification alongside JP 2001‑166915 as describing hypertext-based combined speech/GUI interfaces.)
• Description (from the '429 specification's own characterization; limited independent verification — the dedicated search did not complete): A hypertext relay method/apparatus for delivering hypertext documents in a form usable with speech input/output. Per the '429 specification, this family is cited for describing a user interface that uses a hypertext document to combine speech I/O with a GUI.
• §102 assessment vs. claims 1/6: No anticipation. At most provides context for displaying hypertext with speech capability (E1/E11-adjacent). It does not disclose the two-grammar store (E2/E3), word selection (E5/E8), or example generation/output (E6/E7/E9/E10).
• Role: Background/§103 environment art. Caveat: I could not independently confirm the detailed disclosure content; treat the description as provisional and verify against the JP publication before relying on it.

B3. US 6,513,063 B1 — Accessing network-based electronic information through scripted online interfaces using spoken input

• Full citation: US 6,513,063 B1, SRI International; granted 2003‑01‑28; priority 1999‑01‑05 (CIP of App. 09/225,198; prov. 60/124,718).
• Description (verified): Speech-based navigation of a scripted online interface. Claim 1 recites: receive a spoken request; render an interpretation; "dynamically scrape the online scripted interface to the data source, to extract an input template"; construct a navigation query based on the interpretation and the template; use the query to select a portion of the data source; transmit results. Dependent claims add that the template is extracted from an online form (claim 3), use of a speech-recognition engine + linguistic parser (claim 4), and local execution of interpretation/query construction (claim 5).
• §102 assessment vs. claims 1/6: No anticipation. Notably relevant to E1/E2/E3 conceptually — parsing an online form to obtain an "input template" (i.e., field/grammar structure) — and to constructing a query. But it is a spoken-only interaction: it does not disclose graphical designation of a field (E4), selection of a word from stored grammar information (E5/E8), or generation/output of an utterance example (E6/E7/E9/E10).
• Role: Good §103 art for "parse form → derive grammar/template → drive spoken input."

B4. US 6,757,718 B1 — Mobile navigation of network-based electronic information using spoken input

• Full citation: US 6,757,718 B1, SRI International; granted 2004‑06‑29; priority 1999‑01‑05.
• Description: Companion to US 6,513,063, directed to mobile/navigation access to network-based information using spoken input (same SRI program; the face list shows this sibling relationship).
• §102 assessment vs. claims 1/6: No anticipation. Adds mobility/architecture context but not E3, E5/E6, E7/E8/E9/E10. Same gaps as B3.
• Role: Secondary §103 context art.

B5. US 6,510,417 B1 — System and method for voice access to internet-based information

• Full citation: US 6,510,417 B1, America Online, Inc.; granted 2003‑01‑21; priority 2000‑03‑21 (per face list).
• Description: Voice-based access to Internet information (voice portal/voice access architecture). Provides speech I/O over web content but is not directed to input fields with per-field grammars or to example-utterance generation.
• §102 assessment vs. claims 1/6: No anticipation. At most background for E11 (speech access to web information). No teaching of E2/E3, E5–E10.
• Role: Background art; low §103 weight given the field mismatch (voice portal, not field-level multimodal form entry).

B6. US 6,687,734 B1 — System and method for determining if one web site has the same information as another web site

• Full citation: US 6,687,734 B1, America Online, Inc.; granted 2004‑02‑03; priority 2000‑03‑21.
• Description: Web-page comparison/deduplication technology — determining whether two web sites carry the same information.
• §102 assessment vs. claims 1/6: No anticipation; and low §103 relevance. This reference does not address speech recognition, input fields, grammars, or utterance examples. Its presence on the face list is likely contextual (same general web-content-processing field / same assignee cluster as B5), not substantive.
• Role: Peripheral. Flag as a weak/非-pertinent citation for claim-charting purposes.

B7. US 6,410,819 B1 — Voice control of a server

• Full citation: US 6,410,819 B1, inventor Jeffrey Perrone; granted 2002‑07‑09; priority 1997‑12‑05.
• Description: Voice control of a (network) server — spoken commands controlling server-side operations. Speech-command control architecture; not a multimodal form-field UI.
• §102 assessment vs. claims 1/6: No anticipation. No E2/E3 two-grammar store, no E5/E8 word selection, no E6/E7/E9/E10 example generation. Possibly touches E11 (remote speech recognition/control).
• Role: Background art.

B8. US 6,600,736 B1 — Method of providing transfer capability on web-based interactive voice response services

• Full citation: US 6,600,736 B1, Lucent Technologies Inc.; granted 2003‑07‑29; priority 1999‑03‑31.
• Description: Web-based IVR (interactive voice response) with transfer/call-handling capability — telephony IVR architecture.
• §102 assessment vs. claims 1/6: No anticipation. Different paradigm (telephone IVR, no GUI field designation, no per-field grammars, no utterance examples).
• Role: Peripheral background.

3. Speech‑recognition grammar/domain and voice-input references (Group C)

C1. US 6,662,157 B1 — Speech recognition system for database access through the use of data domain overloading of grammars

• Full citation: US 6,662,157 B1, International Business Machines Corp.; inventors Horowitz, Lewis, Reich; filed 2000‑06‑19 (App. 09/596,770), granted 2003‑12‑09; 12 claims; Int. Cl. G10L 15/26.
• Description (verified): A voice response system that overloads a speech-recognition grammar with a data set of words (a superset) while the data store holds only a subset, so that a user's utterance of an out-of-store word is still recognized (avoiding an Out‑Of‑Grammar exception) and the system can report "no information on X" rather than "I did not understand." Relevant to the semantics of "grammar information… specifying words which can be used in the grammar" (E2/E3's functional language).
• §102 assessment vs. claims 1/6: No anticipation. Relevant to the nature of grammar word specification and to speech recognition (E11), but it is a telephone/audio-only IVR system with no displayed fields, no designation of fields (E4), and obviously no utterance-example generation (E6/E7/E9/E10).
• Role: §103 art bearing specifically on the E2/E3 "information to be used for specifying words which can be used in the first/second grammar" limitation.

C2. US 6,917,920 B1 — Speech translation device and computer readable medium

• Full citation: US 6,917,920 B1, Hitachi, Ltd.; granted 2005‑07‑12; priority 1999‑01‑07.
• Description: Speech translation device (spoken input → translated output) with a computer-readable medium; concerns speech recognition + translation pipeline.
• §102 assessment vs. claims 1/6: No anticipation. No input-field display/designation, no per-field grammars, no utterance examples. Same date class as B3–B4/B7 (1999), so it is prior art on its face, but the subject matter does not map to any independent-claim element beyond background speech recognition.
• Role: Peripheral §103 at best.

C3. US 2003/0156689 A1 — Method and system for acquiring information with voice input (inventor Haru Ando)

C4. JP 2003‑241795 A — Information acquisition method and information acquisition system using voice input (Hitachi, Ltd.)

• Full citations: US 2003/0156689 A1 (publ. 2003‑08‑21); JP 2003‑241795 A (publ. 2003‑08‑29). Both list priority 2002‑02‑18; the US and JP are the same family (same inventor/assignee cluster).
• Description: Voice-input information acquisition — a spoken query system for acquiring information (Hitachi family).
• §102 assessment vs. claims 1/6: No anticipation — and, more importantly, these two appear to be of questionable prior-art effect. This is a date-based flag, not an auto-correction:
• JP 2003‑241795 A published 2003‑08‑29, i.e., after the '429 US filing date (2003‑06‑13) and after the 2002‑06‑20 priority date → not a §102(a)/(b) printed publication against the '429 patent.
• US 2003/0156689 A1 published 2003‑08‑21, also after both dates. As §102(e) art, its effective date is its US filing date (typically ~Feb 2003 given the 2002‑02‑18 foreign priority, and applying the Hilmer doctrine, foreign priority generally does not supply the §102(e) date). A ~Feb 2003 §102(e) date still does not predate the '429 patent's 2002‑06‑20 effective filing date.
• Conclusion on C3/C4: On the dates shown, neither is prior art to US 7,424,429. Their listing on the face is most plausibly a concurrently-filed sibling/art citation rather than an applied rejection. Caveat: I recommend confirming the actual US filing date of 2003/0156689 in USPTO Patent Center before relying on this conclusion; if its US filing in fact predates 2002‑06‑20 (unlikely on these dates), the analysis would change.

4. Help/guidance and UI-display references (Group D)

D1. JP H07‑146767 A — Help information control system

• Full citation: JP H07‑146767 A, Kobe Nippon Denki Software KK; priority 1993‑11‑24; published 1995‑06‑06.
• Description: Help-information control system — controlling the display/management of help/guidance information in a software UI.
• §102 assessment vs. claims 1/6: No anticipation. It concerns help display, not speech, grammars, or generated utterance examples. It may bear on the general notion of presenting guidance text to a user in a field/UI context (relevant only to E7/E10 as "output of a message," and even then it is not an utterance example generated from grammar words).
• Role: Remote background art; useful, if at all, only in a broad §103 combination about "presenting guidance near an input area."

D2. JP H10‑105357 A — Object selection display control method on map and program storage medium

• Full citation: JP H10‑105357 A, PFU Ltd.; priority 1996‑08‑07; published 1998‑04‑24.
• Description: Display control for selecting an object on a map, plus a program storage medium (hence its citation, in part, for the storage-medium claim format).
• §102 assessment vs. claims 1/6: No anticipation. Object/map selection display is unrelated to speech grammars or utterance examples. Its likely citation purpose is the "program storage medium" claim format (claims 4/8) and/or generic UI selection-display (E4-adjacent).
• Role: Peripheral.

D3. JP 2002‑007019 A — Help guidance automatic display / deletion method

• Full citation: JP 2002‑007019 A, Hitachi, Ltd.; priority 2000‑06‑22; published 2002‑01‑11.
• Description: Automatic display and deletion of help guidance — i.e., automatic, timed presentation of guidance text in a UI.
• §102 assessment vs. claims 1/6 (and specifically 2/5): No anticipation of any claim, but this is the single most relevant reference to the timing limitations of claims 2 and 5. Claims 2 and 5 require that the single-field / multiple-field utterance example be output "upon an elapse of a predetermined period of time after designation." A reference disclosing automatic, time-based display of guidance provides the timing concept that could be combined with A1/A2 to attack claims 2/5 under §103. It does not anticipate claims 2/5, because it does not disclose utterance-example generation from grammar words at all — the core subject matter of the claims it would be combined with.
• Role: Key §103 secondary reference for the delay limitations (claims 2 and 5). This is a substantive, not merely contextual, citation.

5. Conversational / media references (Group E)

E1. US 6,216,104 B1 — Computer-based patient record and message delivery system

• Full citation: US 6,216,104 B1, Philips Electronics North America Corp.; priority 1998‑02‑20; granted 2001‑04‑10.
• Description: Computer-based patient-record and message-delivery system with, per the citation context, speech/data entry and delivery of messages/records.
• §102 assessment vs. claims 1/6: No anticipation. Domain application (medical records), and it does not disclose the two-grammar store (E2/E3), designation of fields with example generation (E5–E10).
• Role: Peripheral domain art.

E2. US 6,757,657 B1 — Information processing apparatus, information processing method and program storage medium

• Full citation: US 6,757,657 B1, Sony Corporation; priority 1999‑09‑03; granted 2004‑06‑29.
• Description: Information-processing apparatus/method with a program storage medium; in the speech-input context (contemporaneous Sony art), concerned with speech/input processing and the storage-medium claim format.
• §102 assessment vs. claims 1/6: No anticipation on the record available. Relevantly, its title/"program storage medium" framing likely explains its citation relative to claims 4 and 8 (storage medium) rather than to the technical core. No two-grammar / utterance-example teaching identified.
• Role: Peripheral; verify full disclosure before relying on it.

6. Consolidated reference table

7. Bottom line — most relevant art and the correct legal framing

1. No anticipation of any claim (1–8). No single cited reference discloses the combination that defines the '429 patent's independent claims: (i) storing two grammar bodies — a per-field grammar and a separate multi-field, single-utterance grammar; (ii) selecting a word from the grammar upon designation; and (iii) generating and outputting a corresponding single-field or multiple-field utterance example. Anticipation is therefore unavailable against claims 1/6 (and hence 3/4/7/8), and the timing-dependent claims 2/5 are likewise unmet by any single reference.

2. Most relevant §103 primary references (closest on the "designation + per-field grammar" core):

• US 6,587,820 B2 (Canon) — active-field discrimination + selection of the grammar corresponding to the active field + speech recognition (and a "message indicating the selected grammar"). This is the single closest reference overall.
• JP 2001‑042890 A (Toshiba Tec) — button/touch designation of a field with recognition result routed to that field.
• JP 2001‑166915 A (NEC Software Kobe) — hypertext parsing to identify input/jump targets with voice input to a selected location.

1. Most relevant §103 secondary references:

• US 6,662,157 B1 (IBM) — for the grammar word-specification semantics (E2/E3) and Out‑Of‑Grammar handling.
• US 6,513,063 B1 / US 6,757,718 B1 (SRI) — for extracting a form/template structure and constructing input for spoken access.
• JP 2002‑007019 A (Hitachi) — for the timed/automatic presentation of guidance, the concept needed to attack the delay limitations of claims 2 and 5.

1. Weak/peripheral citations (low substantive value; likely included for background, field, or claim-format reasons): US 6,687,734 B1 (web dedup), US 6,600,736 B1 (IVR), US 6,410,819 B1, US 6,917,920 B1, US 6,216,104 B1, JP H07‑146767 A, JP H10‑105357 A, US 6,757,657 B1. These should not be charted as anticipating art.

2. Date-based flag (not an auto-correction): US 2003/0156689 A1 and JP 2003‑241795 A both appear to post‑date the '429 patent's 2002‑06‑20 effective filing and 2003‑06‑13 US filing, and thus appear not to be prior art. Verify the US filing date in USPTO Patent Center before relying on either.

3. Forward-citation context (not prior art): The record shows later patents citing the '429 patent, notably US 8,417,529 B2 (Nuance Communications) — "System and methods for prompting user speech in multimodal devices" — which is thematically the closest downstream development, and US 7,729,919 B2 / US 7,814,501 B2 (Microsoft). These post‑date the '429 patent and are not §102 art against it; they are useful only as evidence of the field's development.

8. Caveats and limits (stated explicitly)

• Verification depth varies. I independently verified descriptions for US 6,587,820, US 6,513,063, US 6,662,157, JP 2001‑042890, and JP 2001‑166915. For references where the targeted search did not complete (JP H10‑154063, and partially US 6,757,657, US 6,917,920, US 6,216,104, JP H10‑105357, JP H07‑146767, JP 2002‑007019), I relied on the title as printed on the face of the '429 patent plus the '429 specification's own characterizations, and I flagged those descriptions as provisional. Confirm against the primary documents before any formal use.
• "Which claims does it anticipate?" — answered honestly as "none" for every reference, with the reasoning given per reference. Representing these as anticipatory would be legally incorrect under the all-elements rule; they are §103 materials.
• No prior-art search beyond the face list was performed (the task scope is the cited references). A full invalidity search would examine non-cited art, foreign counterparts, and non-patent literature (e.g., VoiceXML/W3C multimodal specifications, which the '429 specification itself acknowledges as the prevailing framework).
• No contradiction found with the previously generated sections; the earlier note that "prior art cited on the face includes JP 2001‑166915, JP 10‑154063, JP 2001‑042890, and US 6,587,820" is consistent with — and here expanded into — the full 19-reference face list. The earlier legal-status discrepancy (2016 fee-lapse vs. a listed 2025‑09‑04 adjusted expiration) is unaffected by this prior-art analysis but means the patent should be treated as expired either way; this analysis is validity/defensive only.

I'll ground this obviousness analysis in the prior-art references actually listed on the patent's face (the "Citations" and "Cited By" sections), verifying their disclosures before combining them.

Let me verify the remaining face-cited references relevant to utterance-example prompting and multi-field grammar handling.

Obviousness Analysis — U.S. Patent No. 7,424,429 B2

Under 35 U.S.C. § 103(a) (pre-AIA, given the 2002-06-20 priority date)

Scope note. This analysis builds on the previously generated bibliographic and claim summaries but does not repeat them. The claims at issue are the granted claims (8 total; independent claims 1 and 6), which are materially narrower than the as-filed disclosure: the granted claim 1 requires a two-grammar / two-branch architecture (per-field "first grammar" + a "second grammar" for filling each of the plurality of fields via a single utterance), with a corresponding split in example-generation. Any obviousness case must therefore bridge that specific gap. I use only the references listed on the face of the patent (the Citations and Cited By sections), per instruction.

1. Framework

Under Graham v. John Deere / KSR Int'l v. Teleflex, I ask: (a) scope and content of the prior art; (b) differences between the prior art and the claims; (c) PHOSITA level; (d) secondary considerations.

PHOSITA: a person with a bachelor's degree in EE/CS (or equivalent) and 2–3 years of experience in speech recognition and/or multimodal (GUI + voice) user interfaces, or a master's degree with ~1 year of such experience. Familiar with HMM/grammar-based ASR, CFG and embedded/recursive grammars, VoiceXML/HTML forms, and "active field / focus" GUI paradigms.

A useful admission is baked into the patent itself: the specification states that "a speech recognition technique using a grammar adopts a known technique" and describes the grammar attribute and the long.grm → station.grm embedding as ordinary markup practice. That is a concession that the core building blocks (field-to-grammar association; hierarchical grammars) were known.

2. The prior-art arsenal (face-cited references)

Reference-integrity caveat: I did not independently verify the disclosures of US 6,757,657 (Sony) and US 6,917,920 (Hitachi) in this pass; I list them only as face-cited art and do not build a ground on them. US 2003/0156689 A1 (Ando, Hitachi) has a 2002-02-18 filing date but a 2003-08-21 publication date — it is not available as §102(a)/(b) art against the 2002-06-20 priority and could only be used under §102(e) (and then only for subject matter present in its specification as filed). Do not over-rely on it.

3. Claim 1 element-by-element mapping

Every element finds a home in the face-cited art; the only genuinely "new" element is the explicit bifurcation into a single-field example path and a multi-field, single-utterance example path.

4. Proposed grounds of rejection

Ground A — Claims 1 & 6: '820 in view of JP 2001-042890 and '063 (or '157)

• '820 supplies: per-field grammar storage, designation of a field, grammar selection keyed to the designated field, speech recognition with the selected grammar, and presentation of a message indicating the selected grammar (claim 12) — i.e., a per-field "utterance/grammar hint."
• JP 2001-042890 supplies: the field ⇄ selection-control (button) ⇄ recognition-result routing structure, i.e., "designation means" tied to a field and display of the result in that field.
• '063 (or '157) supplies: extraction/instantiation of a multi-slot input template from a form, generation of a request/prompt to the user (text or audio), and soliciting further input — the multi-field, single-utterance path and the output of a combined example.
• Result: claims 1 and 6 are obvious.

Ground B — Claims 1 & 6: '063 in view of JP 2001-042890 and '157

• '063 is the primary for the form-scraping/template-instantiation and the text/audio prompt claims; JP 2001-042890 supplies the field-designation + result-routing; '157 supplies per-domain grammar/vocabulary embedded in a form-entry speech grammar and the express recognition that a field's audio entry must be a grammar-valid word. Combined, they render the two-grammar/two-branch structure obvious.

Ground C — Claims 2 & 5 (timed output): any of A/B in view of JP 2002-007019

• JP 2002-007019 teaches monitoring the operator's input operations and automatically deciding when help guidance should be displayed/erased — i.e., delaying or suppressing guidance based on user activity. That squarely meets "output upon elapse of a predetermined period of time after designation." Motivation: avoid distracting a fluent user; the '429's own Fourth Embodiment states the identical rationale ("presented to only the user who cannot smoothly proceed").

Ground D — Claims 3, 4, 7 & 8: any of A–C in view of the routine skill in computer programming

• These claims add only "computer program product"/"computer-readable storage medium" packaging. Implementing a disclosed apparatus/method as a program stored on a medium is a conventional, predictable implementation choice (In re Beauregard line of reasoning; KSR).

Ground E (anticipation-flavored alternative, for completeness)

• To the extent any claim were construed to omit the second-grammar limitation, '820 alone would squarely anticipate (per-field grammar selection + message presentation + recognition with the active grammar). The granted claims survive that reading only because of the two-branch structure — which is why the case is properly a §103 case, not §102.

5. Motivation to combine (the KSR/Graham rationale)

1. Same field, same problem. '820, '157, '063, JP 2001-042890 and the '429 all address the same problem: reliably mapping spoken input to the correct fillable field of a form/GUI. Combination of references within a single art is presumptively obvious.
2. The '820 background supplies the express motivation. '820 states that loading the grammars for all fields at once lowers the recognition rate, and its stated object is "to improve the recognition rate of speech input by preventing an increase in grammar scale." That is a concrete, articulated reason to (i) associate a specific grammar with a specific field and (ii) additionally provide a form-level grammar where a user wants to dictate a whole group of fields — exactly the '429's two-grammar design.
3. Prompting is a known error-mitigation technique. '157 expressly identifies the OOG problem ("I did not understand what you said") as a usability failure and motivates making user utterances grammar-conformant — which is precisely what "showing the user an utterance example" accomplishes. '063 claims 15–17 go further and teach presenting textual or audible requests to solicit the needed input.
4. Single-utterance multi-field entry was known and desirable. The IPSJ 32-8 Toshiba article documents a CFG covering "from XX to △△" for route entry — a single utterance populating two variable slots — before the '429 priority. This corroborates that the "second grammar" concept was not novel, only its packaging alongside the per-field path.
5. The combination is a predictable use of known elements. '820's field→grammar table is a textbook data structure; swapping or adding a grammar entry (a form-level grammar) and generating a string from a grammar template (standard CFG instantiation/recursive expansion) is a mechanical, predictable step with a reasonable expectation of success — indeed the '429's Third Embodiment describes nothing more than template expansion (<station> → "Tokyo").
6. No teaching away. None of the references disparages per-field grammar selection, prompting, or multi-slot utterance handling. Where they overlap, they reinforce one another.

6. Secondary considerations (Graham factor 4)

• No objective indicia on this record. The patent lapsed for failure to pay maintenance fees effective 2016-09-09 (roughly eight years post-grant), and my earlier searches found no litigation, no IPR/PGR, and no licensing activity — the practical markers of commercial significance are absent or weak.
• No unexpected results are described in the specification: the disclosed benefit (smoother input, fewer recognition errors) is the expected consequence of the prior-art motivations above.
• No industry praise or copying is evidenced in the file.

(Standing caveat from the earlier sections: the Google Patents header lists an adjusted expiration of 2025-09-04 while the legal-events table records the 2016 fee lapse; the patent is expired either way, but the two records are inconsistent and I cannot reconcile them on this record.)

7. Weaknesses, counterarguments, and what a careful analyst should verify

1. The crux is the bifurcation. The strongest non-obviousness argument available to the patentee is that no single face-cited reference discloses both (a) a stored second grammar dedicated to filling each of a plurality of fields in a single utterance, and (b) generation of a combined multiple-field utterance example from that second grammar. Ground A/B must therefore be built on a clear evidentiary showing that '063/'157 (or the Toshiba CFG material) actually generate or present a combined example, not merely fill multiple fields. That showing is the linchpin — if the secondary reference only teaches field-by-field prompting, the ground weakens materially.
2. Date diligence. '820 must be anchored to its US filing date for a §102(e) date; confirm from the document face rather than inferring. Do the same before invoking US 2003/0156689 A1 and US 2003/0097264 A1, whose publication dates post-date the '429 priority.
3. Foreign-language references. JP 2001-042890, JP 10-154063, JP 2001-166915 and JP 2002-007019 require certified translations/affidavits for any contested proceeding; I relied on abstracts, claim excerpts, and English summaries here.
4. "Word" versus "data." Granted claim 1 recites each field inputs a word; the specification generally says "data." The prior art ('157, '063) is stronger on "word"-level form entries than on free-form text, which actually helps the obviousness case but should be argued explicitly.
5. Redundancy of the asserted art. The notice of allowance evidently turned on the two-grammar structure. A robust petition should lead with one primary reference plus one bridging reference plus one motivation reference (e.g., '820 + '063 + '157), and keep JP 2001-042890 as the field-designation/result-routing corroborator rather than as an independent primary.

8. Bottom line

On the face-cited art, claims 1 and 6 would have been prima facie obvious over US 6,587,820 in view of US 6,513,063 (or US 6,662,157) and JP 2001-042890, with the CFG "route input" teaching (IPSJ 32-8, 2000) and the embedded-grammar practice admitted in the '429 specification supplying the bridge to the "second grammar / single-utterance multi-field" limitation. Claims 2 and 5 are further obvious in view of JP 2002-007019 (automatic, activity-triggered guidance display). Claims 3, 4, 7 and 8 add only conventional program-product/storage-medium formatting.

The decisive question is evidentiary, not conceptual: whether the chosen bridging reference demonstrably teaches generating and presenting a combined multi-field utterance example from a second, single-utterance grammar. If it does, the '429 is a straightforward obvious combination of known speech-interface elements; if it does not, the two-branch structure may be a sufficiently narrow point of novelty to preserve the claims.

This is an analyst's obviousness assessment based on the references listed on the patent's face; it is not a legal opinion and does not substitute for claim-charting against full reference texts and certified translations.