US 7818176

A technical analysis of US Patent 7,818,176 reveals a foundational patent in the field of voice-based advertising, detailing a system for delivering targeted ads in response to natural language speech.

Title: System and method for selecting and presenting advertisements based on natural language processing of voice-based input.

Assignee at Grant: VoiceBox Technologies, Inc. Subsequent assignment records indicate the patent is now held by VB Assets, LLC.

Inventors: Tom Freeman, Mike Kennewick

Filing Date: February 6, 2007

Issue Date: October 19, 2010

Abstract: The patent describes a system and method for choosing and showing ads based on the natural language processing of spoken input. A user's spoken words are received, and a conversational processor identifies a request. An advertisement is then selected and presented to the user based on this request. The ad can be delivered in a conversational, natural language format. The system also tracks the user's interaction with the ad to build statistical profiles, which helps in refining future ad selections.

Overview of Independent Claims:

This patent has four independent claims: 1, 14, 27, and 40.

Claim 1: The Core Method
In plain language, this claim describes the fundamental process. It starts with receiving a person's spoken request. A speech recognition engine then figures out the words in that request by breaking the speech down into phonetic sounds and generating a preliminary interpretation. A "conversational language processor" then establishes the context of the request. Based on this context, an advertisement is selected and presented to the user.

Claim 14: The "Adaptive Misrecognition" Method
This claim is similar to the first but adds a key feature: error correction. It outlines a method that includes determining if the system misunderstood the user's words. If a mistake is detected (for example, if the user quickly says something else, stops the request, or repeats themselves), the system is designed to reinterpret the original spoken words. This "adaptive misrecognition" is a crucial element for improving the accuracy and usability of the voice-based system.

Claim 27: The System for the Core Method
This claim essentially describes the physical system that carries out the method in Claim 1. It details a system comprised of an input device (like a microphone), a speech recognition engine, and a conversational language processor. The claim specifies that the speech recognition engine is configured to map the phonetic stream of a user's speech to generate a preliminary interpretation of the words. The conversational language processor then interprets these words in context to select and present an ad.

Claim 40: The System with "Adaptive Misrecognition"
Paralleling Claim 14, this claim describes the system that includes the "adaptive misrecognition" capability. It outlines a system with an input device, a speech recognition engine, a conversational language processor, and an "adaptive misrecognition engine." This last component is responsible for detecting if there has been a misinterpretation of the user's speech and triggering a reinterpretation. Recent court documents from a case in the District of Delaware confirm that claim 40 of this patent has been asserted in litigation.

A search of the CAFC dockets for 2026 did not reveal any pending cases specifically referencing US Patent 7,818,176. However, given the patent's subject matter and its assertion in district court, it remains a significant asset in the voice-assistant technology space.