Machine Translates Thoughts Into Speech

An anonymous reader points to this explanation of a brain-machine interface for real-time synthetic speech production, which has been successfully tested in a 26-year-old patient. From the article: "Signals collected from an electrode in the speech motor cortex are amplified and sent wirelessly across the scalp as FM radio signals. The Neuralynx System amplifies, converts, and sorts the signals. The neural decoder then translates the signals into speech commands for the speech synthesizer."

Comments

Hawking

[dreamchaser]

Imagine adapting this type of technology to other forms of input, such as a thought controlled dictation system. Imagine how much more someone like a Hawking could accopmlish.

Read Carefully -This Is How To Do It

[DynaSoar]

Most implant approaches use electrodes shoved in from the outside intending them to work immediately. That invasive technique leaves the person open to infection, and the neurons contacted tend to die fairly quickly, requiring yet another round of more of the same. This approach takes a long time, but eliminates the chance of infection (after the obviously necessary implantation) and lets neurons grow into and around the electrodes, so none of them producing signal are likely to die off soon, allowing long term contact and communication.

I'm sure there will be improvements on this, but this looks to me to be the first really viable direct neural signal collection technique.

"Five years ago, when the volunteer was 21 years old, the scientists implanted an electrode near the boundary between the speech-related premotor and primary motor cortex (specifically, the left ventral premotor cortex). Neurites began growing into the electrode and, in three or four months, the neurites produced signaling patterns on the electrode wires that have been maintained indefinitely.

Three years after implantation, the researchers began testing the brain-machine interface for real-time synthetic speech production. The system is “telemetric” - it requires no wires or connectors passing through the skin, eliminating the risk of infection. Instead, the electrode amplifies and converts neural signals into frequency modulated (FM) radio signals. These signals are wirelessly transmitted across the scalp to two coils, which are attached to the volunteer’s head using a water-soluble paste. The coils act as receiving antenna for the RF signals. The implanted electrode is powered by an induction power supply via a power coil, which is also attached to the head."

Rather than risking killing off speech center neurons in the implant process, they instead implant them in the pathway through which the speech center communicates outbound. Previous attempts by others went directly for the primary processing centers. This small change shows remarkable thinking foresight. I'd call this the first true hack in neural interfacing.

The only point of clarification I'd add is to say "through the scalp" instead of "across"; the latter more often implies a lateral vector. And the only point I'd request is, if only the scalp needs to be traversed, is the transmitter between the skull and scalp? It appears so but isn't stated s such in the paper (the PLoS article's URL is at the bottom of TFA). In any case, the FM transmission through the scalp does away with all the permanent jacks and sockets that SF and Hollywood have always used to signify brain/machine interfacing. With this one implementation, the future image of neural interfacing becomes something like a hair net with buttons sewn into it (we already have EEGs like this). Someone call Larry Niven. Wireheads will be buttonheads.

A future hack will almost certainly be to collect the signal wires running from the scalp to a second transmitter operating between the person and the machine. This will eliminate the direct connection and allow movement, including ambulatory data collection and processing. That not only makes possible testing in realistic situations, but also neural control of machine mediated locomotion for the paralyzed, without being restricted to the length of a cable. An obvious inclusion here would be a transmitter at the machine with receiver on the person, running the signals into the relevant muscle groups. This will also take some power induction that may be greater than the FM systems being used can handle. And are we not on the verge of getting wireless power induction for operating such devices, the same technology intended to refresh batteries and even run laptops?

A bit farther in the future will be to switch from spike analysis of neural firing to time/frequency analysis of synchronized activity such as EEGs examine. The former require computation that's commonly available. The latter require continuous wavelet analysis that s