Brain-to-Text Benchmark '24: Lessons Learned

TL;DR

The Brain-to-Text Benchmark '24 evaluates neural decoding algorithms for speech, highlighting ensembling and language model integration as key to improving accuracy, while architecture changes showed limited benefits.

Contribution

This paper summarizes lessons from the Brain-to-Text Benchmark '24 competition, emphasizing effective ensembling and training strategies for neural speech decoding.

Findings

Ensembling with large language models significantly improved accuracy.

Optimized training methods enhanced baseline RNN performance.

Advanced architectures like transformers did not outperform RNNs yet.

Abstract

Speech brain-computer interfaces aim to decipher what a person is trying to say from neural activity alone, restoring communication to people with paralysis who have lost the ability to speak intelligibly. The Brain-to-Text Benchmark '24 and associated competition was created to foster the advancement of decoding algorithms that convert neural activity to text. Here, we summarize the lessons learned from the competition ending on June 1, 2024 (the top 4 entrants also presented their experiences in a recorded webinar). The largest improvements in accuracy were achieved using an ensembling approach, where the output of multiple independent decoders was merged using a fine-tuned large language model (an approach used by all 3 top entrants). Performance gains were also found by improving how the baseline recurrent neural network (RNN) model was trained, including by optimizing learning rate scheduling and by using a diphone training objective. Improving upon the model architecture itself proved more difficult, however, with attempts to use deep state space models or transformers not yet appearing to offer a benefit over the RNN baseline. The benchmark will remain open indefinitely to support further work towards increasing the accuracy of brain-to-text algorithms.