Welcome to the twentieth episode of Access On, the National Federation of the Blind's Technology podcast.
Episode
Listen to the twentieth episode of the Access On podcast (Browser).
Or listen on your preferred podcast platform.
Timestamps
The show is segmented by chapter, making it easy to move between segments of the podcast if you have an app or player that supports chapters. Below is what's on the show this week, and when you can hear it.
- Introduction 0:00
- Update from Google regarding SMS two-factor authentication 1:50
- Podcast sponsorship opportunities now available 5:45
- Huasong Cao from Agiga discusses the Echo Vision Smart Glasses 6:27
- Inaccessible tools on the job 43:21
- Can't get to the status bar of my iPhone via my Braille display 46:57
- Looking for advice on accessible TVs 48:48
- Tech tip, transcribing audio with ChatGPT 52:16
Transcript
Speaker 1: Live the life you want. Access On.
Jonathan Mosen: Welcome to Access On, the technology podcast of the National Federation of the Blind. Agiga's EchoVision smart glasses may be a strong new entrance to the wearables market for blind people. If you run a business and want to reach the Access On audience, now sponsorship opportunities are available. Another listener reflects on the difficulties people have using software to get their job done. And in response to an audience request, a tech tip looks at transcribing audio using chat GPT. It's good to be back with you from the Jernigan Institute in Baltimore, Maryland for our 20th episode.
Wow, where did the time go? That went fast. I'm recording this when there's a little bit of thunder going on outside the Jernigan Institute, so you may or may not hear a little bit of that as we get through the podcast. It does remind me of one convention, gosh, it must've been about 13 or 14 years ago, and I was trying to do one of those showcases where you go along to the event and you talk about the latest products that you are showcasing for the company that you are working for.
I was the one that was fronting that year and there was a massive thunderstorm and a rainstorm outside and it was so loud that I couldn't even hear myself. So, I doubted the audience could either. So, in the end I just stopped and said, "This sounds like one of those times when we just need to sing about our favorite things." See, that's The Sound of music reference right there.
So, if you have not seen the sound of music, then that's your homework. I think it's audio described out there now. Lovely movie, love that. I'm going to spare you the songs from it though and move on to something that the National Federation of the Blind has been tracking since we learned about this some weeks ago. And this relates to changes that the media have said Google were going to implement relating to how you authenticate when you log into Gmail on your computer.
According to those media reports, Google would be switching very soon to a method where you would have to scan a QR code in order to log into your Gmail account on your computer. Gmail is very widely used around the world and that means a lot of blind people use Gmail too. We enjoy a cordial, constructive relationship with Google. When we learned about this at the National Federation of the Blind, we immediately raised it and they have been great about keeping us informed about discussions pertaining to accessibility.
Google have come back with three key points, which I'll convey to you. One is that in the United States, this change will not be implemented in the near future. So, this means that if you currently receive a text message and you're in the United States, that is not going to change anytime soon. You'll still be able to receive a code via text message to log in.
Now, I would point out that that's not the most secure of methods, so it might be worth investigating other options, but if you really don't want to, it sounds like it's not going away here. The second point Google wanted to convey, for users accessing Google services on a mobile device, nothing should change. And the final thing for desktop or laptop users, Google recommends the use of alternative forms of two-step verification like Google prompt, pass keys or security keys. If you use a password manager, pass keys are a safe, secure and very simple way to log in once they are set up. I use them everywhere they're available and it's so easy to log into sites like Amazon and Uber and so many more sites now that use pass keys. If you use a password manager, and I highly recommend that you do, a good password manager should support pass keys.
1Password does, which is an excellent cross-platform password manager. They went through a little bit of an accessibility glitch for a while when they were changing the way that their apps were coded, but it's in pretty good accessibility shape now. And 1Password is very responsive to any accessibility bugs. They know about accessibility, they expressly care about it and it's very well maintained. 1Password also offers a directory of sites that they are aware of offering pass keys. So, you can go in there and set it all up with one password and it's very easy to do and once it's done, it's done. You don't even need to remember a username or password or anything like that and it's a very quick way to log in. Now, there may well be a learning curve for getting this set up, which not everybody is happy with, so you might want to consider Google prompt.
The way that Google prompt works is that if you are signed in on your mobile device to your Google account and you try to log in using your computer, then if you have Google prompt enabled, it'll just ping your other device, your mobile device that you're logged in on and it'll say you're trying to sign in on another device, is this really you? And you just double tap? Yes, it is really me and you're logged in. You don't have to enter any kind of code or scan anything. It is a very simple process as well. So, you might want to check out Google prompt.
If you're an Android user you are, of course, at an advantage because it's built right into the operating system there, but if you have the Google app on an iPhone or some other Google app, then it will work as well. Google prompt is a very simple way to log into Gmail and other Google services on a new device. That's the word from Google so far and we will keep you informed about any further developments.
From time to time, I've been approached by businesses who've said, can we sponsor Access On and get our message out there? Well, the answer is absolutely, now you can and you can sponsor the Nation's Blind Podcast, which is also produced by the National Federation of the Blind. If you'd like to find out more about this, you can go to nfb.org, then choose the link labeled publications and media and then you'll find a link that says publications sponsorship opportunities. And if you go into that page, you'll find all the details about the sponsorship packages that are available for the Nation's Blind Podcast and this one, Access On. So, nfb.org, then choose publications and media and then publications sponsorship opportunities.
Agiga's EchoVision smart glasses were causing a buzz at the recent CSUN Assistive Technology Conference. They're not out yet, but they promise an attractive pair of glasses with many blindness specific features at a reasonable price point. And to tell us about the company and the glasses and the exciting future they hope is ahead, Huasong Cao joins us. Welcome, it's great to have you on the podcast.
Huasong Cao: Thank you for having me, Jonathan. Glad to be here.
Jonathan Mosen: What's the story behind Agiga's founding and what inspired the development of these EchoVision smart glasses?
Huasong Cao: Good question. We started the whole journey from early 2024. I was an engineer at Google for close to 10 years. The recent project for me was about voice assistant, so bringing voice assistant to different devices. My co-founder, Xiaoran, who's the CEO, she's doing Amazon devices and early member of Alexa. So, we had some exposure.
Mostly Xiaoran had exposure to blind community when she was actually doing her masters and her lab was doing artificial injections to human eyes to help people to see again, but that commercial project didn't get very successful at the moment. We have elderly at home that have degrading visions and they're not categorized as low vision or blind, but we are exposed to some of these needs of vision. So, overall, when we were looking for something that we wanted to do, we wanted to bring our expertise to some field that we can really make an impact.
And considering some of the exposure to blind and low vision community, we felt maybe the background we had in terms of computer vision, in terms of voice assistant might be helpful to actually bring to this community and see what we can do. Around that time last year, the AI really is becoming a hot topic. The latest AI regarding LLM is becoming very hot and vision capability is being developed to really help describe this thing to some extent that's very useful. So, all these combined together, it felt that was a good moment. When I look back now, it's actually a good moment for us to actually start something. But yeah, there's a longer journey than what I described, but hopefully, that's a short description of what happened.
Jonathan Mosen: I remember that you were at our national convention last year and it seemed like you were doing quite a bit of market research at that time. I know you had a pilot going where you were doing some visual interpretation over WhatsApp and getting people's feedback. Was that really just gaining data about how blind people seek visual assistance and the kind of tasks they were seeking assistance for?
Huasong Cao: Yeah. That's one of the most important channel we got connected to, actual people, groups, users in this field. We have elderly at home, but again, they're not really categorized as low vision or blind. So, the first channel we actually went to was the local blind centers in the San Francisco Bay Area, but it's still limited. We have exposure to maybe up to 100 people, but that's still limited. So, we figured there's this online community that's much wider audience, more located widely across US, across globe. So, yeah, to answer your question, that's one of the channels we thought very helpful for us to actually get connected, understand if there's real use case that we can do and build on top of remote, either is in the way of virtual visual interpretation or it's something that the form factor of the wearable could really help.
We didn't really have a device that we can actually bring and give to people. Around that time, we learned there's Meta devices, there's Aira, Be My Eyes service, so we tried to survey what's out there and combining what we can do in terms of our knowledge and time. That seems to be a good channel for us to really connect to what we can find.
Jonathan Mosen: Can you describe then what these Agiga glasses are like? Can we start with a physical description of them first?
Huasong Cao: I hope I can describe it well, but I think the highlight or summary really, it looks like a regular type of glasses that the initial version would look blackish, hopefully stylish to many people. It has this attachment which is a sunglass shade that can be magnetically attached to the front of the glasses and it hopefully is lightweight to many people, that's our goal. And also, it has this capability to connect to USB-C battery bank if needed. I think that's pretty much it when looking. There's definitely these circuits and cameras inside, which is the core part of what we can build on.
Jonathan Mosen: What are the key features of the glasses that differentiate it from the Meta smart glasses? Because I've never seen anything go quite so viral in such a short time as the way the Meta smart glasses have found their way into so many blind people's lives. So, they've done very well. It sounds like what you are describing is an equally attractive product, physically attractive, but from what I can gather, they have a lot more blindness specific features on them.
Huasong Cao: Right. There's definitely a lot of attractive parts and features in Meta Ray-Ban and we're learning a lot from it. What we have in terms of flexibility is that we are basically building it with the blind and for the blind community. So, we have that flexibility to basically tune and optimize the glasses for this group and the community. We've heard a lot of feedback, especially during CSUN regarding what people like. So, I'm going to kind of quote what people like from there.
One biggest one is the description of people in very detail, so it will describe myself as East Asian descendant, black hair, eye color, if I'm wearing glasses or hats, some color of the shirt I might be wearing. And if it can recognize me in that moment it will say this is someone and recognized things like that. That's basically describing people and the face recognition.
The second one is OCR to the best it can be. So, it reads word by word, it reads from beginning to end trying to capture different formats of text, letters, books, bottles, things like that. That's the second feature that people really like. The third one is some of the AI description that's more proactive without the user needing to prompt the glasses to answer certain question. It just tries to capture what's in front and continuously proactively describing things. The last and maybe not least is the transit information.
This is also from our feedback that people do take transit and transit information would help people plan and navigate. That's another one that we provide through the glasses.
Jonathan Mosen: So, there is some AI involved in this depending on the task that you're performing. Correct? And I take it that you are leveraging one of the existing large language models to do that AI work.
Huasong Cao: Yeah, that's right. I guess we are riding the wave of the latest and greatest AI advancements. There's various models that we test, evaluate and try to use. We also develop some tuning and customized small models in-house to accommodate certain cases. So, we don't want to limit ourselves to a specific model and I guess we also have flexibility to choose when some model advance to the next level.
We noted for example, describing people, some model would work better than some other choices. Reading things, OCR, some model would work better. In other case, there's also requirement for certain cases to work offline. So, we basically try to be the wrapper of models and choose the one that best serves a specific purpose and the wrapper try to understand the intention of the user and then choose that specific model for that case.
Jonathan Mosen: There's a live video feed where you can ask a question of the glasses and in real time, they will come back and tell you about your surroundings for example.
Huasong Cao: Yeah. We're trying to achieve live, real time, but there's still some long way to go. Right now, as user feedback too, there is some still delay but we're trying to achieve something that the user doesn't feel there's big latency. Yes, the user can ask question, get feedback, ask follow-up question, things like that.
Jonathan Mosen: One of the things that these large language models who are trying this real time video thing have not cracked yet and it's probably just token availability or who knows, but at the moment, a blind person doesn't have the ability to say something like, I'm going to walk down the street and when you see a recycle bin at the curbside or in the road or whatever, I want you to tell me that you see the recycle bin and help me locate it. What you can do with something like the ChatGPT voice and camera functionality is say to it, do you see a recycle bin? And if it looks at that moment and answers you, yes, it will tell you roughly where it is. But it will not monitor for the presence of an object that may be coming up but is not in the view of the camera yet. Is that something that the EchoVision will be able to do?
Huasong Cao: We heard a lot of requests like the one mention. There's a similar one like can we scan through the bottles and tell me when I look at Tylenol bottle. Another example would be is there a empty chair? Let me know. Things like that. We're still evaluating. At the moment, we haven't achieved that capability yet. But I'm hopeful in the midterm timeline we could be providing something close to real time to answer questions like that. That's really something we want to offer to our user and I believe this is a strong demand. We've heard it so many times. So, I don't have a conclusive answer at the moment, but it's a clear goal that we are trying to get in midterm timeline.
Jonathan Mosen: Something that Access On listeners have talked to us about and it's something that resonates with me too is that they find if they wear hearing aids, the Ray-Ban Metas are difficult to work with, particularly with some sorts of hearing aids that aren't leaning directly against the speaker and this becomes an issue because at the moment those glasses do not permit you to connect an external Bluetooth audio device. Some hearing aids are Bluetooth compatible, but there's just no way to connect an external speaker slash headset device to those glasses. Is that something that the EchoVision is going to allow you to do to pair with another audio device?
Huasong Cao: The quick answer is yes. What that means is that the glasses can be paired with either a Bluetooth headset, earbuds or hearing aid. We've been hearing comments for both sides of these and we're still looking for more feedback and comments on this. One side is that people need this for better audio quality or for something to work with the hearing aid, so that's a need for this. On another hand, I've heard also that people, even though they're wearing hearing aid or something, they don't want the sun to all go to the hearing aid or the Bluetooth headset which can interfere or block their sense of the peripherals.
We probably want to test this little bit more when we have this in the field right now, the capability-wise we have that. It's supported. We actually demoed connecting to Bluetooth headset during CSUN and that helps us to actually remove some background noise during the demos. That worked well. Again, in the actual everyday use case, whether people want to connect these two hearing aids and whether that actually works as what people wish, that's still to be validated.
Jonathan Mosen: Well, that's the thing too, talking about demos that when you have that capability, it also allows people who are trying to demonstrate the glasses to pair them with Bluetooth speakers and that can help a lot. I'm sure you would've found in those noisy CSUN and NFB national convention exhibit halls.
Huasong Cao: Right. Yes.
Jonathan Mosen: You have integration with services such as Be My Eyes and Aira on the glasses. How seamless is that integration and how does it work in practice for the user?
Huasong Cao: The goal is really to provide our users what they need and remote assistance using human interpreter is something that's strong, strong need, That's why Be My Eyes and Aira are the top ones, high on our list to get integrated. The experience is more like we trigger the glasses and ask it to call agent, call volunteer, things like that and it will recognize that intent and try to connect with the correct one.
Aira for example, we believe it's very seamless in the way that we can basically use voice to control and at the same time get GPS location communicated over the agent. In terms of some of the features that's provided by these apps more than the human interpreter service, that we may not integrate it for our initial launch. We don't yet see there's strong demand or need for that if the glasses already provide another AI functionalities.
Jonathan Mosen: Is there an option for users to contact their own friends or family member to seek assistance in a private circle?
Huasong Cao: Yes. There's Be My Eyes way of doing the group. There's Meta Ray-Bans way of doing WhatsApp. It's not exactly the same way that we approach it, but it's similar that we can have some of the friends and family being in the circle and they can use the glasses to call them. But when I say it's not exactly the same way because we may not have access to WhatsApp in the same approach or for example, many people ask about Apple's FaceTime integration that might be still very difficult for us at this moment.
Jonathan Mosen: Can you talk a bit more about your philosophy behind this? It sounds like you are seeking to build a relatively open platform. There are an increasing number of players who want to provide computer vision for various projects and I came across them myself at the CSUN conference and they were expressing some concern that at the moment technology like Ray-Ban Meta is fairly locked down. There's no official SDK, there's no official API, so you've really got a hope for attention there. Are you seeking to provide that open place where someone can get a hold of the SDK or the API and appear on your platform?
Huasong Cao: Yes, that's one of our earlier goal and is still is our goal. When talking to a lot of our early feedback pre-orders or bata users or basically the groups that we connected to, I feel maybe openness is one important thing but when people talk about the openness, they really want the apps that they use every day be available. And the more we discover the details of that, I feel it's more about the functionalities that people need.
The functionalities meaning human interpreter services, AI description in certain ways, describing people, things like that. That's my understanding at the moment. When people talk about the need to have regarding openness of the platform on glasses, I think what people in this group at least, really looking for the functionalities that they have with various apps they already use. And if we can provide those through the glasses, I think it's already a good achievement.
But to answer your question regarding the open platform, it is something that we see as a differentiation part of our classes. We basically want it to be available for apps like, a lot of people mentioned GoodMaps, a lot of people mentioned other applications. We want to be able to work with these app owners and make our camera on the glasses available to those apps so that it makes a good combination of these features. So, openness is one part. We still need to provide some type of open SDK so that it can be used. It may not be the high listed priority at the moment, but it's definitely one of our goal and differentiation thing that we see.
Jonathan Mosen: Where is the camera on the glasses positioned? If you are looking dead ahead, is the camera seeing what's in front of you or is it slightly to the left or right of the glasses?
Huasong Cao: We have the camera on the right-hand side for our current build. If we clearly measure that vision scope, it should be a little bit to the right, but that's not something we really want the users to care too much about. We intentionally made the camera landscape and also wider angle, so from the purpose of basically capturing what's in front in terms of text and seeing it's already good enough. So, for practical purposes I don't think the user need to worry too much about whether something's too far to the left so the camera can't really capture it.
Jonathan Mosen: Tell me about the user interface on the touch panel and also perhaps we could talk a little bit about voice control, which I'm sure it has and how you interact with the glasses.
Huasong Cao: Yeah. Many of these are still being refined, so during CSUN time we actually didn't demo the touchpad even though we have it embedded there. So, the touchpad right now we are basically making it to serve a few purposes based on the context. One context is audio playback. It serves us basically volume control, pause, resume of the audio. In another context, which is the AI conversation, we wanted to serve as the TTS speed control and some of the other controls like go backward, forward in terms of the recognized text.
That's about the touchpad. Again, many of these are still being refined. We're still doing U.S. study of what's making the best sense. There's also discussion whether we can make all these customizable by the user. There's a possibility but we don't want it to be too confusing. So, I don't have the final answer regarding exactly what button and control means what, but we basically trying to tune it to make sense.
Jonathan Mosen: And what kind of battery life can users expect from these?
Huasong Cao: We target six hours' use for regular cases. I noted that people really love the AI description during CSUN demo, so that's a little bit dangerous in terms of battery life at the moment, but we don't have a stress testing result regarding how long it can survive if people continuously use AI live description nonstop. It's something we want to optimize further and our current solution for that is to have a USB battery bank possible to be connected while it's being used. So, regular target battery life for regular use is six hours and some optimization still needs to be done.
Jonathan Mosen: And how do the glasses connect to the internet or a smartphone? Are they using Wi-Fi or Bluetooth or a combination of both?
Huasong Cao: It's a combination of both. It connects to Wi-Fi by default and it connects to... by default in the sense that it needs AI description down, it uses mostly Wi-Fi which means that it can either connect to CSUN's venues conference Wi-Fi or grocery market free Wi-Fi or home Wi-Fi. For connecting and getting information from the phone, for example, music playback controls from the phone upgrading system, things like that, that's Bluetooth connected to the phone.
Jonathan Mosen: Since the device handles a lot of sensitive visual and audio data, what privacy protections can people expect to have in place once the product ships?
Huasong Cao: Yeah, this is something we're still learning. There's a couple of things we already have but we're still learning these and receiving more feedback and also improving on it. One area is about the history of questions, chat and descriptions. That history, we leave it up to the user to decide whether they want to keep it there or delete forever. We make it transparent regarding what we keep versus what people want to choose to delete.
The second part is that a lot of feedback regarding whether things can be done on device versus going into the internet, some of these unfortunately cannot be done on device so we probably give user choice if they don't want something to go internet, we try to provide some degraded experience. OCR for example, if people don't want the text or the letter to ever go outside the phone, then we probably provide a degraded OCR experience but it will stay on the phone without going anywhere. That's the second part. Overall, I think there's trade-off of the functionality and experience versus privacy, but our goal is to at least make sure it's transparent enough and people have option.
Jonathan Mosen: It looks like you're going all in on describing people in great detail. And looking at a video, it also appears that you are willing to do identification of individuals. How does that work and how do you ensure that people's privacy is respected with that feature?
Huasong Cao: I don't have a very good answer at this moment. I think the trade-off is always there. I think we leaning towards making it a little bit more aggressive in terms of trying to describe rather than conservative of not describing. So, when we try more aggressive obviously, there's going to be more risk for us that we get things wrong, get description wrong or going a little bit into the field of concept of privacy.
We are still doing a lot of evaluation of what the models produce in terms of describing things more accurately. There is a limit of what we can do in terms of that. I guess going back to the earlier thought regarding what we can is to provide more transparency of this, of what we are doing and give users a choice of what they want to use versus not want to use.
Jonathan Mosen: Can you share any details about the pricing structure of the EchoVision Smart glasses and when can we have them?
Huasong Cao: Thanks for the question. So, we already have pre-order available from our website. Current pricing is big discount still to our final price, so our final price is 599. Our current discounted price is 449. We try to basically make the devices affordable and try to reach wider audience, that's our goal.
And for whoever pre-order now we provide this free lifetime subscription offer meaning we will never charge pre-orders any subscription fee, whether we will charge and how much we'll charge subscription fee after, we haven't made our decision. It's something we still need to battle against different costs and different functionalities we provide, but we provide this guarantee that for pre-order supporters that we don't charge anything.
Jonathan Mosen: But it is conceivable that somebody may buy the glasses and that this is after your free offer expires and that in addition to paying for the hardware, they may need to pay say an annual subscription for service as well.
Huasong Cao: Yes, there's that possibility. I don't have the answer regarding how much and when we'll start doing that. The goal is that we as Agiga and EchoVision bring some real value to people and people find it valuable and worthwhile to pay that subscription. That's hopefully a win-win situation. Otherwise, I don't think it's sustainable for either one.
Jonathan Mosen: Do you think you'll have these ready to go on the market by the national convention in July? Is that what you're aiming for?
Huasong Cao: We're definitely coming to NFB conference in New Orleans. Whether we'll be ready, we probably will have some version close to the end product that we'll be shipping. The reason that I haven't committed to the final, final one to be available during NFB conference is because of some of the feedbacks that we obtained through CSUN time regarding the sizing and industrial design.
So, we are having a heavy debate regarding whether we tune some of the industrial design based on the feedback we got from CSUN. If we do, that might push the end shipping date a little bit more into the later months. We are going to have some answers soon and we're going to update our pre-orders soon. With that I can come back and answer whether we have the final shipping ready to go to NFB conference.
Jonathan Mosen: I presume you've done your numbers because one of the things that I've heard fed back, and this came up in our panel discussion a couple of weeks ago when we talked about CSUN, is people like the product, they think it's really exciting, it's got some innovative features, but they get concerned that with a company like Ray-Ban Meta they are spreading the cost of production over millions of units and the blind community is a much smaller market and you're coming in at a pretty aggressive price point. Are you confident that you've got a sustainable business model for this?
Huasong Cao: I actually listened to that panel discussion. I think Mike May commented on this. He actually was with us during the demo when Steve Wonder visited us and he came back and asked the similar if not the same question, whether we can sustain. We actually heard similar comment from more than a few supporters in our group, people pre-ordering and I want us to basically be alive when they get the device after a year or two. We definitely want it to be some business that can survive by itself.
We also have this goal and vision that we want everyone blind and low vision in this community can actually have one of our devices. We are exploring different ways and one first approach, I guess, right now is to make it affordable so that we can be one step closer to that final goal. This also comes back to the question you raised, whether we and when we possibly charge subscription that might be additional revenue that we explore when it really makes sense.
Another ways could be some other insurance coverage possible collaboration with not-for-profit organizations. So, our goal is to make sure we can reach some volume so that our manufacturer costs can go down. In the meantime, if we can bring enough value to the people that we really want to serve and more people really buy the device, use the device, provide value back, then I think it's at the end of day a win-win situation.
At the moment, we don't have concern of getting the devices manufactured for the first batch, but to make it sustainable business going long-term, I think the volume that you mentioned there really matters to us. So, thanks for bringing us on board. I think this is a really good way for us to spread the word to more audience and hopefully there's this effect that we can basically bring good device and people help us spread the word.
Jonathan Mosen: Are you giving any thoughts at this stage to the onboarding and the training that might be provided to help new users get started with the product?
Huasong Cao: We hope it's a self-serve for the most part, for onboarding and training. A few avenues that we are exploring, example video tutorials, some type of online, text-based materials and people asked also about accessible materials like Braille-based text manuals. We're exploring all those. We're also trying to see if we can collaborate with blind centers across everywhere and also low vision practitioners who can help us be the frontier to answer questions that they already serve.
It's possible that we can also build ambassadors from our pre-order user base who can help us, basically, help their friends. We don't have a final answer, but we're exploring all those channels, but at least what we can provide is the typical support through YouTube videos, through text-based materials.
Jonathan Mosen: We've talked a little bit about Ray-Ban Meta and that's a consumable item. It's very popular. You also have another well-established competitor specifically in the blindness space and that would be Envision. What do you see distinguishes you from the Envision Smart glasses?
Huasong Cao: It's a difficult question to answer because I don't know if I want to really compare each of the detail and say we are the best, but I think some of the approaches that we are doing might set us, at least, for success if we are not really winning every battle or competition here. One area is that we kind of start fresh, which means that we can try to look at what's the latest hardware, latest AI functionalities and latest integration of wearable technologies. So, we're a little bit lucky that we started late, which means that we can actually take the latest things and make use of that. That might be one area we set us apart.
Another area is we try to build the features and functionalities mostly based on what we hear from our audience. The hope is that this is the strong demand that's already there and when we deliver what people ask, there will be good welcoming part of what we deliver. That's, again, hopefully a win-win situation for our product and people that we serve. I don't think we have to basically be the only one in the market. Many people like us also because they have a choice. So, I think at the end of the day we provide another choice that may do very well in certain areas that people need and there are other choices that people may still need for other scenarios.
Jonathan Mosen: If people want to know about the EchoVision smart glasses and the Agiga, where can people go to potentially make a pre-order and just get news about how the glasses develop?
Huasong Cao: Please come to our website, which is echovision.agiga.ai, A-G-I-G-A.ai and we have newsletter that you can subscribe on the website. Thanks for bringing us on board to Access On. We are happy to come back and provide update here too.
Jonathan Mosen: We look forward to that. Where did the name come from, by the way? It's a palindrome. That's pretty cool.
Huasong Cao: Yeah, we basically want to bring AI technology to everyone. That's what the last piece of this name actually represents. So, AGI is the universal AI that we want to bring to everyone.
Jonathan Mosen: Huasong Cao from Agiga joining us to talk about the EchoVision smart glasses. Certainly, a product to watch and we are hoping to get an evaluation pair here at the International Braille and Technology Center. We'll put it through its paces if that's the case. We've got a bit of time for some listener feedback. We'll get to that after the break.
Speaker 1: Registration is now open for our 2025 National Federation of the Blind National Convention from July 8 to July 13 at New Orleans, Louisiana. Our national convention is the largest gathering of blind people in the world. It is the premier event for training, support and information for the blind community. The national convention offers hundreds of sessions, dynamic presentations and countless opportunities as well as a chance for members to participate in decision-making in organizational policy. Register today at nfb.org/convention.
Jonathan Mosen: This email from Desiree Simeon says, "Hello, Jonathan. I listened to your February 19 episode of Access On. I was interested in all of the content, but particularly your email from Amos Miller from Glidance. As you read his observations of the lack of accessibility of employee tools, I finally felt validated. I have been working at Pearson for the past eight years, and from day one have had trouble accessing employee resources, tools and apps used for task tracking and collaboration.
I work on an accessibility team and while they're very understanding, I feel like I complain a lot when I either cannot access the things they can on a daily basis or it takes me several hours to perform a task or complete a training when they can do it in 30 minutes. I also still cannot independently manage my benefits, my pay stubs, expense business travel, tracking goals and projects and the list goes on. The stress that comes from these daily struggles is real and has definitely taken a toll on me and my family.
These struggles in our community are real and all related. We have trouble finding work because potential employers are hesitant and don't think we can do the job. Then we get the job only to find the tools used and not accessible, but we keep going, trying to compensate for the inaccessibility so that we don't lose the jobs we fought so hard to obtain. We do this day to day on top of the normal struggles all people face in our society. And even when you have well-meaning employers, they don't have great options either. They are told to only use tools that are accessible but have nowhere to turn for help because there are so few tools out there that work. I started this email to you to see if I could get a copy of that email you read on the February 19 episode. I would like to share it with my team, but I feel very strongly about this issue and would love to figure out how we can unite together to make it better. Thanks for all you do. Best of luck to you in your journey with the NFB."
That's kind of you. Thank you, Desiree. It's an amazing journey and thank you for your heartfelt email. Let's get the easy part out of the way first. You can get a transcript of the email as part of the transcript of the episode and we transcribe every Access On episode. So, if you go to nfb.org and then you drill down to Publications, and then from there you will find Access On. When you go into each episode, you will find a full transcript of each one. So, go to the one where we read the Amos Miller email and you'll find the full transcript there.
At the National Federation of the Blind, we believe that the Websites and Software applications Accessibility Act would make a material difference to these sorts of issues. So, do get behind that legislation. You can find information about it in the march edition of The Braille Monitor, a full fact sheet with some calls to action there about how you can promote this legislation becoming law.
You can also hear more about it on this very podcast back in episode 12 when the National Federation of the Blind's president Mark Riccobono talked about what a significant difference this legislation would make if it were to become law in the United States. And of course, because so much of the software is developed here in the United States, it would make a difference worldwide. So, I encourage you to get behind that legislation, learn about the difference that it would make. Let's get it over the line and deal with some of these frustrations once and for all.
Now Karen McDonald is in touch and says, "Good afternoon, Jonathan." Good afternoon to you, Karen. She says, "I'm a member of the at-large chapter of the NFB of West Virginia." Almost heaven, West Virginia. "It is a pleasure," says Karen, "listening to your podcast every week and we are so glad that you are working with us to make life better for blind people everywhere." Thank you very much, Karen. It's really a privilege to work with so many amazing people on this journey. Karen says," I have an iPhone 16 Pro Max running the latest version of iOS. I use a Brailliant to control my phone via Bluetooth.
Pressing the space bar with dots two, three and four should move me to the status bar, but that doesn't happen and hasn't happened since I got my new iPhone. I have also tried this using a BrailleSense 6 paired with my phone with the same results. Is this a bug or is it just my phone? If anyone has any thoughts about this, I'd love to hear them. Thanks for reading."
Karen. Let's see what others have to say. What I can tell you is that I'm using a Mantis so that has a QWERTY keyboard rather than a Braille one. That may make a difference, but I'm not sure. And when I execute the status bar command, which in the case of a QWERTY keyboard is the voice-over key VO with M, it does work for me. I'm popped into the status bar. I would like to see what others come back with and when I get a second I'll pop into the IBTC. We've got so many Braille displays there and just see if I can pair a couple with my iPhone and see what happens. It looks to me though, at a cursory glance like it seems to be working, but let's see what happens.
Accesson@nfb.org if you would like to chime in on this one. Accesson, all joined together, @nfb.org. And if you like Karen are a member of the National Federation of the Blind, let us know what chapter and affiliate and we'd be delighted to give you a shout-out.
Let's go to Chicago for this next email. Yes, there are plenty of songs about Chicago, but I won't try and inflict anything on you. See? The Hills are Alive with The Sound of Music I tell you. And here's Terry Gorman and he says, "Jonathan, I would very much like it if you would do a show or a couple of them explaining how one might get accessible TV. I've joined the television viewing set of late and need help. Should I go Apple TV, Amazon's stick or cube, Samsung, Sony or something else? Oddly, I'm not interested in a large TV, but I'm sure others are. Reading a lot of blogs is tough. Please give it some thought."
Well, some of the things you mentioned there, Terry, are set top boxes that would plug into a TV. So, Apple doesn't make a full TV, but they do have a set top box called Apple TV. You would need to have a television to plug that into or at least a sound bar to plug it into with an HDMI port. Similarly, the Amazon Fire TV stick and cube are set top box type devices, but I believe that Amazon do sell a full television that talks, that has their VoiceView screen reader in it. And VoiceView is a pretty mature product now, it's well-developed. So, you may find that buying an Amazon TV and whatever size you can find with a screen will give you a lot of what you need.
Once you do that, the Amazon Fire TV software will be on that television and that does have applications like Netflix. Obviously, Amazon Prime is built right in there. YouTube and various other services that you might want to access. Plus, you'd be able to watch terrestrial television as well. Samsung have a wide range of TVs with voice guidance built into them. My one criticism of voice guidance on the Samsung TVs is that at least last I checked, you only have one text-to-speech engine and it can be pretty difficult to understand for some people.
That is a bit of a downside on the plus side, certainly in New Zealand when we had a Samsung TV there, I was able to navigate the program guide for the service that we use, choose to record shows. You can connect a hard drive to the TV and record things if you want and it's pretty easy to play them back. There are a lot of accessible features once you turn the voice guidance on, but the speech may put you off.
LG also makes some accessible televisions and I've heard good things about LG. I've not had any direct experience of them. So, if you are using an LG TV, let us know what you like or dislike about that. Some TVs, that run Android TV have TalkBack built in, so you do actually have quite a few accessible choices. One thing you could do is buy a relatively simple TV, one where you can clearly switch to an HDMI port or two and then buy an Apple TV, which has VoiceOver built in.
And if you're an iPhone user you'll be familiar with VoiceOver and maybe also a fire stick if you just want a bit of variety. And then you could use your basic TV to navigate around and you should be able to get enough familiarization to switch to those accessible set-top boxes. That's another strategy that some people choose to deploy.
If you're using an accessible TV, what are you using and what do you like about it and where do you think it could be improved? Let's find out what Access On listeners have to say. Accesson@nfb.org if he'd like to help out Terry with his search for his accessible TV.
We are off to Israel now for an email from Haya Simkin who says, "Hi, Jonathan and listeners, I recently began exploring a volunteer opportunity transcribing the life stories of elderly people, a common practice here." What a wonderful idea. Haya, I bet you get some real pearls of wisdom from being a part of a project like that. Well done. He says, "Typically, a volunteer visits an elderly person's home, records an interview about their life and transcribes it.
A different volunteer edits the transcription and yet another adds graphic design before the work is published. All of these steps are done on a voluntary basis. Because most interviews will be in Hebrew, I need transcription tools that are freely accessible and work with languages other than English. To experiment, I first recorded a thirty-minute phone call, my electric bill payment," it's nice to know electric bill payments are the same everywhere, "on my iPhone's notes app, the minimum length I expect for an interview, which usually lasts one to 1.5 hours. I attempted to use Notebook LM to transcribe it but encountered several difficulties.
Earlier, I tested a short recording where I recited a children's passage using the Voice Memos app, saved it to the files app and uploaded it to Google's notebook LM site with instructions to transcribe the Hebrew audio. Although that short, clear recording featuring only one voice and well-known text was transcribed perfectly, I found the text insufficient because it did not replicate the complexity of a full interview. When I returned to the electric bill recording, I faced issues saving it from notes to files. The phone repeatedly shows a preparing message for several minutes, then refreshes to display the recording in notes with no sign of it in files."
I'm just going to stop there and say this is a bug we know about. I've been helping Gary Wonder who's doing some work at the moment trying to fix this very problem with a large call recording that he made in the Notes app and we can't get it to save to files either. We've contacted Apple about this and have reported the issue, so we could only hope that it gets fixed soon. Haya continues, "I also tried emailing the recording to myself after a similar preparing message and refresh. The email arrived without the attachment. I'm not sure if I'm saving or attaching the file incorrectly. I also read that ChatGPT using OpenAI's Whisper API might handle this, but I haven't managed to share a recording with ChatGPT. I'm not familiar with APIs or coding, so I'm hesitant to pursue that route.
I did find an iPhone app that claims to work with Whisper, but when I used a short recording with auto-detect language, it produced many errors. Manually Selecting the correct language only resulted in a stalled transcription process. Do you have any experience with Whisper API or know of any free accessible transcription tools that work with languages other than English? I'm not sure when the organization might assign me a project, but if I can't secure reliable transcription tools, I may have to abandon this plan given the workload involved. Thank you in advance."
Well, to answer this question I thought we'd just get ChatGPT on the show. So, welcome, GP. Welcome back to Access On from the National Federation of the Blind.
Gp: Thanks for having me. It's great to be here. What's on the agenda today?
Jonathan Mosen: We have got a listener who was inquiring about transcribing audio, so if I upload a file to you, can you transcribe audio?
Gp: Absolutely. I can transcribe audio files for you. Just upload the file and I'll get started.
Jonathan Mosen: Is there a length limit to how long the file can be?
Gp: There's usually a limit, but it depends on the specific platform or tool you're using. Typically, for a single upload it can range from 30 minutes to a couple of hours. If it's longer, we might need to split it up.
Jonathan Mosen: Okay, and what languages do you support?
Gp: I can handle a bunch of languages, English, Spanish, French, German, Mandarin, you name it. If you've got a specific language in mind, just let me know.
Jonathan Mosen: Hebrew, mate. Do you do Hebrew?
Gp: Yes, I can transcribe Hebrew as well.
Jonathan Mosen: Could you talk me through how the process works? We're talking to a blind audience here. How would they upload the audio? Let's take the website first, if you would?
Gp: Got it. On the ChatGPT website, you'd navigate to the interface where you usually type your messages. There's an option to attach files or upload audio directly. You'd use your screen reader to find the attach or upload button, usually near the text input area. Once you select it, a file browser will open. You pick your audio file, hit open and then send it. I'll take care of the transcription from there.
Jonathan Mosen: And can you do the same thing with the iOS and Android app?
Gp: Yes, definitely. On iOS and Android, you'd open the ChatGPT app and there's a similar attach or upload button near where you type. Tap that button, choose your audio file from your phone storage and then send it in. It's pretty straightforward.
Jonathan Mosen: Genius. Thank you very much for your help. I appreciate that.
Gp: You're welcome. Glad to help. If there's anything else you need, just let me know.
Jonathan Mosen: Got anything to say to the members of the National Federation of the Blind?
Gp: Absolutely. Live the life you want and let's make it a great one.
Jonathan Mosen: And if you've had any experience with transcribing audio using ChatGPT, let us know how well it works for you. Is it viable? Accesson@nfb.org if you would like to get in touch. And since that was such a handy hint about transcribing audio using ChatGPT, I put that at the end there to serve as this week's tech tip. We are always looking for new tech tips by the way, so if you have one, you can get in touch, accesson@nfb.org and put tech tip in the subject, and we'll put you a little tech tip in the segment here at the end of each episode.
That concludes this episode of Access On, the technology podcast of the National Federation of the Blind. To send in a contribution for a future episode, email us. Attach an audio clip or just write it down and send it to accesson@nfb.org. That's accesson@nfb.org. To keep up to date with Access On, follow us on Mastodon, accesson@nfb.social, that's accesson@nfb.social on Mastodon. Subscribe to an announcement-only email list about upcoming episodes. Send a blank message to accesson-announce-subscribe@nfbnet.org. That's accesson-announce-subscribe@nfbnet.org. To learn more about the National Federation of the Blind, visit our website, nfb.org or phone us 410-659-9314. That's 410-659-9314. And be sure to check out the Nation's Blind Podcast right from where you heard this podcast.