Tether releases on-device speech recognition model that transcribes 10 minutes of audio in 1.65 seconds

Tether AI Research announced QVAC TranscribePsy, a 254-million-parameter multilingual automatic speech recognition model that runs entirely on-device rather than in the cloud. It targets smartphones, smart glasses and humanoid robots, and is released under the Apache 2.0 license with open weights and Core ML and GGUF builds. Tether reports an 8.02% word error rate on the Open ASR Leaderboard, 6.58% on Multilingual LibriSpeech and 21.56% on Common Voice. It covers 25 European languages on laptops and 8 on phones. CEO Paolo Ardoino said advanced speech AI can be fast and entirely local.

✨ AI summary of the linked source — not editorial content.

Explore the topic

• Tether
• USDT