Peer-reviewed research
An instantaneous voice synthesis neuroprosthesis
Maitreyee Wairagkar et al., Nature, 2025 · DOI 10.1038/s41586-025-09127-3
Read the originalRead on October 17, 2026
- What it is
- A study of one man with ALS and severe dysarthria, implanted with 256 microelectrodes, whose neural activity was decoded into a synthesized voice in near real time. The paper reports how quickly the decoder runs and how intelligible the voice was to human listeners.
- How much weight it carries
- Peer-reviewed in Nature, with detailed timing and listener tests. One participant, and the voice was intelligible only often rather than consistently. The 10 ms figure is the decoder's processing step, not the full delay a listener hears, so it should not be read as end-to-end latency.
Facts taken from this source
- In the 2025 Nature study, the decoder turned the participant's neural activity into voice causally, within 10 milliseconds.
“decoding his neural activity causally within 10 ms” (Abstract)
- In the 2025 Nature study, the voice synthesized from the participant's neural activity was often intelligible, but not consistently so.
“The resulting synthesized voice was often (but not consistently) intelligible” (Abstract)