When not harvesting online browsing data of nearly every person in the world in order to surround them with their content bubbles, content with being showered in advertising 24-7 in a highly invasive manner, all with the goal of increasing shareholder value and making yet another step toward a troublesome future threatening to take away some of the freedoms we have previously been taking for granted, Google likes to do some highly targeted good in order to make at least a few of its critics ambivalent about continuing to voice their perfectly legitimate concerns about allowing the American technology juggernaut into the country.
Cynicism aside, it's not the corporate beast known as Google or Alphabet that is primarily responsible for this small miracle but a handful of its researchers who have been focused on one thing and one thing only in recent times – improving the accessibility of another thing we take for granted – voice calls.
That's the short story behind Live Relay, a project Google Software Engineer Sapir Caduri introduced several hours back as part of the first day of the firm's I/O 2019 conference, the latest edition of its annual gathering aimed at developers and hardware partners. If you'd deem Live Relay an odd fit for an I/O launch – no arguing there, it is. However, it appears Google is now positioning this yearly event as more of a catch-all happening that includes a little bit of everything. It's not like anyone that's anyone among tech media hasn't been attending it for years now, so the company apparently figured it might as well try to make the most of that attention for once, especially given how the days of exclusively positive media coverage of Google are long over.
Alright, no more digressions taking away from Live Relay, a project that does a wonderful job at reminding everyone how magical tech can be when used for good. Ms. Caduri, who's credited with developing and implementing the original idea behind the initiative, came up with a solution combining speech recognition with a text-to-speech service, as well as a speech-to-text one. Using a dedicated app, a user is able to call someone disadvantaged by having their phones relay their speech in writing, then waiting for the person on the other side to respond via a text message that gets sent back to the source while also being processed by another algorithm that vocalizes it for more convenience.
While no references to DeepMind's amazing natural-language tech can be found among the publicly released Live Relay materials, the manner wherein Live Relay voices written messages appears to be pointing in that direction. On the other hand, Ms. Caduri, an Israeli-born graduate from the Tel Aviv University, did confirm the platform incorporates Smart Compose and Smart Reply, a pair of self-explanatory features meant to make the process of responding to a transcribed call as painless as possible. Yes. those are the very same capabilities previously integrated into the Gmail and its controversially discontinued counterpart – Inbox.
As an added bonus and contrary to Google's general policy regarding free mobile products, Live Relay isn't monetized and doesn't collect personal information, with all of your commands being processed locally, i.e. never being as much as pointed in the direction of the cloud. In other words, it functions via a traditional telephone connection, so it even supports scenarios wherein one conversation participant is using a landline (depending on the model).
Unfortunately, Live Relay still isn't widely available, with Ms. Caduri only confirming the team running the project will continue experimenting as it observes the curious contraption over time.