Juliet England: New help for group conversations from Holland

If you have any degree of hearing loss, you’ll know only too well the often intense frustration of not catching something in a group setting, from a work meeting to a night out with friends in a noisy bar or restaurant to a school or college class.

Last year, we wrote about Dutch start-up SpeakSee, which reckons it might just have found a solution to the problem.

Now the company has been in the UK, demonstrating its product ahead of the launch this month of its pre-order campaign. The private demo took place in London – and I invited myself along.

Put simply, the device consists of a set of up to nine microphones that allow speech to be transcribed to text in real-time using Artificial Intelligence (AI)-powered speech recognition. The mics get handed out to everyone present, and are attached to each speaker’s clothing below their mouth, while the person with a hearing loss reads the conversation via an app on a mobile device. Punctuation is included, and you can add your own ‘dictionary’ complete with personal jargon if you wish.

Different speakers are identified, and background noise eliminated, while the thing can also be plugged into conference calls systems and TVs.

The product is the brainchild of young Dutchman Jari Hazelebach, SpeakSee CEO and co-founder with business partner Marcel van der Ven. The company has seven employees, and the device has been in the development pipeline for 18 months.

Hazelebach has seen first-hand some of the frustrations of group chats, having grown up with deaf parents. He previously worked for a different assistive tech app before focusing full-time on his start-up, founded in 2016.

He explains that the device can handle up to 80 (mostly western) languages, including US English, with English and Spanish being the most accurately transcribed ones.

The business duo insist that previous apps weren’t able to handle group chats or background noise. Theirs can, thanks to beamforming technology which ensures that only sound coming from the direction of the speaker’s mouth is picked up.

The beauty of SpeakSee, they say, is that no apps have to be downloaded on to all speakers’ mobile phones, which can take time to download. Equally, not everyone wants to do that just to take part in a group conversation, especially if they don’t have a hearing loss themselves.

In contrast, SpeakSee is one, highly portable device that can be taken wherever you want to go and you can just hand out the mics and start to chat straight away.

When I ask Hazelebach whether even getting people to rig themselves up with mics before, say, a night out isn’t a bit of a chore, he replies: “Well, that’s the barrier to be overcome. People do have to be comfortable with that.”

And what about privacy? Aren’t users worried they might be recorded? (My brother and one good friend won’t talk to me on my textphone, for example, just because they can’t bear the third-party involvement, however innocuous.)

“We’ve developed it to ensure privacy, and encrypted the hardware as well as the software. There are no human third parties involved transcribing the speech to text, and no recordings or transcripts made of whatever is said.

“We’ll use anonymous data to improve the performance of the system, and this will be stated in our privacy policy. The user will have the option to disable this if they are not comfortable with us using data in this way.”

Interestingly, one young man at the demo said that, while he would be happy using this solution at work, he wasn’t sure about it for a night out with friends.

SpeakSee is, in some ways, still in development, with Hazelebach and van der Ven still working on how to handle, for example, what happens when conversations split up into multiple chats, which they can do, say, in a restaurant. The algorithm to do this is still being worked on.

There are also long-term plans to have the product use just one mic, and testing is ongoing for how the current multiple mics can work more intelligently together.

For now, the team is getting feedback from one British user in Cumbria, whose response has, says Hazelebach, “been really positive”.

Reservations worldwide are in the order of 80, including some from the UK. From June 18, SpeakSee will be on sale with an introductory price of $399 for a basic set of three mics and a charging dock, with the option to extend to nine microphones.

When I tested this with Hazelebach and one other speaker, it seemed to work as well as, say, TV subtitles, i.e. pretty well most of the time barring the occasional slip-up in transcription.

Because the app for mobile gadgets is also still in development, I saw SpeakSee on the tablet-sized screen that is currently part of the product. Speakers’ words can be viewed clearly, with differently coloured text and separately numbered mics identifying who is saying what.

And as long you can persuade everyone to take part, afford the asking price and remember to take the device with you everywhere, there’s no reason why SpeakSee shouldn’t have the potential to offer a genuine answer to those for whom group conversations can be a nightmare.

Read more of Juliet’s articles for us here.

Juliet England does freelance social media and PR work for cSeeker.