Court Invalidated Zentian Speech Recognition Patent
The Federal Circuit ruled that the claims in U.S. Patent No. 10,062,377 are unpatentable as obvious.
Updated on Sept. 28, 2026 in Artificial Intelligence
Live Poll
Do you trust the current patent system to fairly balance innovation with existing property rights?
The U.S. Court of Appeals for the Federal Circuit has affirmed a Patent Trial and Appeal Board decision to invalidate six claims from a Zentian Ltd. speech recognition patent. The ruling settles a challenge brought by Apple Inc. regarding the scope of the patent's underlying technology.
Why it matters
This decision clarifies how technical terms like feature vectors are interpreted in legal disputes, potentially affecting future litigation over how speech data is processed into usable digital signals. It highlights the influence of prior art in determining the validity of AI-related software patents.
The court upheld a construction where the term feature vector includes both extracted and derived quantities, finding that prior art reference Jiang (U.S. Patent No. 6,374,219) used codewords that fall within this scope.
The players
Zentian Ltd.
An entity holding patents related to speech recognition and signal processing technology.
Apple Inc.
A global technology company that develops consumer electronics, including voice-controlled interfaces and signal processing stacks.
The details
The patent in question covers a speech recognition system that converts audio signals into feature vectors—mathematical representations of audio data used for machine learning. The court determined that the term feature vector, as used in the claim, was broad enough to encompass codewords found in previous research. By treating codewords as derived representations of audio streams rather than distinct entities, the court found the original claims obvious in light of prior art.
Timeline
September 25, 2026: The U.S. Court of Appeals for the Federal Circuit issued the decision.
The Tech Race
This ruling follows a pattern set by previous Patent Trial and Appeal Board reviews that reject patent claims found to be broad or covered by existing prior art. It underscores the difficulty of securing intellectual property for standard signal processing techniques that have been utilized in academic and commercial literature for years.
This decision reduces the legal risk for companies developing speech recognition tools by clearing broad, contested patents from the landscape. It does not immediately change any products currently available to consumers, but it sets a precedent for how future voice-interface technologies are evaluated.
The takeaway
The invalidation of these patent claims removes barriers for developers working on standardized speech recognition techniques. Interested parties should watch for future filings from Zentian Ltd. to see if they seek further review of the Federal Circuit's interpretation of feature vector definitions.
Further reading
For broader context on how patent law is shaping the software landscape, explore our coverage in Artificial Intelligence.
Source note: This article includes information reported by IPWatchdog.com | Patents & Patent Law.
Live Poll
Do you trust the current patent system to fairly balance innovation with existing property rights?
