(intelligent Voice)’s CEO Nigel Cannings says the breakthrough came when he decided to see what would happen if he pointed a machine-learning system at the waveform of the voice data – its pattern of spikes and troughs – rather than the audio recording directly. It worked brilliantly.
Training his system on this visual representation let him harness powerful existing techniques designed for image classification. “I built this dialect classification system based on pictures of the human voice,” he says.
In design, they’re simple. They’re very traditional earbuds, with a small circular bulb that holds a few microphones and a processor. All the processing of sound is done individually on each bud, and they can be calibrated to each user's individual separate ears. Their software leverages tried-and-true acoustic techniques for noise cancellation and effects, but also machine learning algorithms able to adapt to your surroundings.
The app associated with the Here buds has three tabs: a volume knob, which is the master switch for the volume of your world; an equalizer, coupled with effects like echo and reverberation; and a tab split into Tune In and Tune Out, which are pre-made filters meant to either enhance the soundscape or cut it out tailored to certain situations.
For people rendered unable to speak by cerebral palsy, stroke, or traumatic brain injury, VocaliD’s software creates custom speech patterns that sound like the original person’s voice.
After recording whatever sounds a person can produce, VocaliD’s software sifts through its database of 15,000 volunteers’ voice samples for the closest match, based on factors like age and accent. The software blends the subject’s sample with the surrogate voice’s thousands of prerecorded sentences to turn typed text into speech.
To supplement his old-school hearing aid, he favors a $350 iPhone-linked earpiece made by Sound World Solutions, a hearing-hardware maker in Park Ridge, Ill., for whom he’s begun to consult. With the Sound World device on, he can amplify phone calls and streaming music as well as his surroundings. A third, $500 earpiece was custom-made by Ultimate Ears in Irvine, Calif., to help him detect a wider range of musical tones while composing. For restaurants and theaters, he has a $45 directional microphone that pairs with a $5 app to isolate desired voices. And for especially cacophonous places, he has spare $700 microphones, made by Etymotic Research in Elk Grove Village, Ill., that he can strap to companions.
When viewers tune into the Academy Awards Sunday, they can be forgiven for thinking the swelling music is coming from an orchestra hidden somewhere inside the 3,400-seat Dolby Theatre. The truth is a little less glamorous - they are a mile away, playing live at Capitol Records.
The sound is then piped through fiber optic cables back to the theater - in only 2.7 milliseconds.
"We just keep trying to get that latency down as close to zero, so that performers can hear exactly what the orchestra's doing and the orchestra can respond,"
Sound waves tend to travel every which way, meaning if you can hear something, you can usually be heard. This acoustic circulator focuses sound waves so they travel in one direction and blocks them from moving back.
Time interview with Jimmy Iovine and Dr Dre as they move into streaming music beyond their hugely successful headsets
And the younger generation had no idea what the music was supposed to sound like. That’s why both of us take great pride in the fact that this company is turning a lot of young people onto quality sound, when what they were getting before was really bad earbuds and computers where the speakers faced the table. Dre was very frustrated with that, and we talked about that quite often, because my beginning was as a recording engineer as well. So we’ve always shared that – it was our connection from day one, speakers and audio. And this came about and we did this, and we’re very proud of the fact that an entire generation now is turned onto audio.