Tether's BrainWhisperer brain signal decoding accuracy reaches 98.3%, ranking fourth in the Kaggle competition.

CycleProphet · 2026-03-10T08:28:07+00:00

The BrainWhisperer project under Tether achieved an accuracy of 98.3% in converting brain signals to text in the latest testing, ranking fourth in the Kaggle competition. The system combines the OpenAI Whisper model and LoRA fine-tuning technology to decode EEG signals into text. Meanwhile, Tether is also developing a cross-individual signal decoding framework and non-invasive BCI devices.

CycleProphet

2026-03-10 08:28:07

Abstract generation in progress

Deep Tide TechFlow News, March 10 — According to Techcrunch, Tether’s BrainWhisperer project is currently in testing, achieving a 98.3% accuracy rate in converting brain signals to text. Among 466 teams participating in the Brain-to-Text '25 Kaggle competition, it ranks fourth with a 1.78% WER. The system is built on the OpenAI Whisper model, combined with LoRA fine-tuning technology, and uses multi-model ensemble pipelines to decode cortical electrical signals into text. Tether is also advancing cross-individual signal decoding frameworks and non-invasive BCI device development, and has released an open-source brain operating system called Brain OS based on the QVAC platform.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.