There is a lot of activity in the areas of speech recognition and "the Internet of Things", but a natural application of both has been relatively novel: helping the deaf and hard of hearing to take part in everyday conversations. SpeakSee aims to do this (after crowdfunding, of course) with an intelligent hardware design that minimizes the friction of the installation and allows everyone to communicate naturally.
It is intended to be used in situations where a hearing-impaired person needs to speak with a handful of other people – a meeting, a dinner conversation, asking for directions and so on. There are speech synthesis applications that can transcribe what someone says, but they are not really fit for purpose.
"Many deaf people ran into a huge obstacle by asking people to download the application and hold the phone near their mouths.These limitations in the interface meant that no one was continuing to "But because we designed our own hardware, we were able to customize it for the situations in which it will be used."
SpeakSee is simple to use: a set of clamp microphones is in a small charger box, and when the user wants to have a conversation, he hands these microphones to whoever wants to talk. The case acts as a wireless hub for the microphones and relays the audio to the smartphone with which it is paired. This audio is sent, transcribed quickly somewhere in the cloud and displayed on the phone of the deaf user.
However, from a critical point of view, each microphone also intelligently and locally accounts for its loudspeaker noise and background noise.
"Of course, pickups pick up the speech of several people," Hazelebach said. "So we've included sensors that tell the microphone where the sound is coming from, and the microphones are exchanging those values, so we can determine which microphone should pick up who's speech."
The result is a rapidly transcribed speech divided by a speaker, delivered quickly and with decent accuracy (there is always a compromise between the time of execution and how the process is). And no one has to do anything but wear a microphone. (They have a patent pending for this multi-microphone system.)
Hazelebach's parents are deaf and he grew up seeing how their ability to interact under ordinary circumstances was limited.
"As you can imagine, my parents were the first to test that," he said. "At first we had a lot of problems but we quickly started to engage with others, we wrote a post on a deaf blog, and 200 people registered, so we did field testing with groups in the United States, as well as in the United States and the Netherlands. "
At present, voice recognition in English is significantly superior to that of Dutch and other languages, so the transcripts will be better for older ones, but the devices should work with one of the 120 languages supported by the cloud. The transcript is free for up to 5 hours of audio per month, after which it is a $ 10 / month subscription. But if it works, it can be worth more than worth it.
The team has a finished prototype, but is looking for crowdfunding to start production. "We need to improve the electronics to meet specifications, such as battery life.We plan to ship in February 2019," said Hazelebach. Preorders are set at $ 350 for a dock and three pickups.
The usual warnings (mainly "emptor") apply when one supports an Indiegogo-type campaign – but at the very least, having spoken to the creator, I'm pretty sure that it is a real product that needs to be boosted to the market.