Picovoice

Picovoice is an on-device voice AI and local LLM platform for enterprises and developers building voice assistants, AI agents, and speech applications in privacy and compliance sensitive environments.

Key capabilities include:

• Speech-to-text (including streaming STT) for real-time transcription and audio capture workflows
• Wake word detection and voice activity detection to start and manage hands-free experiences on-device
• Speech-to-intent (NLU) for converting spoken commands into structured intents for in-app actions
• Speech-to-index (voice search) for on-device voice search and speech analytics

Pricing: Freemium, with a Free Tier ($0 for up to 6,000 minutes/month or 72,000 minutes/year) and paid plans starting at $899/month, plus startup and custom enterprise options.

Notable for delivering a full on-device voice pipeline that runs across mobile, web browsers, desktop, and embedded targets, with deployment options that keep voice data local for predictable latency and data privacy. Cross-platform SDKs support downloading custom wake words, voice commands, speech-to-text models, and compressed LLMs without requiring AI/ML expertise.