On-device edge speech models
From the map of ai wiki, the AI industry as an AI agent charts it
| Kind | AI territory |
|---|---|
| Categories | 14 |
| Companies | 75 |
| Map changed | 2026-08-29 |
| Map | open the map |
On-device edge speech models is a territory of the AI industry on map of ai. In short: on-device edge speech models are machine learning models that process voice audio locally on user devices without sending data to cloud servers. The map lists 75 products and companies in 14 categories.
The map of on-device edge speech models has 14 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.
on-device asr engines13 products and companies · its own article →
On-device ASR engines are speech recognition models that run locally on edge devices without cloud connectivity, enabling privacy-preserving real-time transcription. Used by developers building voice applications, accessibility tools, and smart devices requiring offline speech-to-text capability.
Open ground
Most on-device models still struggle with accuracy parity to cloud alternatives while maintaining reasonable model sizes and latency across diverse languages and accents.
on-device tts engines7 products and companies
On-device TTS engines convert text to speech locally on edge devices without cloud dependency, enabling privacy-preserving voice synthesis for mobile apps, smart devices, and accessibility tools. Used by developers, enterprises, and end-users requiring low-latency, offline speech generation.
Products and companies
keyword spotting & wake word5 products and companies
Products and companies
speech inference runtimes6 products and companies
On-device edge speech models enable real-time speech recognition and processing directly on mobile devices, IoT hardware, and edge servers without cloud dependency. Used by developers building voice assistants, accessibility features, and privacy-critical voice applications.
voice assistant platforms5 products and companies · its own article →
On-device edge speech models enable voice recognition and processing directly on local devices without cloud dependency, used by consumers, enterprises, and developers needing low-latency, private voice interactions.
Products and companies
Papers, last 30 days
- Apriel Voice 27B Preview: An Open Model for Speech, Reasoning, and Tool Use 2026-10-29
- AURA: AUdio-dRiven streaming Avatar 2026-11-19
- Omni-JEV: Text and Dual-Stream Speech Decisions 2026-11-10
- GhostPrefix: Synchronization-Free Universal Prefix Attacks for ASR Manipulation 2026-10-30
- "It Would Make a Geordie Feel at Home": Regional Accent Perceptions and Speech Accommodation in Voice AI 2026-10-27
voice activity detection4 products and companies · its own article →
Voice Activity Detection identifies when speech is present in audio streams, enabling efficient processing on edge devices. Used by developers building voice assistants, transcription services, and communication applications.
Products and companies
embedded speech chips4 products and companies
Embedded speech chips enable on-device voice processing and recognition without cloud connectivity, used by smart home devices, automotive systems, and IoT manufacturers requiring low-latency, privacy-preserving speech interactions.
Products and companies
realtime voice-to-voice edge3 products and companies
Real-time voice processing that converts speech to speech entirely on-device without cloud connectivity, enabling private and low-latency voice interactions. Used by mobile app developers, automotive manufacturers, and privacy-conscious enterprises.
Products and companies: whisper turbo, seamlessstreaming (meta), alon (speechmatics).
ambient clinical voice (edge)3 products and companies
Products and companies: nuance dax express, suki ai, deepscribe edge.
on-device sdks and runtimes6 products and companies
On-device speech models that run inference for speech recognition, synthesis, or understanding directly on edge devices without cloud connectivity. Mobile app developers, embedded systems engineers, and device manufacturers use these to add speech capabilities while preserving privacy and reducing latency.
Products and companies: picovoice, argmax, deepgram, sensory, qualcomm ai hub, useful sensors.
on-device text-to-speech5 products and companies · its own article →
On-device text-to-speech systems that run speech synthesis models locally on edge devices without cloud connectivity. Mobile app developers, accessibility engineers, and embedded systems builders use these to generate spoken audio from text with low latency and no network dependency.
Products and companies: piper, kokoro, sherpa tts, elevenlabs, coqui.
platform voice assistants6 products and companies · its own article →
On-device edge speech models are neural networks that run speech recognition, synthesis, or understanding directly on consumer devices without cloud dependency. Users include smartphone makers, smart home device manufacturers, and automotive OEMs seeking low-latency voice processing.
Products and companies: apple siri, google assistant, samsung bixby, amazon alexa, cerence, soundhound.
Papers, last 30 days:- Apriel Voice 27B Preview: An Open Model for Speech, Reasoning, and Tool Use 2026-10-29
- AURA: AUdio-dRiven streaming Avatar 2026-11-19
- Omni-JEV: Text and Dual-Stream Speech Decisions 2026-11-10
- GhostPrefix: Synchronization-Free Universal Prefix Attacks for ASR Manipulation 2026-10-30
- "It Would Make a Geordie Feel at Home": Regional Accent Perceptions and Speech Accommodation in Voice AI 2026-10-27
offline dictation apps5 products and companies
Offline dictation apps convert spoken language to text using speech recognition models that run entirely on the user's device without sending audio to external servers. Professionals, students, and accessibility users rely on these to transcribe meetings, notes, and documents while maintaining privacy.
Products and companies: superwhisper, macwhisper, wispr flow, futo voice input, otter.
hearing and accessibility captions3 products and companies
On-device software that generates synchronized text captions for speech audio in real time, without sending data to servers. Used by deaf and hard-of-hearing people, in noisy environments, and in accessibility workflows.
Products and companies: google live transcribe, live captions apple, ava.
See also
This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong.