← Back to live feed1 story
Tether has released QVAC TranscribePsy, a 254 million parameter speech-to-text model that runs on local hardware to enable offline transcription on smartphones and laptops. The model supports 25 European languages on laptops and 8 languages on mobile devices, and reportedly transcribed a 10 minute recording in 1.65 seconds on an iPhone 17 Pro Max's Neural Engine. Tether's QVAC division emphasized that local processing via the parakeet engine is critical for privacy as devices shift toward voice interfaces.
Sign in to suggest
Key sources
- SOURCE@qvac“on an iPhone 17 Pro Max's Neural Engine the 8-language build transcribes a 10-minute recording in 1.65 seconds”x.com
- SUPPORT@theblockco“Tether has launched a 254-million-parameter AI speech recognition model that transcribes audio entirely on-device without sending data to cloud servers.”x.com
- SOURCEmarketbrief.now