Free APIVision & SpeechAPI keyOfficial
Picovoice Leopard & Cheetah
On-device speech-to-text, free for up to 100 hours of audio per month including commercial use
Picovoice ships on-device speech-to-text SDKs rather than a cloud API: Leopard for batch transcription of recorded audio and Cheetah for real-time streaming. Both models are under 20 MB and run entirely on the device, so audio never leaves the machine. You authenticate SDK calls with an AccessKey from the free Picovoice Console.
Free tier
Free tier: up to 100 hours of audio/month including commercial use, no card; paid plans from ~$999/moRate limits
No network rate limit (runs on-device); bounded by the 100h/month free allowance and device CPUAuth
API keyBase URL
https://console.picovoice.aiWhat you'd build with it
- Privacy-preserving on-device transcription where audio cannot leave the machine
- Offline voice commands in an edge or embedded automation
- Real-time dictation in a desktop app without per-minute cloud costs
Key endpoints
- GET
/ (Console AccessKey)Obtain a free AccessKey from the Picovoice Console to unlock the SDKs - POST
Leopard.process(audio) [SDK]Transcribe a recorded audio buffer locally - POST
Cheetah.process(frame) [SDK]Stream-transcribe live audio frames locally
Members
Members get the runnable recipe
Sign in free to copy the example request, see a sample response, and read the gotchas.
Tags
speech-to-texton-deviceofflinevoiceedge