Episode 14 – Inclusive Voice Design

Voice experiences are extremely difficult to craft in a way that will be truly accessible and inclusive for all users. Problems can arise from things like cultural biases and local colloquialisms, social backgrounds, and accented language and terms, among others. In this episode, guest Diana Deibel joins to talk about how to raise awareness for these inclusive challenges and some tips on how to address them in your voice experiences.

Guest – Diana Deibel

Diana Deibel is a Brazilian-American VUI Designer and a Lead Designer at the digital product design consultancy, Grand Studio, in Chicago, leading their voice and future tech initiatives. She’s the founder of the Chicago chapter of the Ubiquitous Voice Society and a speaker at conferences and colleges across the country. This year, you’ll see her at SXSW and SpeechTEK, as well as a few other events. Over the last 7 years, she has led teams in content and VUI design, crafting voice-first omnichannel conversations, modernizing corporate IVRs and creating engaging chatbots.

Transcript

Jeremy Wilken
Welcome to Design For Voice. Today our topic is covering inclusive design and how we can incorporate all people in all walks of life into our voice experiences as we develop them. I’m your host, Jeremy Wilken. And today I’m joined by Diana Deibel, welcome to the show.

Diana Deibel
Thanks so much for having me.

Jeremy Wilken
Why don’t you give us a bit of your background in the voice space and how you came to be a voice designer?

Diana Deibel
Sure. So I am a Brazilian American voice designer. And right now I work at a digital product design consultancy in Chicago called Grand Studio. And at my current job, we specialize in solving kind of messy problems for pretty big companies. And most of the time, we’re either getting brought in to fix an underperforming product or service or create something from scratch. And it’s really kind of ambiguous. So usually what we’re creating is like these solutions that are rarely one channel or one type of design. And I get to work with a bunch of designers to create more like holistic systems, for which I’m usually a person jumping in for voice. And I kind of got into this through, I have a playwriting degree. And I worked in entertainment for a while. And I was doing production and script writing work for a bunch of different companies, including Blue Man Group and animal planet. And eventually found my way to Chicago, where I was doing a lot more of commercial work. And from there, I kind of lost lost a little bit of meaning in my life when I was making like Sara Lee turkey commercials, and found my way into health tech, where I was working as a content designer. And as soon as they had some voice products come in line, I started working on those and really found that that was like the perfect marriage of everything that I had done up into that point, that made me feel like I was delivering something of value to the world, and still getting to write dialogue all day long.

Jeremy Wilken
It’s fascinating. I love having a story where it’s bringing all of your skills together and finding this alternative path that still utilizes those skills in new interesting ways, which is, how many of us I think have found our way into voice over time by taking things that we’ve gotten from another place, and bringing them to the table to voice. So thank you for joining me today. And we were going to talk about inclusive design. But before we dive into that too much, I want to maybe define the term a little bit more. So what do you mean by the term inclusive design? And how do we think about that with voice?

Diana Deibel
Yeah, I think. So there, there is a definition for inclusive design, I tend to think of it like basically taking the opportunity to interact glued users at every design decision point that comes up. And it really the point of it is to enable people with diverse characteristics to use a product in a variety of different environments or circumstances, and really make something that, you know, we all sort of talked about, which is something that everyone can use. And I know when we think about voice in particular, there are some really great use cases for access and in particular accessibility in the way that we kind of traditionally think about it with impairments and being able to aid people in that way. But the thing about voice is that conversation is just so social. And it’s embedded with all of these cultural aspects that a lot of times we just take for granted. So things like slang, or colloquialisms or cultural references and touch points. Sometimes we have subculture memories or larger cultural memories that are embedded in our in our conversation and the manners that we use. So all of this kind of becomes its own package. And it can change from macro group to micro group. So if you’re literally not speaking to someone in their own language, then they’re not going to engage with you. And they’re not going to engage with your product, and they’re going to stop using it. And obviously, all of us want to make something that people use, and that makes their lives better. So it’s really important for voice to think about how you’re including or exclude people to make sure that the product that you’re making is getting used by the people who you want to use it.

Jeremy Wilken
I think it’s good to point out though accessibility is a piece of it. But a lot of times we think of accessibility as building a ramp into a building or color contrast on a screen. And those are physical limitations that folks may have that we want to support. But we’re also talking about the the cultural experiences that enable people to communicate and and those differences. So it’s it’s a bit more than just accessibility in that sense. And I like I like your definition. So thank you for giving us that outline. And you mentioned a few of them. So let’s dive into a little bit of detail on some of these. So you mentioned a few of the challenges. But what do you see the top challenges are that affect users when it comes to inclusive at invoice.

Diana Deibel
So I kind of break them out into three sort of categories for myself, one being physical challenges. So that’s where you’re thinking about your traditional speech impediments or even accents, but stuff that is more about the the actual voice or the forming of speech, and the way it gets delivered. Then there are cultural challenges, which is more kind of like what I was talking about before with the different words that you’re using and reference says, and manners and all that kind of stuff. And then there are platform or channel challenges. So thinking about access in terms of who are my users, and what platform do they use? Are they even in this channel? Or do they want to be and that kind of questioning can be, can be another path that you can go down?

Jeremy Wilken
I think the cultural ones are the most interesting that we we often do think about, all right, our accents in the physical intonation of somebody speaking to a voice technology as something that kind of the platform itself typically is going to take care of when we think of the voice assistance in particular, like Alexa and Google, because a lot of that is handled at at the core platform level. And then any third party experiences benefit from those technologies at the core. So let’s focus on the cultural ones. And give us a couple examples of what cultural inclusive design might look like.

Diana Deibel
So one is just like originality. So if you think about like, the way that we speak in different parts of this country, and I even thinking globally, if we just take the US. And we think about the way somebody speaks in the southeast, versus the way somebody speaks in the northeast, so we don’t even have to get off the east coast. It’s so different. So the pace at which they speak is much different. Northeast, and to speak much more quickly. And obviously, we have a whole pretty solid image, I think in all of our minds of the pace at which southerners speak. And there’s colloquialisms like my favorite one is, oh, bless your heart. We’re in the northeast, that is a much more aggressive way of saying bless your heart. And it just sort of reflects back the the existing social structure in each of those areas. And if you were to use the vulgarity from the north east, in the south, like that would just be way over the top. But if you were to say bless your heart in the northeast, and people would not quite follow necessarily what you’re saying. So like these, just the the pure like colloquialism bit of it changes region to region. And then depending on the group that you have in the region, it could, it could be something that exists for the most of that population, but in a subgroup of that population doesn’t exist. So thinking about? Well, there’s one, one time that I was designing something for some people who were in Ohio, and the persona we created was really friendly. And the idea was, this is going to be somebody who is reaching out via phone on behalf of a physician to help people get like their annual physical scheduled, and this is something that’s covered for everybody under all the insurances. during a time when I think the Affordable Care Act has just passed everybody, almost everybody was on insurance. So this was like, Okay, let’s get people in, get them a relationship with their physicians. And I had, instead of really doing my research, and this was a lesson in that, I went off of the data that had been given to me, which was from clinicians who were not necessarily boots on the ground. And I got the information that okay, this is a very basic, traditional Midwestern group of people, they like friendly, they like convenience. So yeah, like, write something around that, that sounds fine. And the language I used was, you know, I know you’re busy. So let me take this off your plate and get you scheduled for this appointment. Which I think under normal circumstances might have been fine. But the problem is that I didn’t do research. So I didn’t really understand who the users were. And there was a large population of immigrants from, I want to say, I know it was an Eastern African country, I don’t remember which one. And they had been the emigrated because of a food shortage. So the concept of somebody saying to them, I will take food off your plate was threatening, it was not friendly. And I quickly realized that that was we were seeing a lot of people dropping off, and we were able to track that. And realize that, oh, that language is not translating, I thought it was friendly. And it is actually just confusing to people. And, you know, confusing it best and threatening, at worst. So was able to change that. But I think that is another like point that you when you think you, you might know, kind of the macro group, you still need to take a look at what that group is really made up of, to ensure that you’re speaking to everybody in a language that everybody can understand.

Jeremy Wilken
That’s really, really interesting story and really difficult because you can’t possibly know every single person, right? And how do you collect that kind of information ahead of time to potentially cover every single case. And I think we have to acknowledge that there will certainly be gaps at the end of the day. But there’s lots of stuff that we can do to help identify these things ahead of time. So that’s a really good example of something that didn’t go so well. But what kinds of things can we do with our designs to counteract as we think about the scripts or our users at early and often how do we identify when we’re using colloquialisms or or phrases that may not make sense in the right context. Because often, there’s so second nature, we don’t always even know that they’re colloquialisms. Sometimes we think, Oh, that’s a strange thing to say, if you really step back and look at it, but it it makes total sense.

Diana Deibel
Yeah, I’ve actually I’ve taken myself through an exercise. And certainly after that I was like, hyper aware of every time I was writing something that wasn’t just straight up plain English, like as crystal clear as it could possibly be. I think I probably aired more on this night have less personality. So I’m finally starting to come back into the middle. I took myself through an exercise of writing out whatever sort of friendly phrase I thought it was in my head, and then writing out like five to 10 different ways that I could possibly say that. And usually what I would do that I would sort of look at it and be able to step back and be like, Oh, I see now there is a clear way to say that I’m always bringing in other people to is a great way to kind of check yourself because sometimes we you know, to your point, we can’t see our biases if they just exist. And until they’re pointed out to us, I don’t think we’re aware of them. And I would oftentimes check things. Think I try to use something like hunky dory or like some sort of Sarah Palin esque folksy ism, that was too far too far down that road, but I was after this, take food off your plate instance, where I don’t think I would have caught it otherwise. And then I read it out loud and was like, Wait a second. There might I don’t know that everybody’s going to understand that. And I ran that by just like people out in the hallway to see what their reaction was to that. And of like the five sort of random people that I pulled who all work in the same areas me live in the same areas me. There were a couple that didn’t know that. And I thought, well, that’s a pretty good indicator, because this is already in a in an area has a lot of commonalities with my background. So if this is not reading here, it’s definitely going to not read out in the world.

Jeremy Wilken
So I have a similar story where I was given a talk in Germany, and I use the term willy nilly, which some people will know what that means. But it just means kind of haphazardly, right, you just, you’re just sort of doing something without thinking very carefully about it. And so, I looked at people’s faces, and they’re all like their heads kind of altered a little bit like what, and I stopped was like, I’m sorry, I just used a term that none of you probably understand. Right? And it was really interesting, got that immediate feedback, but, and they had a little moment of fun with that, too. But what about wanting to use those kinds of things, though, for on purpose, right to build a brand or to build some kind of persona that you are, your brand tries to embody is Is it okay to still try to push that a little bit? Or how do you balance that?

Diana Deibel
Yeah, I think so. It depends on two things. One is your brand and who your brand is. And the other is, you know, what, what level of information are you trying to get across? Like if this is a crucial piece of information, and you need to tell people that they should not be like haphazardly invest investing their money and you say willy nilly investing their money, then that might be kind of a higher stakes situation where you want to pull back on the the personality and the way you’re describing it, to make sure that it’s clear versus it’s fun. And then if your brand is not aligned with that, then it doesn’t make sense to like this. This happens a lot. I think where we, my company is a lot of work with financial institutions. And banks are traditionally pretty straight laced, like there aren’t a lot of to use another phrase that is probably not understandable to everybody, loosey goosey, like, they’re just not very flexible and fun, then you don’t think of them as the guy in the backwards baseball cap at the party that you’re at, like, they’re definitely the dad coming home. So you, you want to take that under consideration when you’re, you know, like you would with any persona, a lot of times the marketing and brand departments have already done a lot of the work for you and defining who the brand is. And it’s just taking that list of characteristics and that place in the market, and translating that into a personality that makes sense for that. That place and that that work that the team has already done. So that you’re not creating a persona or a personality, or any kind of verbiage that would relate to those that fights against the that because that’s where people get confused. If all of a sudden their their banking app is saying like, What up, bro, I think I got your money. Like, that’s not great, that doesn’t make me trust that this bank knows what they’re doing, or that they’re actually doing what I want them to do with my money.

Jeremy Wilken
So it makes sense that the branding is there. I mean, not everybody has branding, when we think about smaller scale voice experiences, which is the majority of what’s being built today outside of what agencies might be building. And there’s still a lot of room for that growth. But a lot of people that I talked to are one person shops are building them kind of for fun on us on the side. And so it’s always a little bit interesting to see how much thought and how they can apply these lessons into their experiences and how they’re building them ahead of time without having to wait for things like you described where people falling off of the the experience or negative reviews on on the app stores or whatever, to be able to identify, oh, this is something that causing trouble, although is important to be able to track at least if you have that you you at least have a better chance of finding when you are losing people for possibly these kinds of reasons.

Diana Deibel
Yeah, I think um, so to that end, you know, just sort of knowing who you want to be listening to, this is always a good place to start. And I know everybody always wants to start with like, well, I want this to be for everyone. And that’s awesome. But it does help, I think, especially when you are a one person team of thinking about Okay, yes, I want this to be for everyone. And is there a way that I could? Like, are there specific groups that I can think about first, that might help me figure out sort of the scope of what I mean by everyone, because I think everybody has a different idea of that in their head. They also don’t think that you do have to make everything for everyone doesn’t always make sense to do that. So defining that for yourself. And there’s kind of gut check that my colleague Ilana Shallow. She and I used to work together, we would do this to each other all the time, we would just ask who’s here? Who isn’t? And are we okay with that. And I think that’s just a good place to start of like, it doesn’t have to be. It doesn’t make sense if you’re creating something for the for people who smoke like, I don’t know, if Marlboros making a voice app, or they really making it for everyone, are they making it for everyone who smokes. So like, it just, it doesn’t make sense to actually make it for everybody, you have to just define what that looks like for yourself, and then go listen to the way that they’re having conversations.

Jeremy Wilken
I think you’re right about if you understand that the users that you are really interested in. And then I think there’s still ways to try to address it in a way that’s still inclusive without maybe targeting, right that maybe there’s a difference there. And that comes with careful consideration of terms. And the one of the challenges with, especially with voice assistance, or even IVR is if someone calls into a phone line, or someone invokes it on their Alexa devices that you have no idea what their background is, they could be in a foreign country. Because you know, English is natively supported pretty much everywhere, even if they’re in a country that doesn’t have native language support. So you just don’t know the background in most cases. So, you know, you’ve got to maybe hold that back, I think is another possibility. You hold back some of that branding until you’ve established rapport or you established a foundation with that person, maybe ease into it, rather than throwing it all out there at the beginning. How do you think about that?

Diana Deibel
So I think that if you need to, you know, we all want to have some sort of kind of personality or brand with it. But again, trying to think about how clearly Am I stating this is really helpful. And thinking about the pace at which you’re speaking, is another way to be a little bit more inclusive. So, you know, like we talked about, people in the, at least in the US in the south, speak slower than in the north, if you have something that’s going on pretty quick clip, it’s going to be really difficult cognitively, for people who speak at a slower pace to really parse that and like, pay attention to it respond in a timely fashion, a lot of times, the response we get is that that feels really rushed to me, I couldn’t keep up with it. And likewise, on the flip side, you’ve got somebody who’s used to speaking a little faster, and the voice pattern is going much slowly, much more slowly than you are irritating them and they’re not patient with it, they want to get out of there. So thinking about like, Okay, well, can we how do we accommodate both of those groups. So those are pretty opposite sides of the spectrum. And a lot of times, just simple stuff, like, you could keep it at sort of the quicker pace but have more enunciation and articulation in the voice you’re actually using, which sometimes means using a voiceover artists instead of using the TTS, but it allows you to kind of accommodate in some regard a little bit of both. So that until you know who your users really are, like, at least where they are that you can, you can tune that a little bit more as you move forward. But at least it gives you a starting place where you’re you’re doing your best to accommodate everybody. And by the same token, yeah, you’re not going to get everybody’s language you’re going to get everybody’s reference points, the beginning, but you can question the ones that you’re putting in, even think about, like, is this a reference point that truly everyone has? One example of that is my husband’s friends were talking about Commando. And they were shocked that that like, none of the women in the room, particularly any other women of color had heard of watch this movie, like, some of us didn’t even know what it was. And they had to explain like, Oh, well, it’s just like Arnold Schwarzenegger movie from the 80s. It’s amazing. How have you never seen this? Like basic stuff like that, that may seem to you kind of like oh, yeah, this is a reference point that everybody has that same kind of idea of backing off of it, writing out those five examples or 10 examples of the colloquialism, same kind of thing with the reference points. Are there ways that you can make lists for yourself or run them by other people to really determine if those are are as inclusive as you think they are?

Jeremy Wilken
Do you have any other suggestions like that? So we’ve got writing out several variations of the same thing to see kind of the different ways that it can be said and find possible colloquialisms hidden throughout. Do you have other techniques to identify those things?

Diana Deibel
Yeah, I mean, this is a really, it’s a little creepy, but it works really well. But it’s eavesdropping on people. And I mean that like don’t go stalking people. But just like sitting at a bus stop, we’re in a coffee shop or someplace where real people are, particularly if you are trying to, you know make this as inclusive as possible, you might want to go outside of your normal spots to hear how different people speak to each other, and write down exactly what they’re saying. And the reason for that is thinking about, like an art, they tell you draw what you see not what you think you see. And it’s the same sort of application for any sort of dialogue writing, just write what people say not what you think they say. And writing down somebody’s conversation allows you to see where they cut each other off, where what kind of words they’re using, where they have non sequiturs, or refer back to something else earlier in the conversation. All that kind of stuff is going to inform not only the structure and the functionality that you’ll need in your voice spot in your system. But it also helps you figure out like, what even are the manners that I need to include here? What are people expecting out of the conversation, something as simple as allowing people to barge in is really kind of tricky, because if you think about it, like it’s just as interrupting. And it’s a power dynamic, and you, as a human usually feel like you are in the power position. And when the system gets switched over to being in the power position, it feels uncomfortable. I think that’s why most users, and I will include myself in this get very frustrated, when we’re using a voice application, and we get compensation repair messages that are like, I don’t understand, you need to rephrase that. I think that makes it feel like the human messed up that is sort of reinforcing that power position that the system has over the user. So you can just think about like, there are different instances where in different cultures where people will interrupt, and it’s perfectly fine. And there are others where they interrupted, it’s not so just considering what am I saying with this, and what sort of seems to be the norm and the culture that I’m designing for, in order to accommodate that.

Jeremy Wilken
I think this idea of the system and being uncomfortable with the dynamics of that is actually really, really important. Because it might be the same with another human, you might be talking your students, your talk to the principal, it might not be a good conversation. And there’s certainly power dynamics. And it’s not just true for students and principles. But in many relationships, there is a little bit of that, even in close relationships, there may be from time to time, and keeping track of that. And I honestly, I agree, I have those moments where it’s like, why can’t understand me, I just said it 10 times clearly, but it’s it’s missing out. And sometimes it blames me or It feels like it is. And there’s a lot of emotion to it. I think you have a list of questions that you have written up that are pretty good ideas for things that you can ask yourself as you’re working through a design or working through a voice experience to help identify if there’s anything that you might be able to improve about it for inclusivity. Why don’t we go through a couple of those and as examples of some of these questions and ways to review before we close out the show.

Diana Deibel
Yeah. So one of the ones that I do is when someone who’s never used a product, understand how to use us. Like that’s sort of the test that I think a lot of people refer to as the like kindergartener grandma test like what a kindergartener understand this with my grandma understand this. And I think the idea is just that, that novice, that first time user coming through, how are you really making this interesting and engaging and easy to use for them, which, of course, is like every design, initial experience, we all try to do that. There are some people who have been in this for a very long time. And there are a lot of people who have been in this for a very short time. And trying to understand how we translate some of this information that we have from places like engineering, or UX design, or content, writing of some kind, and translating all of this into the stuff that people who’ve been around doing this for a while. Understand, but even for them, they’re on these new platforms. And we’re all sort of figuring out like, everybody who uses this has only been using this for what, three years now something like that’s how long the..

Jeremy Wilken
2015, yeah three to four.

Diana Deibel
So I mean, our users, even when they’re super excited about it, heavy users are still figuring out with us how it all works. And we all are learning from each other. So really thinking about like, Okay, this is not second nature in terms of using an artificial intelligence to have a conversation. It is very common nature to have a conversation, though. So it’s writing that line between what does somebody expect from from a robot essentially? And what does somebody expect from a conversation? Because we expect a lot from a conversation and we don’t know what to expect, from bots,

Jeremy Wilken
We anthropomorphize our devices very easily, very subconsciously. And when social norms are broken, we’re quickly going to jump on them as failures. And I guess a lot of people from some of the research I’ve seen, actually have low expectations. So when they break social cues are things are suboptimal, for whatever reason, people are kind of like, okay, it’s just a bot. But that expectation is going to keep rising and getting higher and higher. And I think, as you described with these things, with inclusive it, that will continue to also increase as far as the number of people that you need to be able to accommodate in a given conversation and trying to balance that with your persona and your brand. And your overall experience is going to be more and more challenging. But some of these questions you if someone’s never used it before, how would they perceive it? You have another one about? What kinds of assumptions might I be making? Essentially? And those are really hard to tease out because you’re trying to answer a question that you should have thought of before you wrote it all down, I guess, then that’s, but that’s what makes humans different is that we can have thoughts about our thoughts and analyze it and maybe give it a day and see it in a new light. So I think you’ve got good advice here around all the things that we can do for inclusive it, and essentially just being more critical of what we write and sharing that with others to get feedback quick and early. And often, rather than waiting to see when you ship it, if people are falling off the experience at some point.

Diana Deibel
Yeah, I am, if you don’t have a user researcher in your life, I would highly recommend befriending one. And that was probably the most eye opening experience I had for me was really working with somebody who was trained in this kind of thing of, of assumptions and breaking things down and understanding who who the audience is investigating who they are, and teaching me that how I could look through even simple things like chat rooms online, I don’t even have to go outside, I can just see how people are communicating with from the comfort of my own home. And showing me where I was making assumptions on things really, really helped. So if you know anybody or like, can join a slack or follow people on Twitter, it’s so useful to have somebody who’s trained in that on your side and and helping you figure that stuff out.

Jeremy Wilken
Certainly, we’re at the close of our show here, the section I like to call endpoint detection. So before we get to kind of the last few questions here, let’s look at the top takeaway from today, related to inclusive it and voice design.

Diana Deibel
Yeah, so inclusive, it is I think is in specially in voice is really under gaining your audience and their lives, how they are, how they are behaving and how they are communicating. And I think for that, that just means if you’re in a position of power, create an environment of curiosity, where everybody can really be encouraged and empowered to ask questions, and listen to the questions that are being asked, and really have this this curiosity around what are we creating? And why are we creating it? And who are we allowing into this experience? And if you’re not in a position of power, then just really listening to your users. So you can bring that data to the table with your team and have some backup when you do post those questions and, and bring up these ideas.

Jeremy Wilken
Excellent. So do you have an interesting voice experience that you’ve had recently?

Diana Deibel
Yeah. I tried. Somebody recommended the listeners. It’s an Alexa skill. It’s so weird. It’s the it’s the craziest and like, most fun skill that I played with, because it’s just like, how I don’t even know how to describe it’s like creepy people listening in on you. But they want to talk about your feelings. And then there. They are just a wee and they’re everywhere. And it’s very bizarre, but so innovative, and I can’t stop playing with it.

Jeremy Wilken
That’s a new one. So check it out. But seems like they’re pushing the boundaries on that, especially the emotional front or the social front of voice there.

Diana Deibel
Yeah, they are. It’s it’s not like anything I’ve ever experienced.

Jeremy Wilken
All right. Another question. What resources do you recommend for people who want to learn about voice design?

Diana Deibel
My go to is always Cathy Pearls’ book, designing voices for interfaces. It’s so easy to read and I think really hits on all the the main points that you need going into to designing a VUI.

Jeremy Wilken
Its a popular suggestion. And finally, last question, how can people learn more about you and your work?

Diana Deibel
Yeah. So I’m at brand studio, and our website is grand studio. com. We’re also grand studio design on Twitter and Instagram. And I am also on Twitter and my own at Diana does this. So anywhere there you can find out about usually what I’m doing. I’m talking about who I’m having conversations with so you can eavesdrop on me to start a conversation. I love those.

Jeremy Wilken
Awesome. Well, thank you so much for joining today and talking about inclusive it. There’s a lot of good insights here and things that we can take and actually put into our voice experience in our designs. So thank you for coming and joining me today. It was a pleasure to have you Diana,

Diana Deibel
Thank you so much for having me.

Jeremy Wilken
Thank you for listening to today’s episode and if you liked the show, please rate us on your favorite podcast player. All the show notes are available on design for voice.com