Monday, March 15 2021

Automating Animation: Patentable; Automating Captions: Hold On

by James Watson

The U.S. Court of Appeals for the Federal Circuit has issued a decision affirming that patent claims to automating the process of making closed captions for audio-visual media are ineligible for patent protection. In Enco Sys. v. Davincia, the Court affirmed a lower court’s early resolution of a patent infringement lawsuit, granting a motion to dismiss on grounds that the asserted claims were invalid. The claims at issue involved automatically generating captioned text given just an audio-visual signal. Specifically, the claimed invention was software for identifying speech in an AV file, converting the speech to text, and synchronizing the text with the video to provide closed captioning automatically.

The Analysis

The panel applied the now-familiar test for subject-matter eligibility. First, it looked to whether the claims were directed to subject matter ineligible for patent protection. It concluded that the process of converting audio to text and associating the text with corresponding video was nothing more than an abstract idea. The Court relied on the patent’s own description of the prior art and the purported advance over that art. Specifically, the patent acknowledged that closed captioning was conventionally generated manually by a person viewing the AV media. When describing how the claimed system would differ from that process, the patent merely relied on “conventional speech-to-text software application such as such as Dragon Dictate®” to generate the text. This patent claimed priority to a provisional application originally filed in the year 2000, when speech-to-text software was still in the process of improvement. Despite this early priority date and the space for improvement, the patent did not describe any actual improvements to the process; rather, it merely applied existing software to the particular environment of generating captions. The Court criticized this lack of innovation, stating that using “known computer techniques for automation of known processes” amounted to nothing more than an unpatentable abstract idea.

Second, the Court analyzed whether there was anything more to the claims that might transform them into a patentable application. It reasoned that the claims did not “incorporate anything more beyond conventional computing hardware and software.” To support this conclusion, it again pointed to the patent’s own description of the claimed elements. There, the description of each element in the system acknowledged that purely “conventional” components and techniques were sufficient to generate and execute the claimed software. Highlighting one element, the Court noted that the claimed step of training the software to learn words in the signal was described in the patent as being accomplished using software already available at the time. Merely combining known software techniques on conventional hardware to achieve the kinds of efficiency gains that always accompany computer automation was insufficient to render the claims patentable.

Lessons in Contrast with Prior Decision

The technology in this decision was similar to the Court’s prior McRO decision in some ways. There, the court held that claims for automatically generating facial animations for 3D models to match a speech file were patentable. The primary difference was that the claims in the McRO patent set forth a specific, unconventional set of techniques for telling the computer how to animate the 3D model’s mouth. In particular, the panel overseeing that decision highlighted that the inventors had come up with a unique way for instructing the computer how to generate the facial animations that differed from the conventional, manual approach to doing so.

These decisions together highlight the importance the claims and description play in a software innovation’s patentability. For example, the claims must include at least one feature that a decision-maker can point to as being a genuine, significant, and preferably technical advance over the prior art. In addition, the patent’s own description of whether some feature is conventional or represents true innovation has significant influence on a decision-maker’s perspective when analyzing the claimed invention. When navigating the continually evolving expectations and best practices for complying with these standards, innovators in the software space should consider turning to a professional, such as the practitioners at TraskBritt, to increase the likelihood any patent covering their inventions will give them the competitive edge they desire.