Keen Research
Keen Research develops the KeenASR SDK, an on-device automatic speech recognition platform using proprietary deep neural network acoustic models, serving EdTech, enterprise/frontline workers, gaming, and healthcare developers across iOS, Android, Web, Unity, and custom hardware.
- Company typePrivate
- Founded2016
- HeadquartersSausalito, United States
- Headcount1–10
- GTM typeB2B
- OfferingSoftware
What Keen Research does
Keen Research (Keen Research Inc.), a privately held California LLC founded in 2016 by Ognjen Todic and headquartered in Mill Valley, develops the KeenASR SDK, an on-device automatic speech recognition platform for mobile, web, and custom hardware applications. The SDK is built around a proprietary Deep Neural Network decoder with in-house trained acoustic models — including specialized children's-voice models — and supports English (adults and children), Spanish, German, and French across iOS (Objective-C), Android (Java), Web (JavaScript/WebAssembly), Linux, and Unity (C#) targets. All processing occurs locally on the device, yielding a self-attested COPPA/GDPR/HIPAA-compliant architecture with offline operation, phoneme-level Goodness of Pronunciation scoring, contextual decoding graphs, and customizable language and acoustic models. Development tooling includes a cloud-based Dashboard for transcription, scoring, and configuration inspection during the development phase.
The company monetizes via annual SDK subscription licenses for mobile and web apps (no per-installation, per-user, or usage-based fees), per-device fees for custom hardware OEMs, and professional services for SDK integration, proof-of-concept development, custom acoustic/language model training, and porting to custom hardware platforms. A freemium trial SDK with GitHub PoCs and live web demos supports a product-led, developer-first go-to-market, supplemented by quote-based enterprise sales. Keen Research's named customers span EdTech (PBS Kids, Osmo, Rally Reader, StudyCat, Teaching Strategies, Hopalai, Almedia), enterprise and frontline workers (eoStar, Numina Group, Ambifi, MiraCheck), gaming and audio (AudioKit Pro, Novel Effect), media (ViacomCBS/Noggin, discontinued), and academic research (University of Louisville). The company is bootstrapped, has grown organically from one to approximately four team members, and continues to ship quarterly SDK releases, with a stated intention to add Natural Language Understanding capability on top of the ASR engine.
The principal risks for an investor are concentration on a small team with no external capital, dependence on the continued health of key EdTech customers (several of whose deployments have already been discontinued), and competition from cloud ASR leaders; the principal opportunity is the structural privacy/offline advantage in COPPA-regulated children's apps, healthcare, and frontline-worker use cases where cloud ASR is poorly suited.
Keen Research firmographics
Firmographics- Name
- Keen Research
- Legal name
- Keen Research Inc.
- Website
- https://keenresearch.com
- Company type
- Private
- Founded year
- 2016
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- Keen Research develops the KeenASR SDK, an on-device automatic speech recognition platform using proprietary deep neural network acoustic models, serving EdTech, enterprise/frontline workers, gaming, and healthcare developers across iOS, Android, Web, Unity, and custom hardware.
- Ownership category
- akta.pro rank
Keen Research industry classification
Industry- Product category
- Speech Recognition SDK
- NAICS
- Software Publishers (513210), Software Publishers (5132)
- SIC
- Services-Prepackaged Software (7372)
- akta.pro primary industry
- On-Device Inference Runtimes & SDKs (mobile/embedded) (HDAAAJAB)
- akta.pro secondary industry
- Developer Tooling & SDK Platforms (BPAMAOAD)
Keywords
Where Keen Research is headquartered
LocationHeadquarters
- HQ city
- Sausalito
- HQ country
- United States
- HQ region
- North America
Offices1 record
Markets served
Keen Research business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Infrastructure, Operations, Marketing or Sales
Revenue model
- SDK Licensing - Mobile and Web Apps: Yearly recurring fee model for mobile and web applications. License covers use of SDK in a specific product across all supported platforms (iOS, Android, web, Unity). No fees associated with number of installations, number of users, or usage volume.
- SDK Licensing - Custom Hardware Devices: Per-device fee model for custom hardware platforms. Licensing terms structured differently from mobile/web app licensing.
- Professional Services: Professional services including proof-of-concept development, SDK integration assistance, evaluation of specific task domains, customization of acoustic and language models, and porting to custom hardware platforms. Offered through Professional Services Agreement.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Subscription | Annual | Mobile and Web App Licensing |
| Unit Pricing | Multi-year contract | Custom Hardware Device Licensing |
| Freemium | Pay-as-you-go | Trial SDK |
Go-to-market motion2 records
Distribution channels3 records
Marketing channels7 records
Keen Research product offering
Product offeringCore offering
Keen Research develops and licenses the KeenASR SDK, an on-device automatic speech recognition (ASR) software development kit powered by Deep Neural Network decoder and acoustic models. The SDK runs natively on iOS (Objective-C), Android (Java), Web (JavaScript/WebAssembly), Unity (C# plugin), Linux, and custom hardware platforms, processing all audio locally on-device without sending data to the cloud. The product is sold via annual recurring licenses to developers and enterprises building voice-enabled applications for EdTech, frontline workers, gaming, healthcare, and consumer use cases, with additional professional services for integration, acoustic/language model customization, and custom hardware porting.
Product overview
Keen Research offers the KeenASR SDK as a unified on-device automatic speech recognition platform. The product portfolio includes platform-specific SDKs for iOS (Objective-C), Android (Java), Web (JavaScript/WebAssembly), and Unity (C# plugin), all powered by the core KeenASR engine using Deep Neural Networks. Each SDK includes ASR Bundles with acoustic models for English (with children-optimized variants), Spanish, German, and French. Developer tools include Dashboard, a cloud-based debugging and analytics tool. Professional services are available for SDK integration, customization, and custom hardware porting. The SDK enables offline operation with all processing occurring locally on-device.
Differentiator
Problem solved
Functional benefit
Brands
- KeenASR SDK: On-device automatic speech recognition Software Development Kit for iOS, Android, Web, and custom hardware platforms
- Dashboard
Products and services
- KeenASR SDK On-device automatic speech recognition (ASR) SDK based on Deep Neural Networks, sold to developers and enterprises building voice-enabled apps. Runs natively on iOS, Android, Web (JavaScript/WebAssembly), Unity, Linux, and custom hardware platforms, supports English (adults and children), Spanish, German, and French, and performs all processing locally on-device with no cloud dependency.
- KeenASR iOS SDK Objective-C framework (also released as XCFramework) for on-device speech recognition on iOS devices, supporting iOS 13 and later with Swift interoperability documentation. Sold as the iOS component of the KeenASR SDK license for developers building iPhone and iPad apps.
- KeenASR Android SDK Java AAR library for on-device speech recognition on Android devices, supporting Android 5.0+ (API level 21) and ChromeOS, including x86_64 ABI for Intel-based Chromebooks. Sold as the Android component of the KeenASR SDK license.
- KeenASR Web SDK JavaScript library that performs on-device speech recognition directly in the browser via WebAssembly, enabling web apps to run speech recognition locally without streaming audio to backend servers and to work offline when bundled with ASR Bundles. Sold as the web component of the KeenASR SDK license.
- KeenASR Unity Plugin C# interface plugin for the Unity game engine that targets iOS and Android, giving game developers full access to KeenASR's on-device speech recognition API with parity to the native SDKs. Sold as the game-engine component of the KeenASR SDK license.
- Dashboard Cloud-based development tool for the KeenASR SDK that syncs on-device audio recordings and speech-recognition metadata, enables transcription and scoring, and exposes recognizer configuration inspection. Free for customers during app development; disabled by default to preserve privacy compliance.
- ASR Bundles Acoustic model and language resources shipped with the KeenASR SDK for English (adults and children), Spanish, German, and French; available as standard bundles or custom-trained bundles built for specific acoustic environments, populations, or memory/CPU constraints. Sold as part of the KeenASR SDK license.
- Professional Services Engagement-based services covering SDK integration, proof-of-concept development for voice user interfaces, evaluation of specific task domains, customization of language and acoustic models, and porting the KeenASR SDK to custom hardware platforms, delivered under a Professional Services Agreement.
Quantifiable outcome
- 34% of students score below reading level in fourth grade (U.S. NCES). 54% of U.S. adults lack reading proficiency (U.S. DOE). Products using KeenASR help address literacy challenges through engaging, voice-enabled learning apps.
- +1 more outcomes
Companies that use Keen Research
Customer profileNamed customers15 records
Segments5 records
Ideal customer profiles4 records
Keen Research technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
AI capability2 records
Feature9 records
Keen Research partnerships and signals
Strategic signalPartnerships
One partnership is on record.
- Augmented Reality for Enterprise Alliance (AREA)coreContent partnership where Keen Research contributes articles to AREA website. Article 'Beyond the Headset: Empowering Frontline Inspections with the Untapped Potential of Voice' was originally published on AREA website. AREA is a leading voice in adoption of immersive technologies for enterprise.
Scale indicators3 records
Recent moves6 records
Expansion highlights7 records
Keen Research competitors and assessment
Company assessmentEmerging players
- Rhasspy: Open-source voice assistant toolkit for on-device speech recognition, primarily targeting home automation and hobbyist developers. Represents the open-source alternative in the on-device ASR space.
- Deepgram: Cloud-based ASR API with strong developer focus and real-time streaming capabilities. Competes on accuracy and developer experience, though its cloud-only model differs from Keen's on-device approach.
- AssemblyAI: Cloud-based speech-to-text API platform targeting developers with easy-to-integrate speech recognition models. Competes for the same developer audience though delivers ASR as a cloud service rather than on-device.
- Speechmatics: Provides ASR technology with broad language support and enterprise focus. Competes on accuracy and language coverage though primarily cloud-delivered rather than on-device.
- Rev.ai: Speech recognition API with focus on transcription accuracy. Competes in the ASR developer tooling space, though delivers cloud-based services rather than on-device SDKs.
Broad incumbents
- Apple Speech Recognition: iOS-bundled on-device speech recognition built into Apple platforms. Free and integrated by default for iOS developers, creating pricing pressure and friction for paid SDK alternatives on Apple devices.
- Cerence: Specializes in automotive and embedded speech recognition at scale. Broad incumbent in on-device ASR with focus on larger automotive OEMs, less specialized in mobile/EdTech.
- SoundHound: Voice AI platform offering speech recognition and natural language understanding across automotive, restaurant, and IoT. Broad incumbent with overlapping on-device capabilities and enterprise focus.
Others
- OpenAI Whisper: Open-source ASR model that can be deployed on-device for offline speech recognition. Represents a competitive threat to commercial on-device SDKs by offering a free, high-quality alternative for developers willing to handle integration themselves.
Direct peers
- Picovoice: Provides on-device voice AI SDKs (Leopard ASR, Cheetah STT) for mobile, web, and embedded platforms. Directly competes with Keen's on-device speech recognition architecture and developer tooling approach.
Market position
Strengths4 records
Weaknesses4 records
Competitive moat4 records
Key risks5 records
Key highlights6 records
Customer concentration
Keen Research social profiles
Digital presenceKeen Research compliance and trust
Trust signalCompliance2 records
Keen Research financial estimates
Financial estimateRevenue estimate
Valuation estimate
Keen Research leadership team
Management profileNumber of profiles
Profiles2 records
Keen Research funding detail
Funding detailFunding overview
Funding rounds
Investors
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Keen Research M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Keen Research
What does Keen Research do?
Keen Research develops and licenses the KeenASR SDK, an on-device automatic speech recognition (ASR) software development kit powered by Deep Neural Network decoder and acoustic models. The SDK runs natively on iOS (Objective-C), Android (Java), Web (JavaScript/WebAssembly), Unity (C# plugin), Linux, and custom hardware platforms, processing all audio locally on-device without sending data to the cloud. The product is sold via annual recurring licenses to developers and enterprises building voice-enabled applications for EdTech, frontline workers, gaming, healthcare, and consumer use cases, with additional professional services for integration, acoustic/language model customization, and custom hardware porting.
Is Keen Research a public or private company?
Keen Research is a private company. It is classified as founder individual operated bootstrapped and is currently operating.
When was Keen Research founded?
Keen Research was founded in 2016. It employs 1 to 10 people.
Where is Keen Research based?
Keen Research is headquartered in Sausalito, United States, in the North America region.
How does Keen Research make money?
Three revenue lines are on record. SDK Licensing - Mobile and Web Apps are the primary driver. The others are SDK Licensing - Custom Hardware Devices and professional Services.
Who are Keen Research's main competitors?
Emerging players on record are Rhasspy, Deepgram, AssemblyAI, Speechmatics and Rev.ai. Broad incumbents are Apple Speech Recognition, Cerence and SoundHound. OpenAI Whisper is listed as an others. Picovoice is listed as a direct peer.
Does Keen Research have an API?
Yes. KeenASR SDK provides platform-specific APIs for iOS (Objective-C), Android (Java), Web (JavaScript), and Unity (C#). The SDK includes createDecodingGraph methods for building decoding graphs, PrepareForListening methods for recognition setup, VAD parameters for endpointing control, and response callbacks for partial and final results. Trial SDK is fully functional but limited to 15 minutes per session. Developer documentation is at keenresearch.com/keenasr-docs.
What industry is Keen Research in?
Keen Research's product category is Speech Recognition SDK. Its primary akta.pro industry code is HDAAAJAB, On-Device Inference Runtimes & SDKs (mobile/embedded), with a secondary code of BPAMAOAD, Developer Tooling & SDK Platforms. Its NAICS code is 513210 and its SIC code is 7372.