Audio codec entropy and arithmetic coding patents: who holds the ground and where it is still open

The assignee ranking in this corpus is concentrated: the top five assignees combined hold 605 of the 995 records in scope, or 60.8%, and the top ten hold 794, or 79.8%. Names appearing in the ranking include Dolby Laboratories, Samsung Electronics, Dolby International, Fraunhofer, DTS, LG Electronics, Qualcomm, Ericsson, Electronics & Telecommunications Research Institute and Nokia. Because a handful of assignees hold most of the filings, a newcomer's freedom-to-operate review should start with these players' independent claims rather than scanning the full 100-company ranking evenly.

No — filing peaked in 2014 at 90 records and has trended down since, with the last fully comparable window (2021 to 2024) showing a drop from 31 to 19 records, a 39% decline. That said, publication typically lags filing by around 18 months, so the 2025 and 2026 counts in this dataset are still incomplete and should not be read as continuing evidence of decline. Anyone tracking this space should re-check the trend once the most recent two years have finished publishing.

All three are lossless compression techniques used at the back end of an audio codec, and this search deliberately groups them because patent claims often use the terms interchangeably or claim variants of each. Huffman coding assigns fixed-length code trees to symbols by frequency; arithmetic coding (and its context-adaptive binary variants) represents a whole sequence as a single fractional interval and generally achieves tighter compression; entropy coding is the umbrella term covering both plus related techniques. Patent claims in this corpus most often specify a particular entropy coding scheme (e.g., context-adaptive arithmetic coding) applied to a specific set of audio parameters, such as frequency-domain coefficients.

G10L (speech and audio analysis/synthesis) is the dominant class, appearing in 74.5% of the 995 records in this corpus, but it is not sufficient alone. H03M (coding and code conversion) covers 27.1% of records and captures filings that claim the coding method itself rather than the codec system, while H04N (pictorial communication) appears in 20.9% of records due to shared techniques with video compression. A thorough search should combine G10L and H03M at minimum, and include H04N, H04B, H04S, G11B, G06F and H04H depending on whether the target application involves broadcast, storage or stereophonic processing.

The clearest signal is the gap between G10L's 74.5% coverage and H03M's 27.1%: most filings treat entropy coding as a feature bundled into a larger codec claim rather than protecting the coding method on its own terms, which leaves room for narrowly drawn method claims that isolate a specific coding technique. Smaller classes such as H04H (broadcast, 4.0%) and G06F (general data processing, 4.3%) are comparatively under-claimed relative to the corpus size, suggesting less crowded prior art for entropy coding methods applied in those contexts. Any white-space claim should still be checked against the top-ten assignees' portfolios first, since they hold nearly 80% of the field.