Tag Archives for " MIT "
The biggest thing that every musician and engineer lives in fear of is losing one’s hearing. To make matters worse, it happens naturally to all of us as we grow older, although its gradually enough that we can unconsciously compensate. That said, all it takes is a loud concert, or a sudden loud feedback, or cymbals in your ears on stage, and your ears will ring for days and some of your hearing may never return to the way it was. The good news is that there’s been a new breakthrough by research scientists at Harvard and MIT that might make permanent hearing loss a thing of the past.
We actually have two sets of auditory cells in our ears that are long and thin like hairs (see the photo on the left). Hair cells grow in bundles in the inner ear and are mostly concerned with balance. In the cochlea, the hearing organ deep in the ear canal, there are two kinds of specialized hair cells – outer hair cells that amplify pitch, and inner hair cells that convert sound into electrical signals sent to the brain. The problem is that each cochleae (one in each ear) has only about 16,000 hair cells, and once they’re damaged, they don’t regenerate – as in, gone for good!
That’s what happens with most mammals, but fish, birds, lizards and amphibians can have cochlear hair cells that die but can be regenerated in as fast as a few days. Coming back to mammals, mice and other small mammals also have these regeneration characteristics when they are newly born, and that’s where the researchers came in. The team managed to grow up to 11,500 hair cells from the stem cells of one baby mouse ear, which could bring hope to people of all ages already hard of hearing or getting that way.
These laboratory-grown hair cells aren’t perfect however, even though they appear to have many of the characteristics of actual inner and outer hair cells. They might not end up being fully functional, so the most immediate use for this new technique might be to create a large set of the cells to test drugs and to identify compounds that can heal damaged hair cells or regrow them and restore hearing.
Either way, that’s good news for just about everyone who depends on their hearing for a living. Now turn that music down!
I have great respect for sound designers in that they not only have to create effects that not only sound totally real but, in many cases, better than real. That last part is the key that will continue to keep them working despite a new algorithm from some MIT researchers that can independently add realistic sound effects to silent videos.
The researchers from MITCSAIL used artificial intelligence to enable a computer to learn the movements and surfaces occurring in a video and insert the appropriate sound effects. As you can see from the video below, the results are impressive.
The computer associates what it sees in the video with the appropriate sound from a database, then inserts it as needed. That should send a chill down a few sound designer’s spines.
In order to prove that the method was effective, the researchers did an A/B test on a on a number of test subjects. They showed one video that had the sound effects inserted using normal foley techniques and the other using the algorithm. In most cases, the test subjects failed to notice a difference between the two, and in some cases even preferred the one generated by the algorithm.
If you’re a sound designer reading this and fearing for your job, you needn’t be worried – yet. Although the algorithm shows promise, it has a long way to go. It’s only useful for very short clips as it tends to misfire on longer ones and play sounds at the wrong times, and it’s dependent on the sound library that it has available.
These problems will no doubt be worked out as development continues, but remember, only a sound designer good judgement can make something sound “better than live,” and judgement isn’t exactly the strong suit of computers, at least not today.