Dr. Joan Palmiter Bajorek, Founder and President of Women In Voice, talked about how emerging technologies in speech and voice are causing a systemic change in the industry.
Voice and conversational AI are emerging fields centered around speech technology. Human-computer interactions with voice technology and the people behind it are changing.
In an interview, Dr. Joan Palmiter Bajorek, Founder and President of Women in Voice and CEO and founder of Clarity AI, talked about how emerging technologies in speech and voice are causing systemic change in the industry. She spoke about how we can overcome gender bias in voice technology and the role of enterprise in pushing the needle in the right direction.
Dr Bajorek is a technical advisor to several companies and startups. She has previously been a Senior Conversational Experience Designer at Nuance and a Principal User Experience Researcher at the University of Arizona. Harvard Business Review published her PhD research. She regularly contributes to Cambridge University Press, SoundHound, Adobe XD, and UXmatters. Her expertise includes voice products, bias in AI, platform disruption, and future multimodal and multilingual interfaces.
Excerpts from the interview;
Tell us a bit about the journey.
I am the founder of Women in Voice. I work in conversational AI. I love languages, and that’s how I started in this field. I have a master’s in linguistics and a PhD in speech and language technology. Linguistics is a multimodal interface and human-computer interaction, and that is how we are pushing innovation and optimization in all parts of the conversational AI tech stack.
How are emerging technologies in speech and voice causing a systemic change in the industry?
We have yet to see what we expect to be realized in speech and voice technology. It is one of the big things. We think about IoT and how far this has come. But the lights change when we think about the opportunity for voice and speech, that people can interact with their devices by speaking and have different outputs.
We are seeing a lot of people trying out different devices. In the United States, many people have smart home devices, whether they use them or not. But we see expectations changing in customer behavior, whether for an enterprise where people expect to speak into an interface or for day-to-day personal things to get information in a timely way.
You can find different places by using Google Maps. So, we certainly see a change in the industry, but it’s still in the early days.
How can we overcome gender bias in voice technologies?
We are benchmarking for different languages and dialects. How well do these systems support this? In my research, looking at these interfaces, I realized that we have huge gender and race biases in these systems. Frequently, that’s directly related to the data set being used and leveraged to build these systems. Unfortunately, little biases in these systems produce pretty compounding effects.
So, we are all working on this problem. It’s also who’s at the table to decide what gets deployed at the end of the day. That’s crucial for my research but also for the role of the enterprise. They are constantly trying different tech stacks. They’re benchmarking performance. The role of the enterprise in pushing the needle would be to continually look at these numbers, look at demographic data, and consider what demographics the technology is supporting or not.
So, it’s up to all. Consumers expect it will perform well for everyone, and enterprises need to invest in how the products are built, deployed, and optimized.
Tell us about your voice and multimodal augmented reality (AR) and virtual reality (VR) research.
We can use technology specifically to interact with devices. It’s a kind of voice that you speak to the device. It understands you, and it speaks back. But I see a world where – you ask the lights to turn on, on or off, and it is done.
When we think about leveraging different modalities, whether we are thinking about AR, VR, or speaking to it – this is enmeshed. Modalities can be leveraged in cool ways that are impactful for users. I’ve done the research for my PhD, looking at an immersive educational technology, platform, and product with an embedded speech recognition system. In this immersive environment, people were speaking into the interface and frequently forgot that they were not in that experience.
Especially as we see Web 3.0 and the Metaverse taking off, the idea that these modalities will be leveraged in different ways is a regular thing that users are discussing. It’s not necessarily easy to build, but users will expect it as these things get built out.
How can people and companies get involved with Women in Voice?
Everyone is welcome at Women in Voice. We have a narrative about joining the party; we would love everyone to attend. Women in Voice’s mission is to build community, amplify and celebrate the work and talent of women and gender minorities, and equip everyone with professional development and opportunities.
We frequently run the career development annual summit. Membership is something that people can sign up for. And we have a regular newsletter with many free resources for everyone worldwide.
Allies are welcome as well. But this community’s beating heart can be found across our social media. Women in Voice’s social media hits about 25,000 people daily across platforms. So, I recommend following, joining, and becoming a member.
One of the wonderful things about Women in Voice is that we are international by design. We have roughly 20 chapters in 15 countries and have just launched in Africa. These phenomenal chapters have local events, pre-pandemic and regular, that people could attend.
We don’t yet have ambassadors and chapters in Saudi Arabia. So I would be very interested in that, but it’s the local community. As different markets grow and evolve, women can be ambassadors in their regions. These are usually mid to senior-level women, three to five in a region to set up a chapter.
