Tether AI Research Launches On-Device Speech Recognition Model QVAC TranscribePsy

ChainCatcher report: Tether AI Research (referenced herein as Tether Data, S.A. de C.V.), a subsidiary of stablecoin issuer Tether, has released QVAC TranscribePsy, an open-source multilingual automatic speech recognition model. With 254 million parameters, the model runs entirely on-device, making it suitable for smartphones, smart glasses, and humanoid robots—without requiring audio to be uploaded to the cloud. Smart glasses can process audio near-real-time via Bluetooth or Wi-Fi, and the model can also operate on onboard robotic processors. Evaluations show a word error rate of 8.02% on the Open ASR Leaderboard, 6.58% on Multilingual LibriSpeech, and 21.56% on Common Voice. It outperforms Qwen3-ASR-0.6B, omniASR-CTC-1B-v2, OWSM-CTC-v4, and Whisper-Medium on Multilingual LibriSpeech. On six tested smartphones, it is the fastest model—2.6 times faster than the second-fastest and 5.7 times faster than Whisper-small. The Core ML version transcribes 10 minutes of audio in just 1.65 seconds on an iPhone 17 Pro Max, uses less than half the disk space of a tested Core ML competitor, and requires 1.7 times less memory.