TechFlow News: On March 10, according to TechCrunch, Tether’s BrainWhisperer project achieved a 98.3% accuracy rate in converting brain signals to text during its latest testing phase. In the Brain-to-Text ’25 Kaggle competition—featuring 466 participating teams—the system ranked fourth with a Word Error Rate (WER) of 1.78%. Built upon OpenAI’s Whisper model and enhanced with LoRA fine-tuning techniques, the system employs a multi-model ensemble pipeline to decode cortical electrical signals into text. Concurrently, Tether is advancing cross-subject signal decoding frameworks and developing non-invasive brain-computer interface (BCI) devices. It has also released Brain OS, an open-source brain operating system built on the QVAC platform.
Add to Favorites
Share to Social Media
