aiqbench analysis
This research evaluates whether selective state-space models like Mamba can outperform recurrent decoders in intracortical brain-to-text systems using phonetic versus character-level targets. The findings indicate that while hybrid architectures remain competitive, traditional GRU-based baselines currently achieve superior performance on the Brain-to-Text '25 benchmark across both error metrics. Error analysis further reveals distinct failure modes tied to specific representation choices between articulatory and lexical boundaries.
Phoneme- vs. Character-Level Targets and Selective State-Space Models for Intracortical Brain-to-Text
arXiv:2607.26751v1 Announce Type: cross Abstract: State-of-the-art intracortical brain-to-text systems pair a neural-sequence phone decoder with an external language model. Two design axes remain underexplored: whether selective state-space models (Mamba) improve on recurrent decoders, and how the output target (phonetic vs.\ character) interacts with that choice. On the public Brain-to-Text '25 benchmark, we study a controlled 2x2 grid (GRU vs.\ hybrid Mamba decoder; phonetic vs.\ character targets) trained with a CTC objective under one reproducible protocol. The recurrent baseline remains strongest: the best phonetic GRU reaches 12.62\% PER and 21.19\% WER, while the best textual GRU after LM rescoring reaches 13.39\% CER and 26.28\% WER. The Mamba hybrid is competitive but does not surpass it. Ablations isolate architectural contributions, and error analysis shows representation-dependent failures: articulatory-like phoneme confusions vs.\ lexical and word-boundary errors.
Title:Phoneme- vs. Character-Level Targets and Selective State-Space Models for Intracortical Brain-to-Text
Abstract:State-of-the-art intracortical brain-to-text systems pair a neural-sequence phone decoder with an external language model. Two design axes remain underexplored: whether selective state-space models (Mamba) improve on recurrent decoders, and how the output target (phonetic vs.\ character) interacts with that choice. On the public Brain-to-Text '25 benchmark, we study a controlled 2x2 grid (GRU vs.\ hybrid Mamba decoder; phonetic vs.\ character targets) trained with a CTC objective under one reproducible protocol. The recurrent baseline remains strongest: the best phonetic GRU reaches 12.62\% PER and 21.19\% WER, while the best textual GRU after LM rescoring reaches 13.39\% CER and 26.28\% WER. The Mamba hybrid is competitive but does not surpass it. Ablations isolate architectural contributions, and error analysis shows representation-dependent failures: articulatory-like phoneme confusions vs.\ lexical and word-boundary errors.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
arXiv:2607.26751 [cs.CL]
(or arXiv:2607.26751v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2607.26751 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 239 words
