Automatic speech-to-text for groups finally a reality?

SL First has previously run articles about speech to text being available on phones and tablets but now a new system is leading the way and is able to cope with group discussions and even multiple people talking at the same time!

The new technology is known as Transcense and has been created by four US graduates from around the world. Transcense is an application for your mobile phone or tablet (Android/iPhone/iPad) that is able to translate group conversations into text, in real-time, and also shows you who said what.

Most of our readers will already know that when it comes to lip-reading, only around 25-30% of English words are able to be read accurately, so the other 70% is complete guesswork. This means it’s really hard for people to have a one-to-one conversation when relying on lip-reading, let alone a group discussion or meeting.

There are already existing solutions for this situation; for those that know a signed language, they can make use of an interpreter, and those who don’t sometimes choose to use a Palantypist. Transcense feel that they beat the competition by being available 24/7 and in return for a yearly subscription fee of around $350 (£220) you get unlimited use of the app. If using the technology in a work meeting or with family - only one person needs to have paid for a subscription in the group - the others can download the app and join the conversation for free. We asked the team about future pricing and whether they had thought about UK prices yet, Thibault told us that even the current US dollar prices are still being tweaked at the moment, so we won’t know final costs until the service launches properly – likely early next year.

They will offer a free service which they currently estimate will be 5-10 hours per month and beyond those hours, the user will be charged for the month. Thibault also says “Many solutions claim to provide autonomy to the user. However, these so-called solutions are not always available, are highly dependent on other people and are super expensive. Transcense provides real autonomy to its users.”

They recently launched their project on the crowd-funding website IndieGoGo and hoped to achieve a funding goal of $25,000 (£15,600) to allow them to continue and expand the work they’ve done so far. Amazingly, they’ve already passed that goal, currently standing at $32,000 (£20,000) with another 21 days to go.

How does it work?

Transcense is a mobile application that creates a conversation network by connecting multiple mobile devices and their microphones together so that they can capture all that is being said in the group. All of the speech information is sent to Transcense where it's quickly processed and returned in text to the screen of the deaf or hard-of-hearing person. They claim that this should be happening in almost real time - the words appearing on the screen around only one second after they were spoken. The deaf person is then free to choose whether they want to reply in spoken English themselves, or they can type a message and the system will speak those words for them.

The Team
The four men involved are Thibault Duchemin, Pieter Doevendans, Skinner Cheng and Noore Kadri; a mix of deaf and hearing individuals whose lives have all been influenced by deafness in different ways. They have also received feedback, support and advice from over 300 people, most of which are deaf and hard of hearing.

Transcense aims to be a 24/7 solution for deaf and hard of hearing people to engage in any conversation at any time, giving them full autonomy, and increasing their social and professional opportunities. If you’d like to help fund their campaign to break down those communication barriers and get early access to the mobile apps you can do that on their IndieGoGo page. You can also find out more information about them from their own website at www.transcense.com.