US 10163451

I'll search for this specific patent number and check for any litigation/docket activity.

Let me verify the granted claims (which can differ from the published application) and check docket activity more specifically.

Let me try to pin down the granted (issued) claim set, which can differ from the pre-grant publication.

US Patent 10,163,451 — "Accent translation"

Bottom line: I could confirm the patent's bibliographic data from multiple sources (Google Patents, Justia, and an Amazon 10-K-adjacent/patent aggregator). I found no CAFC 2026 docket, no district-court case, and no PTAB proceeding naming this patent. One 2026 PTAB matter surfaced (IPR2026-00274), but it concerns a different accent patent (referred to as "the '745 Patent," an accent-conversion patent by different inventors/owner) — not US 10,163,451. Two claims-related caveats are flagged below.

Bibliographic data (high confidence)

Source: https://patents.google.com/patent/US10163451/en ; https://patents.justia.com/patent/20180174595

Abstract

"Techniques for accent translation are described herein. A plurality of audio samples may be received, and each of the plurality of audio samples may be associated with at least one of a plurality of accents. Audio samples associated with at least a first accent of the plurality of accents may be compared to audio samples associated with at least one other accent of the plurality of accents. A translation model between the first accent and a second accent may be generated. An input audio portion in a first spoken language may be received. It may be determined whether the input audio portion is substantially associated with the first accent, and if so, an output audio portion substantially associated with the second accent in the first spoken language may be outputted based, at least in part, on the translation model."

⚠️ Caveats on the claims

1. Granted vs. published claim text. The authoritative Google Patents text supplied for this patent is truncated before the claims. The claim text I located online is from the pre-grant publication page for US 2018/0174595 A1, which is the published version of the same application (15/387,038). Claim 1 in that listing includes a "determining first combined audio characteristics…" step that does not appear in the abstract's summary of the invention, which suggests the published/issued claim language may have been amended during prosecution. I could not independently verify the exact wording of the issued claims of US 10,163,451 against the official USPTO patent copy. Treat the claim summaries below as the best available reconstruction, not a verbatim reproduction of the granted claims.
2. Number-coincidence warning (per your "interpret literally" rule). The string "10163451" also appears as European patent application 10163451.7 (HTC Corporation, "Handheld electronic device," filed 2010-05-20) — a completely unrelated family. It is not the same as US 10,163,451 and must not be conflated.

Independent claims — plain language (20 claims total; 3 independent)

Claim 1 — System (independent; claims 2–4 depend from it)
A system with processor(s) and memory storing instructions to:

• receive many audio samples and tag each with one of several accents;
• compute "combined audio characteristics" for a first accent by combining the sample-level audio characteristics of samples tagged with that accent (i.e., build an accent profile);
• compare that first-accent profile against characteristics of at least one other accent;
• generate a translation model between the first accent and a second accent;
• receive an input audio portion in a spoken language; and
• determine whether the input is "substantially associated with" the first accent, and if so, output an audio portion in the same spoken language but carrying the second accent, using the translation model.

Dependents: second accent may be neutral / a modification / a combination of other accents (cl. 2); outputting adjusts pitch, tone, stress, or melody (cl. 3); input is speech by a first party in a call with a second party, and the converted output is played to the second party (cl. 4).

Claim 5 — Method (independent; claims 6–13 depend from it)
Same core pipeline expressed as method steps: build combined audio characteristics for a first accent; compare to other accents' characteristics; generate a translation model between the first accent and a second accent; receive an input audio portion substantially associated with the first accent in a spoken language; and output (same language) an output audio portion substantially associated with the second accent, based on the model.

Dependents add: determining the input's accent (cl. 6); accent determination based on geographic location of the capturing device (cl. 7); accent determination by matching input audio characteristics to the first accent's sample characteristics (cl. 8); computing FFT and/or MFCC on the input to identify its audio characteristics (cl. 9); neutral/modified/combined second accent (cl. 10); adjusting pitch/tone/stress/melody (cl. 11); model determined at least in part by artificial neural networks (cl. 12).

Claim 14 — Non-transitory computer-readable medium (independent; claims 15–20 depend from it)
The same operations as claim 1 encoded as instructions on a non-transitory CRM. Dependents mirror the method claims (e.g., determining the input's accent; geographic-location-based accent determination; adjusting pitch/tone/stress/melody).

Litigation / docket status (as of searches on 2026-09-28)

• No CAFC 2026 appeal and no district-court or ITC case identified that asserts US 10,163,451. Searches of Federal Circuit 2026 dockets, Amazon's litigation disclosures, and patent-case aggregators surfaced Amazon patent suits involving other patents (e.g., InterDigital, Nokia, BitHarmony, Broadband iTV, Dialect) but none naming 10,163,451.
• IPR2026-00274 exists in 2026 but targets a different patent ("the '745 Patent," an accent-conversion patent with different owner/priority) — do not attribute it to 10,163,451.
• Uncertainty statement: My tooling here is web search only, not a live PACER/Docket Alarm/PTAB query for this exact patent number. It is possible that (a) a very recent or sealed filing, or (b) a case captioned by a related family member rather than by this exact number, exists and simply did not surface. I have no authoritative information on any litigation, IPR, or CAFC appeal specifically involving US 10,163,451 as of today, and I will not assert one exists.

Practical/technical context

The disclosure is Amazon's "accent translation" concept — collecting accented speech (from phone calls, Alexa-style voice devices, recorded media), clustering it into accent sample sets, deriving accent profiles (pitch, tone, stress, melody; via FFT/MFCC), and training accent-to-accent conversion models (LSTM/neural-net based, continuously refined). Use cases listed include two-party calls (both parties speak the same language but different accents), live presenting, and playback of recorded media, plus translation to a "neutral"/combined accent. Media coverage of the pre-grant publication: TechCrunch (2018-08-02), https://techcrunch.com/2018/08/02/amazon-patents-a-real-time-accent-translator/

If you need the verbatim issued claim set, the authoritative source is the USPTO PatentCenter/Patent Public Search record for application 15/387,038 (or the granted PDF of US 10,163,451), which I could not retrieve with the tools available here.