Picovoice
Picovoice 是一家來自加拿大溫哥華的人工智慧新創公司，專注於開發可在本地端（On-device）運行的生成式AI平台。成立於2018年，Picovoice 由一群科學家與工程師組成，致力於打造體積小、效能強、易於整合的邊緣AI解決方案，協助開發者在手機、電腦、嵌入式裝置甚至瀏覽器上部署AI模型，而無需依賴雲端資源。
核心產品包含：
‧ 生成式AI技術（picoLLM）：壓縮與推論小型大型語言模型
‧ 語音AI模組：即時語音辨識、語音合成、喚醒詞辨識、降噪、說話人辨識等
Picovoice 的 SDK 支援 Python、JavaScript、Android、iOS 等主流平台與語言，讓企業能更快將AI產品上市，並同時解決隱私、安全與成本等雲端部署的痛點。
Picovoice is a Vancouver-based AI startup pioneering on-device generative AI platforms. Founded in 2018, the company is led by a team of scientists and engineers committed to building compact, high-performance, and easy-to-integrate AI solutions for edge devices. Their technology enables developers to deploy AI models across smartphones, desktops, embedded systems, and even web browsers — all without relying on the cloud.
Key offerings include:
‧ Generative AI (picoLLM): efficient LLM inference, compression, and training tools
‧ Voice AI: real-time speech-to-text, text-to-speech, wake word detection, noise suppression, and speaker recognition
Picovoice provides SDKs for all major platforms and programming languages, empowering businesses to accelerate time-to-market while maintaining privacy, reducing latency, and avoiding the high cost of cloud compute.