Table 1.

Percentage (± standard error) of electrodes showing significant similarity between neural data and the produced speech audio or stimulus audio, as established by the coherence method and the Roussel et al.’s (2020) method. The coherence method was run independently for the baseline, listening and speech production epochs. The Roussel et al.’s method used the entire duration of the session. Hierarchical bootstrapping was used to estimate the standard error.

Audio	Produced speech audio				Stimulus audio
Method Epoch	Coherence baseline	listening	speech	Roussel all	Coherence baseline	listening	speech	Roussel all
ECoG	0.2±0.1%	0.2±0.1%	52±6%	54±5%	0.4±0.3%	0.2±0.1%	1.4±0.7%	2.2±0.5%
Macro	0%	0%	42±8%	25±5%	0%	0%	0%	4±2%
Micro	0%	0%	56±7%	96±7%	0%	0%	0%	0.6±0.8%
DBS Lead	0%	0%	47±10%	59±10%	6±6%	7±6%	10±8%	12±8%
Blank	10±4%	12±5%	65±8%	71±6%	2±2%	15±8	3±2%	12±7%

Coherence indices are independent of anatomic location.