Phase 1: Testing and feedback
The original Voiceitt app helps people with impaired speech to communicate and control smart home technology using their own voice. In phase 1 we asked early participants to test the app in ways that would be useful to them, including:
- Socialising, ordering food and drink or using public transport
- Playing music, audiobooks or the radio through a smart speaker
- Controlling smart lights, smart plugs, windows, blinds or TV
- Asking a smart assistant about the weather, time, news or sports results
The video below describes the testing phase and shows some of our participants demonstrating how they tested the app.
‘Nuvoic Project Video’ text description
Some of our participants show how they’ve been using Voiceitt’s app to control smart home technology like lights, blinds and TVs, ask Alexa about the weather or play music or the radio, using their own voices. The video starts with a series of photos of participants using the Voiceitt app in various different settings, including four young men, two young women and a middle-aged woman with her support worker. As the voiceover starts, the Karten Network logo and Voiceitt logos are displayed, and captions display what’s being said.
A young woman, Eleanor, uses the Voiceitt app on a tablet mounted on her wheelchair to turn her bedroom light blue. A young man, Karl, has the app running on a smartphone on the arm of his wheelchair, the screen reads ‘Movie time’, and you see his lights dim, his blinds go down and his TV turn on. A young woman, Sarah, sitting on her sofa uses the app on an iPad on her coffee table to ask Alexa about the weather. She then says “I really enjoy taking part in the Voiceitt trial. I have never been able to use Alexa before because it didn’t recognise my voice”.
Another young woman, Ashley, sits on a sofa and uses the app on an iPad resting on her lap to play Magic Radio through her smart speaker. A young woman, Lissa, sits in her office and describes how her team at Hill House Care Home have supported people to try out the Voiceitt app. There are clips of lots of several different participants using the Voiceitt app or smiling at the camera as the voiceover finishes.
We asked participants to keep in touch and share their feedback, which we passed to the Voiceitt team. Participants came up with loads of great suggestions on how the app could be made better or easier to use and Voiceitt used these to make some big improvements, like:
- Making it quicker to train new phrases
- Adding XP and achievements to make it more fun to use
- Making navigation simpler
- Enabling predictive text, wake words and switch access
Feedback interviews
Thirty-three participants took part in feedback interviews where we asked about their experiences of testing the app, what they liked about it and what they thought could be improved. We had lots of really useful suggestions, here are just a few quotes:
Friends don’t always understand and I have a lot to say
Using the interface was challenging at times due to my dexterity
My carer helped at the start, showing me what to do, and after that I carried on independently
Before the new update it would usually take 20-40 [repeats to unlock new phrases] but now it’s under 20
The recognition seems to work much better the more I use it
I was happy and surprised the first time it understood my voice
When it’s noisy it doesn’t work as well and I usually have to repeat a few times
When it works it’s great and it understands me better than Alexa
It makes it easier to do some things for myself
It helps me when I want to go out to places and communicate with new people
[I want it to] allow me to navigate through the app using my voice
Symbols for every phrase would be useful
I like how it allows me to use my voice to control technology
It’s amazing!
Phase 2: Voice Sampling
Something that lots of people asked for was continuous recognition – for the app to understand whatever they say and not just the phrases they’ve trained. To do this, Voiceitt needed lots more speech data from people with non-standard speech, so in phase 2 we asked participants to take part in speech sampling. Lots of people donated recordings of their speech to Voiceitt’s database, which were used to create the new Voiceitt webapp, launched in 2022.
Users train Voiceitt’s new app by making 1,000 recordings which are used to make a ‘model’ of their unique speech patterns. After the training phase, the app is designed to recognise whatever users want to say. It can be used to help with conversations, generate captions in a video call or dictate text into a document, email, message or social media post.
‘Claire demonstrates Voiceitt’s new dictation app’ text description
One of our participants, Claire, demonstrates Voiceitt’s new dictation app. She says the app allows her to write emails whenever she wants without needing her assistant available to help her. As she speaks, her words appear on the screen to show how the app works.
As Nuvoic draws to a close, the webapp is still under development, but the results look promising! We’re delighted to have been involved in Nuvoic and Voiceitt’s development work, and we hope that it will benefit many more people in the future. We’re no longer accepting new participants, but please visit voiceitt.com if you’d like more information, or if you or someone you know could benefit from Voiceitt’s technology.