Picovoice
On-device voice AI SDKs for wake words, voice activity detection, speech-to-intent, streaming and batch speech-to-text, text-to-speech, speaker recognition and noise suppression that run without a cloud connection.
About Picovoice
Voice features in apps and devices usually send audio to a cloud API, which adds latency, needs a network connection and moves user audio off the device.
Who it’s for
- Product teams adding wake words or voice commands to mobile, desktop, web or embedded devices
- Enterprises that need speech recognition to run locally for privacy or offline use
When to consider it
Consider Picovoice when voice features such as a custom wake word or local transcription must run on the device itself; the Porcupine wake word engine, for example, runs on Arm Cortex-M boards, Raspberry Pi, Android, iOS, major browsers, Linux, macOS and Windows.
Tradeoffs & limitations
- Every SDK needs an AccessKey from Picovoice Console; usage is checked against your account limits and resets every 30 days.
- Picovoice sells to businesses: the Free Trial is a one-time evaluation that cannot be extended, there are no free or paid plans for personal or non-commercial use, and paid pricing is not published (the pricing page sends you to sales).
SOTA overview · Documentation-based assessment · Sources & review method
Updates
No updates shared yet.
Discussion
Newest firstAsk a question or share how you use Picovoice.
Keep it helpful. Community rules
Loading discussion…
Sources & review method
Documentation-based assessment · Nov 1, 2026 · Prepared with AI assistance; not a hands-on benchmark.
- Picovoice official website
- Porcupine Wake Word documentation
- Picovoice FAQ: business model and pricing
Checked by SOTA · AI-assisted documentation review. Selection advice is our assessment; verify current requirements for your deployment.
Import history & original evidence
Picovoice official website
On-device voice AI SDKs for wake words, voice activity detection, speech-to-intent, streaming and batch speech-to-text, text-to-speech, speaker recognition and noise suppression that run without a cloud connection.
Based on official pages and announcements checked on 2026-11-01. No hands-on test, performance benchmark or popularity ranking is claimed.