Picovoice
Script error: No such module "AfC submission catcheck".
 | |
| ISIN | 🆔 |
|---|---|
| Industry | |
| Founded 📆 | 2018 |
| Founder 👔 | |
| Headquarters 🏙️ | Canada |
Area served 🗺️ | |
| Products 📟 | Cheetah Cobra Leopard Octopus Porcupine Rhino Shepherd |
| Members | |
Number of employees | |
| 🌐 Website | https://picovoice.ai |
| 📇 Address | |
| 📞 telephone | |
Picovoice is a privately-owned technology company based in Vancouver, Canada. It develops Edge Voice AI technology that is private-by-design and runs across platforms including low-cost microcontrollers.[1]
Picovoice builds lightweight speech recognition engines using deep neural networks trained in real-world environments. Picovoice proprietary algorithm is developed based on the principles of transfer learning and hardware-aware training.[2]
History[edit]
Picovoice is founded in 2018 to build technology that runs even on boards with very little CPU power like Raspberry Pi and within web browsers that require very small models to run in order not to adversely affect website load times without sacrificing accuracy or user privacy.[3] Picovoice received a 500,000 USD investment from Founders Co-op in 2020. The company names its products after animals.[4]
Products[edit]
Picovice stack processes voice data locally on-device without requiring an internet connection, ensuring privacy and reliability.
- Speech-to Text: Cheetah is for real-time and Leopard is for file-based transcription.
- Wake word Detection: Porcupine Wake Word Detection Engine is for keyword spotting [5] Porcupine creates custom branded wake words similar to Sensory Inc TrulyHandsFree and Soundhound Hound.
- Voice Control and Command: Rhino, Speech-to-Intent Engine, is a large vocabulary domain-specific NLU engine to create voice commands.[6]
- Speech Indexing: Octopus is like Google’s search engine, but for audio files to search for keywords. Octopus indexes massive audio files, eliminates the out-of-vocabulary and competing hypothesis (e.g. homophones) problems.[7]
- Voice Activity Detection: Cobra detects the presence of a human voice within an audio stream and outperforms Google WebRTC. [8]
- Picovoice Console: Picovoice technology is offered through Picovoice Console, which is the platform for designing, training, and testing voice user interfaces instantly within a web browser. [9]
- No Code Platform: Shepherd is a no-code development platform for adding voice to microcontrollers.
References[edit]
This article "Picovoice" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Picovoice. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
- ↑ Thompson, Clive. "Worried About Privacy at Home? There's an AI for That". Wired. ISSN 1059-1028. Retrieved 2022-01-16.
- ↑ "Porcupine™ Wake Word Engine - V1.8 Feature Tour". picovoice.ai. Retrieved 2022-01-16.
- ↑ "Picovoice Launches Speech-to-Text Platform 'On the Edge'". Voicebot.ai. 2019-09-03. Retrieved 2022-01-16.
- ↑ "Arm Developer: Machine Learning: Alireza Kenarsari from PicoVoice on Apple Podcasts". Apple Podcasts. Retrieved 2022-01-16.
- ↑ "More on the Picovoice Toolkit". Hackster.io. Retrieved 2022-01-16.
- ↑ Kenarsari, Alireza (2018-12-27). "Picovoice's Speech-to-Intent Engine". Medium. Retrieved 2022-01-16.
- ↑ "Octopus: Picovoice's Voice Search Engine". picovoice.ai. Retrieved 2022-01-16.
- ↑ "Picovoice Cobra Voice Activity Detection Engine shown to outperform Google WebRTC VAD - CNX Software". CNX Software - Embedded Systems News. 2021-10-28. Retrieved 2022-01-16.
- ↑ "Picovoice's web console lets device makers create their own voice assistants". VentureBeat. 2019-12-04. Retrieved 2022-01-16.
