Train, develop and deploy
custom voice features
in 15 minutes or less

Picovoice is the end-to-end platform for adding voice to anything on your terms

Start Building for FreeWork with an expert

Cost-effective audio transcription

Match the accuracy of cloud-based speech recognition with on-device ASR, with no compromises. Adopt models specific to your domain to boost accuracy even more. Transcribe at a fraction of cost while maintaining ownership of your data.

Leopard Speech-to-Text

Scalable voice search engines

Make audio files searchable and discoverable with direct audio indexing, without relying on transcription. Achieve superior accuracy and speed compared to any speech-to-text solution. Find keywords, including proper names in audio files.

Octopus Speech-to-Index

Always-listening yet private voice assistants

Train custom keywords and phrases and detect them in conversations and recordings. Wake word detection, also known as hotword detection or keyword spotting. Build hands-free, always listening voice experiences and detect keywords in conversations.

Porcupine Wake Word

Context-aware voice assistants with unlimited interactions

Fused voice recognition and NLU for intent inference from complex spoken commands, outperforming cloud NLU accuracy. Add custom voice commands to control devices, applications and websites.

Rhino Speech-to-Intent

Voice-activated products across platforms

Determine the presence or absence of human speech. The only cross-platform VAD that can run anywhere. Lightweight and efficient engine with superior accuracy.

Cobra Voice Activity Detection

Why Picovoice?

End-to-End Platform

From keyword spotting to automatic speech recognition, Picovoice meets all voice needs.


Picovoice publishes open-source benchmarks and makes its technology freely accessible to anyone.

Hyper customizable

Custom wake words, voice commands and speech-to-text models. Self-service training, instantly.


Runs anywhere, including tiny MCUs and load-time-sensitive web browsers with all modern SDKs.


Resilient to noise and reverberation. Outperforms cloud-based solutions with high margins.


All voice data is processed on-device. Intrinsically HIPAA and GDPR compliant.


Predictable and consistent response time. Eliminates unpredictable network delays.


100% of voice recognition is on-device. No concurrency limit, downtime or connectivity costs.