[ti:Scientists Create Device to Turn Brain Signals into Speech] [by:蜗蜗牛小游戏网 www.liandama.cn] [00:00.00]更多听力请访问www.liandama.cn [00:00.04]Scientists say they have created a new device [00:04.28]that can turn brain signals into electronic speech. [00:10.92]The invention could one day give people [00:13.90]who have lost the ability to speak [00:16.88]a better way of communicating than current methods. [00:23.28]The device was developed by researchers [00:26.62]from the University of California, San Francisco. [00:32.12]Their results were recently published [00:35.00]in a study in the journal Nature. [00:39.01]Scientists created a "brain machine interface" [00:44.12]that is implanted in the brain. [00:47.96]The device was built to read and record brain signals [00:52.68]that help control the muscles that produce speech. [00:57.12]These include the lips, larynx, tongue and jaw. [01:04.04]The experiment involved a two-step process. [01:08.87]First, the researchers used a "decoder" [01:12.92]to turn electrical brain signals [01:15.72]into representations of human vocal movements. [01:21.80]A synthesizer then turns the representations [01:26.20]into spoken sentences. [01:29.92]Other brain-computer interfaces already exist [01:34.56]to help people who cannot speak on their own. [01:38.94]Often these systems are trained to follow eye [01:42.98]or facial movements of people who have learned [01:46.37]to spell out their thoughts letter-by-letter. [01:51.76]But researchers say this method [01:54.20]can produce many errors and is very slow, [01:58.52]permitting at most about 10 spoken words per minute. [02:04.01]This compares to between 100 and 150 words [02:09.80]per minute used in natural speech. [02:14.60]Edward Chang is a professor of neurological [02:18.16]and member of the university's Weill Institute for Neuroscience. [02:25.00]He was a lead researcher on the project. [02:29.12]In a statement, he said the new two-step method [02:33.04]presents a "proof of principle" with great possibilities [02:37.64]for "real-time communication" in the future. [02:41.68]"For the first time, this study demonstrates [02:45.87]that we can generate entire spoken sentences [02:49.76]based on an individual's brain activity," Chang said. [02:55.52]The study involved five volunteer patients [02:59.62]who were being treated for epilepsy. [03:03.44]The individuals had the ability to speak [03:06.40]and already had electrodes implanted in their brains. [03:11.88]The volunteers were asked to read several hundred sentences aloud [03:17.16]while the researchers recorded their brain activity. [03:22.60]The researchers used audio recordings of the voice readings [03:27.28]to reproduce the vocal muscle movements [03:30.68]needed to produce human speech. [03:34.41]This process permitted the scientists to create [03:39.04]a realistic "virtual voice" for each individual, [03:43.76]controlled by their brain activity. [03:48.20]Future studies will test the technology on people [03:52.16]who are unable to speak. [03:55.44]Josh Chartier is a speech scientist and doctoral student [04:00.56]at the University of California, San Francisco. [04:05.28]He said the research team was "shocked" [04:08.29]when it first heard the synthesized speech results. [04:14.00]The study reports the spoken sentences [04:17.32]were understandable to hundreds of human listeners [04:21.20]asked to write out what they heard. [04:24.25]The listeners were able to write out 43 percent [04:28.69]of sentences with perfect accuracy. [04:33.40]The researchers noted that [04:35.80]- as is the case with natural speech [04:39.04]- listeners had the highest success rate [04:41.92]identifying shorter sentences. [04:45.68]The team also reported more success [04:48.71]synthesizing slower speech sounds like "sh," [04:53.68]and less success with harder sounds like "b" or "p." [04:59.33]Chartier admitted that much more research of [05:03.80]the system will be needed to reach the goal of [05:07.20]perfectly reproducing spoken language. [05:10.47]But he added: "The levels of accuracy we produced here [05:16.36]would be an amazing improvement [05:18.88]in real-time communication [05:21.08]compared to what's currently available." [05:25.20]I'm Bryan Lynn. 更多听力请访问www.liandama.cn