Speech Recognition & Voice AI Patent Landscape 2026
Speech Recognition & Voice AI Patent Landscape in 2026
Speech Recognition & Voice AI is a mature, highly concentrated field in which the top five filers dominate among the hundred largest applicants and Google holds a commanding lead over all challengers. Annual filing volume peaked in 2019 and has eased since, though publication lag means the most recent periods are still under-counted.
Google leads a concentrated field with a clear tier gap to challengers
Google LLC is the clear leader in Speech Recognition & Voice AI, holding the top position in the applicant ranking by a substantial margin over Microsoft Technology Licensing and Samsung Electronics, which occupy the second and third positions respectively.
The top five filers — Google, Microsoft Technology Licensing, Samsung Electronics, IBM, and Amazon — together account for 37% of the combined patent records of the hundred largest filers, signalling a highly concentrated competitive structure with a pronounced gap between this leading cluster and the remainder of the ranking.
| # | Applicant | Patent records | Share |
|---|---|---|---|
| 1 | Google LLC | 8,356 | |
| 2 | Microsoft Technology Licensing LLC | 6,194 | |
| 3 | Samsung Electronics Co., Ltd. | 5,753 | |
| 4 | International Business Machines Corporation | 3,741 | |
| 5 | Amazon Technologies Inc. | 2,488 | |
| 6 | Intel Corporation | 2,220 | |
| 7 | Apple Inc. | 2,170 | |
| 8 | Nuance Communications Inc. | 2,068 | |
| 9 | NEC Corporation | 1,755 | |
| 10 | Huawei Technologies Co., Ltd. | 1,520 |
| # | Applicant | Patent records | Share |
|---|---|---|---|
| 11 | Sony Group Corporation | 1,506 | |
| 12 | NVIDIA Corporation | 1,437 | |
| 13 | Panasonic Holdings Corporation | 1,319 | |
| 14 | Toshiba Corporation | 1,301 | |
| 15 | Qualcomm Incorporated | 1,249 | |
| 16 | Tencent Technology (Shenzhen) Co., Ltd. | 1,189 | |
| 17 | Koninklijke Philips NV | 1,096 | |
| 18 | Fujitsu Limited | 1,039 | |
| 19 | LG Electronics Inc. | 1,032 | |
| 20 | NIPPON TELEGRAPH & TELEPHONE CORP | 988 |
The leaders’ entrenched positions, built on deep G10L (speech and audio analysis/synthesis) portfolios, create meaningful barriers for new entrants seeking to compete on core automatic speech recognition technology; differentiation in adjacent spaces such as AI models (G06N) or domain-specific applications may offer more accessible entry points.
Filing counts for 2024 and 2025 are materially under-counted due to patent publication lag and should not be interpreted as reflecting the true pace of recent activity. Longer-window growth, applicant concentration, and technology-route coverage are therefore more reliable signals than the latest-year bar alone.
Volume peaked in 2019; core speech technology still dominates the mix
Two complementary views — the annual filing trend and the IPC technology breakdown — reveal both the life-cycle trajectory of the field and the structural concentration of activity in core speech processing classes.
Annual filing trend
Annual filing activity grew from 894 records in 2017 to a peak of 1,461 in 2019, then eased steadily through 2022 and into 2023. Figures for 2024 onward reflect publication lag and significantly understate actual activity; do not read the apparent drop as a continuation of the trend.
↗ Hover for values · click a bar to ask EurekaTechnology composition
G10L (Speech & audio analysis/synthesis) dominates the technology mix by a wide margin, reflecting the field’s deep roots in core acoustic and language modelling. G06F (Electric digital data processing) is a distant second, with G06N (AI models), H04M (Telephonic communication), and H04L (Digital information transmission) each holding smaller but meaningful shares — these adjacent classes are the likeliest sites for emerging competitive activity.
↗ Hover for values · click a bar to ask EurekaHighly cited patent families surfaced by the query
Citation-heavy patent families returned by the query. Use this section as citation context, not as a curated list of the most topic-specific patents.
Joint end-to-end spoken language understanding and…
A method includes receiving an utterance from an audio input device. The method also includes determining a context associated with the utterance. The method also includes providing the utterance as an input to a joint model for automatic speech recognition (ASR) and spoken language understanding (SLU), wherein the joint model operates in a single mode to… (excerpt from the patent abstract)
| # | Patent | Citations |
|---|---|---|
| 1 | Using Context Information To Facilitate Processing… | 1,495 |
| 2 | Universal IP-based and scalable architectures acro… | 899 |
| 3 | Mobile systems and methods of supporting natural l… | 885 |
| 4 | Voice user interface with personality | 875 |
| 5 | Consolidating Speech Recognition Results | 873 |
| 6 | Knowledge-based speech recognition system and meth… | 851 |
| 7 | Spoken language interface | 825 |
| 8 | Method and apparatus of specifying and performing … | 803 |
Ranked by total forward citations. Citation counts favour older and broadly cited patent families, and broad or adjacent patents may appear when they match the search scope. Treat this section as citation context, not as a curated list of the most topic-specific patents. Some patent titles may be shown in their original, non-English language where an accurate translation could not be guaranteed.
What the competitive structure means for R&D investment decisions
The combination of life-cycle stage, concentration, collaboration patterns, and geographic distribution shapes where incremental R&D investment is likely to be effective and where established players hold defensible moats.
Field in decline phase from a 2019 peak
The lifecycle evidence places Speech Recognition & Voice AI in a Decline stage, with annual filings easing back from the 2019 peak of 1,461 records. This signals that foundational core-ASR territory is heavily staked out and incremental filing in mainstream G10L subclasses faces a crowded, mature landscape. R&D investment is more likely to yield protectable positions in adjacent application domains or at the intersection with large language models and multimodal AI.
Lifecycle: DeclineFive players hold 37% of the top-100 filers’ combined records
The top five filers collectively account for 37% of the combined patent records among the hundred largest applicants, and all five have deep, multi-year portfolios concentrated in G10L subclasses. This degree of concentration means that freedom-to-operate analysis is essential before entering core speech recognition architecture spaces. The tier gap between the top five and ranks six through twenty (Intel at 2,220 down to NTT at 988) is significant but not insurmountable in adjacent technology branches.
High concentrationPhilips internal entities and IBM drive the most active co-filing
The most frequent co-filing relationship is between Koninklijke Philips NV and Philips Intellectual Property & Standards GmbH, with 60 joint records — primarily an intra-group coordination pattern. IBM is the most externally collaborative large filer, appearing in co-filings with its China subsidiary (17 records), IBM United Kingdom (16 records), Toyota Technological Institute at Chicago (6 records), and Ohio State University (2 records). Samsung Electronics co-files with the University of Montreal (6 records) and the SNU R&DB Foundation (4 records), indicating selective academic partnerships at the research frontier.
Intra-group dominantUS is the primary filing jurisdiction; Europe and PCT are secondary
The United States is by far the lead jurisdiction, reflecting both the headquarters concentration of top filers and the commercial primacy of the US market for voice AI products. Europe (EPO) and WIPO (PCT) serve as the main secondary validation routes. Canada and Australia follow at lower volumes, while filings in Singapore, New Zealand, South Africa, and the Philippines are sparse, suggesting limited defensive coverage in those markets.
US-centric coverageGo beyond the landscape: Eureka’s TRIZ Solution agent breaks down an R&D problem and returns patented concept solutions, each with a technical approach and cited patent & literature evidence.
| Applicant | Collaborator | Co-filings |
|---|---|---|
| Koninklijke Philips NV | Philips Intellectual Property & Standards GmbH | 60 |
| International Business Machines Corporation | IBM China Co., Ltd. | 17 |
| International Business Machines Corporation | IBM UNITED KINGDOM LTD INTELLECTUAL PROPERTY DEPAR… | 16 |
| Samsung Electronics Co., Ltd. | Université de Montréal | 6 |
| International Business Machines Corporation | Toyota Technological Institute at Chicago | 6 |
| Samsung Electronics Co., Ltd. | SNU R&DB Foundation | 4 |
| Koninklijke Philips NV | PHILIPS PATENTVERWALTUNG GMBH | 4 |
| Koninklijke Philips NV | Philips Norden AB | 3 |
| International Business Machines Corporation | Motorola Inc. | 2 |
| International Business Machines Corporation | The Ohio State University | 2 |
Co-filing pairs, ranked by the number of jointly-filed patent families.
Google leads in volume and momentum; most challengers are pulling back
The top two positions are held by US-headquartered hyperscalers, but their trajectories diverge sharply: Google is growing while Microsoft Technology Licensing, Samsung, IBM, and Amazon are all filing at materially lower recent rates.
Google LLC
Google holds the top position with 8,356 patent records and is the only major filer in this field showing positive recent momentum, with recent filings up 6% versus the prior period. Its technology emphasis is concentrated in G10L 15 (speech recognition), G06F 3 (data processing interfaces), and G06N 3 (neural network AI models) — the last of these signalling active investment at the intersection of speech and large-scale AI. This combination of scale and positive trajectory makes Google the benchmark against which all other players must be measured.
patent records: 8,356Microsoft Technology Licensing LLC
Microsoft Technology Licensing ranks second with 6,194 patent records, though recent filing activity is down 52% versus the prior period — a significant pullback suggesting portfolio consolidation or a shift in filing strategy following the Nuance Communications acquisition. Its technology focus spans G10L 15 (core speech recognition), G06N 3 (AI models), and G06F 40 (natural language processing), reflecting the company’s pivot toward large language model integration. The combined Microsoft entity (including legacy Microsoft Corp at rank 44) represents a substantial accumulated portfolio even as new filing rates ease.
patent records: 6,194| Applicant | Recent (3 yrs) | Trend |
|---|---|---|
| Google LLC | 673 | ▲ +6% |
| Samsung Electronics Co., Ltd. | 250 | ▼ -46% |
| International Business Machines Corporation | 64 | ▼ -60% |
| Amazon Technologies Inc. | 148 | ▼ -35% |
| Nuance Communications Inc. | 21 | ▼ -52% |
| Microsoft Technology Licensing LLC | 89 | ▼ -52% |
| Apple Inc. | 32 | ▼ -51% |
| Koninklijke Philips NV | 8 | ▲ new entrant |
AI-model and telecom branches are under-served relative to core speech
Against a dominant G10L core, several adjacent IPC branches show low relative share but plausible technical linkage to voice AI — making them worth monitoring for teams seeking less-crowded filing territory.
G06N · Computing based on AI models
G06N accounts for only 5% of the branch distribution, despite being the technical foundation of modern transformer-based speech and language models. The sparsity likely reflects legacy classification practices that route neural-network speech work into G10L rather than G06N, but applicants explicitly framing inventions around underlying AI architectures — such as attention mechanisms, continual learning, or foundation model fine-tuning for speech — may find less direct competition here. Entry path: claim novelty at the model-architecture level rather than the acoustic-feature level.
Search this in Eureka →G06Q · Business, commerce & admin data processing
G06Q holds only 2% of the branch distribution, yet voice AI is increasingly deployed in enterprise workflows — call-centre automation, voice-driven commerce, and compliance transcription. The relative sparsity in this class suggests that application-layer patents tying speech recognition outputs to specific business-process steps or transactional workflows remain an under-served area. Entry path: inventions that combine a recognised speech component with a novel downstream business-process step are more likely to find differentiated claim space here than in core G10L.
Search this in Eureka →How leaders differ by technology route emphasis
Strength of each leader across the main technology routes.
| Player | G10L 15 · Speech & audio analysis/synthesis | G10L 25 · Speech & audio analysis/synthesis | G06F 3 · Electric digital data processing | G10L 21 · Speech & audio analysis/synthesis | G10L 17 · Speech & audio analysis/synthesis |
|---|---|---|---|---|---|
| Google LLC | Strong · 1,794 | Emerging · 214 | Emerging · 289 | Emerging · 146 | Emerging · 159 |
| Samsung Electronics Co., Ltd. | Strong · 1,025 | Emerging · 163 | Moderate · 217 | Emerging · 84 | Emerging · 139 |
| International Business Machines Corporation | Strong · 1,007 | Emerging · 120 | Emerging · 61 | Emerging · 106 | Emerging · 96 |
| Amazon Technologies Inc. | Strong · 701 | Emerging · 116 | Emerging · 106 | Emerging · 49 | Emerging · 119 |
| Nuance Communications Inc. | Strong · 549 | Emerging · 53 | Absent | Emerging · 87 | Emerging · 49 |
| Microsoft Technology Licensing LLC | Strong · 436 | Emerging · 49 | Emerging · 48 | Absent | Emerging · 56 |
| Apple Inc. | Strong · 308 | Emerging · 44 | Moderate · 73 | Absent | Absent |
Frequently asked questions
The corpus covers 9,230 patent families in the Speech Recognition & Voice AI space. This is the universe used to assess concentration, technology composition, and lifecycle stage.
Google LLC is the top filer with 8,356 patent records in the ranking, ahead of Microsoft Technology Licensing (6,194) and Samsung Electronics (5,753). Google is also the only major filer currently showing positive recent filing momentum, up 6% versus the prior period.
The evidence places the field in a Decline stage. Annual filings peaked at 1,461 records in 2019 and have eased steadily since. Note that 2024 and 2025 data are materially under-counted due to publication lag and do not represent the full extent of recent activity.
G10L (Speech & audio analysis/synthesis) is by far the dominant IPC class. G06F (Electric digital data processing) is a distant second, followed by G06N (Computing based on AI models) and H04M (Telephonic communication), each at 5% of the branch distribution.
The United States is the primary jurisdiction, followed by Europe (EPO) and WIPO (PCT) as the main secondary routes. Canada and Australia follow at lower volumes. Coverage in Singapore, New Zealand, South Africa, and the Philippines is sparse.
The most active co-filing pair is Koninklijke Philips NV with its internal Philips IP & Standards entity (60 joint records), which is primarily an intra-group arrangement. IBM is the most externally collaborative large filer, with co-filings alongside its China entity, IBM United Kingdom, Toyota Technological Institute at Chicago, and Ohio State University. Samsung Electronics has formed selective academic partnerships with the University of Montreal and the SNU R&DB Foundation.
Ready to map your own patent landscape?
Join 18,000+ innovators using PatSnap Eureka to map any technology landscape: search 2B+ patents and papers, surface key assignees, and generate a report like this in minutes.
Disclaimer. This page is generated from PatSnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. PatSnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.