July 2022 Month In Review

 

Automated Transcription

OK hello hello it is Tyler Bryden here. I hope everything’s going well. This is July 22 month in review. In the past I’ve broken this into many things specific specifically speak AI software company that have run doing natural language processing, transcription. All this wonderful stuff continues to grow blah blah personal life updates, YouTube channel. All those wonderful fun things and then even maybe some professional growth or things that I’m learning today. I’m going to try to combine that all into one video timing. Myself with 15 minutes, 15 minutes Max. That’s the, that’s the the the the race against the clock that I’m doing here. But first I’m going to start with PKI which is very important part of my life and I’m so grateful for people who have followed us along all the supporters, whether that’s from afar friends, whether that’s people who are interested in investing in us, that’s people who are just most, I think, most importantly customers and users of the speak application. We’re so grateful for you, and we’ve learned so much working with you.

Along the way, and some of the progress specifically in this month, has really been spurred by requests by the customers. The partners that we’re working with, and so we’re very excited to detail that, and I think, just generally one of the trends we’ve seen working with. I would say research firms, market research firms, or even companies trying to find competitive advantages or insights in the large sort of exponential exponentially increasing data. Set of language information that’s online. They’re looking for different tools to do this, and one of some you know. Some of the companies to us, but one in particular said, hey, we’re looking to, you know, identify all sort of mentions of this specific term and then products associated with this term. We want to analyze it, and as you know suggested, hey, why don’t you work with, you know, a social listening tool and they said, because we’ve tailored the insights and the analysis. So in so much detail within the speaker application, we can’t get that insight anywhere else. And so we want to use.

Hate to do this and so you know can you help us build some of these scraping capabilities and functionalities to unlock insights from the web and then tailor them specifically to the insights that are in the speaker account? So specifically what that looked like in July was doing pulling, basically building a full scale Twitter scraping engine and a full scale Amazon scraping engine with the goal to automatically be able to identify. On Amazon, basically product and then on Twitter be able to identify by search you know words or hashtags, but then also be able to even go hone in more specifically and say hey I want to look at this country. I want to look at this region or these cities. Are this languages and so that was some of the work that we have done this month. I wanted to talk a little bit about this and share a couple insights but just generally Twitter. Although maybe there’s some you know, disagreements about how many people are real on Twitter versus bots. It’s still a huge conversation hub.

Rich language data insights are possible here and what we wanted to do was be able to go all the way back to 2006, scrape tweets and then automatically be able to analyze those we needed to build multi language into that and then wanted that to connect very simply into the speak platform where the categories the custom insights that you would build into the system were then unlocked. And so if you are interested in this, if you’re interested in like trends, products, events or just people, the way that people are are are are, you know having conversations online specifically on Twitter. Related to your. Or whatever product you’re in or whatever competition you’re in, or just whatever your marketing, marketing, or researching feel courage to send us a message. We love this stuff and we’re actually building out more scraping tools to be able to identify these different levels. Pieces on the web and then bring them into speak for analysis. But I wanted to hover on Twitter for a second just because I think it’s really interesting what is actually possible. And then it also shows you a little bit of what you know what, what, what we’re getting, and then how we’re working through it so.

And I’ll pull up a screen share what I’ve done. I shared a video on COG Video which is like text to video generation. Super interested, something I’m interested in and then you can see that we can query this Twitter engine and we start to then organize this data. This data is very extensive that we can get and you know goes all the way from like one with treat was tweet was created but then when the user was created and then the actual tweet and then the translated version of that if it’s coming in different languages and we automatically identify the country, the language and then make that translation the screen name. How many people are following these users?

As the center original tweet, and is there any hashtags associated with this in a URL associated with this as well as actual geolocation and details on these on these on these tweets and you can even specify and say I only want ones where geolocation is enabled, so that you know that you’re looking at a specific place and then what’s happening here is we’re then combining that into. Basically what is a speak import template and I can pull this up. We’re sort of aggregating some of the columns and rows together to make. Incense and to speak and then those can be imported so that the tags are actually there. So you now know that the query that you looked for was cog video. You have to tweet and then if there’s a location assigned to it then there is a location. It actually honors the creation date and then you’ve got the raw text there as well too. So fantastic way to sort of push all this information into speak in a way that is compliant with the system and we’re actually looking to extend the system beyond what it is so that you’re not necessarily stuck to just sort of the tag and column and row.

Structure that we have. I guess the row structure will be, but the column structure we’re looking to advance so that you can bring in things like location and age and gender and some sort of standardized valuable inputs that we’re seeing consistent across datasets. So that is something we’re really excited about, and then we’re able to import that into speak, and you’re able to do analysis very quickly. And I’ll show this out a little bit more on Amazon, which is Amazon is obviously an incredible place where there are a ton of reviews that are happening. In my case, I have been.

Looking at this damn this damn outdoor lighting string and there’s like challenging reviews on here. You know what good, bad? And then I’m looking at different locations and even times of like when you know when are these? Are these new positive reviews new or are they old and have things change etcetera etcetera? So obviously this is me individually looking at this but if I’m looking at you know if I’m a company building products in a specific market and I’m trying to understand you know how do I differentiate across them and how can I compete or or build better products? This all the sudden.

Things really insightful and so now all I have to do is get some URLs. Basically pass them to us and then we will scrape all the reviews and if I pull it up you’ll be able to see the reviews one second here, which is beautiful right here. This was a lot of preparation to get this ready to apologize and now we’ve got the description. We’ve got the product name there. We’ve got the username. Who did it? The actual star rating and then the location. And then we’ve got the text. We’ve got to create that date. And then we’ve got some tags to help you organize it.

And we’re working through it. It’s annoying in Excel we found out that there’s an issue with sort of like handling emojis and stuff that seems more prevalent on Twitter, but a couple of things that we’re working around already and making sure that this is useful, and I think that process is gonna continue to get better and then what’s nice is that hops in to speak into the platform here, so you’ve got the umbright. You see, a person wasn’t very happy, got a couple insights that are coming out of it, and then you’ve got some sentiment analysis as well too, so you can see you know. Obviously, overall sentiment, which was about 5050 and then and then you can actually filter by, you know, filter by sentiment to find the most positive and negative moments. If I want I can also go into, you know, and there was only ten reviews in this specific one that we scraped, but I can go into one of these items and see if there’s anything more worthwhile. So for a couple of times.

Classic was mentioned several times in the. In the in different tweets and this one here. This it was actually one person in general comparing vintage to classic and you can see and extract that very you know very quickly. So from that Explorer Insights you can start to filter. We can look at Amazon reviews if you start to get more you can start to see you know what’s emerging most and if you have specific you know categories of insights that you’re looking at. This becomes very valuable for you to move forward with and so we’re you know, excited to make this even better. We’re going to add a lot more capabilities with sort of columns and being able to analyze.

Everything and look forward to hearing from you about how we can make this even more valuable for you. So that’s one piece when something that popped up that was really fascinating. I’m not going to show this in in terms on the screen, but we also had some requests. Once the data was generated that they wanted to enrich this data with gender and age. So we found a ability to basically look at the the names of people who are leaving reviews or tweeting, and then build predictions of their age and gender, and so you can say, hey.

You know with we’ve looked at a data set of 1000 names we have this understanding they’ve been labeled and we have a 85% confidence that Alice is a female, and so if you’re starting to look at Amazon reviews or tweets and you’re trying to group how maybe different demographic types are interacting with your product or service or whatever it is. This system allows us to enrich your data very quickly, automatically, and then with even giving you an insight into what confidence level it is. Being able to start to group those and get even richer insights from your language datasets that were helping you extract. So we’re really happy about that and look forward to hearing you know how in other ways can we enrich your information. So that’s something that I’m I’m personally very excited about and I find very interesting, although maybe challenging in the environment where you know gender and how we understand.

Cells and label ourselves as different, so also working through sort of ethics and things like that to figure out the best way to to to honor that. And then I move on to what I would say is automatic audio stream clipping analysis and this is really interesting. We’ve been hooking into audio streams that are running and those are running all the times and sometimes they’re silent. Sometimes they’ve got a buzz to them and only sometimes are people actually speak and there’s music playing. Only sometimes are speak people speaking and there’s actual words to transcribe, and so we’ve built this little mechanism to hook into an audio stream you can hook into that.

And then you can automatically identify when words are being spoken, turn that into a clip, and then that clip can be brought in to speak for transcription analysis and the goal here is that you are not wasting any processing time or expending extra money analyzing audio that is not, you know, not filled with language information and and just creating this noisy data set where you have to filter through and look for the really meaningful moments. We’re helping you identify it with this automatic. Listening to the screen clipping mechanism, finding when someone speaking and then producing it as a file that then can be transcribed and analyzed directly and speak so really quite excited about that. The goal is to get you the language rich data that you’re looking for simply without the cost of unnecessary processing of this information and. Those are some of the big pieces like that’s a lot right there that we worked on in July and excited to bring those together. Make them even more accessible, even more democratized within the application. And then we’re looking at a couple of things that are next and those things that are next are relationship graphs. And really, the goal is now we’ve dumped all this data into the system. How? How are these terms connected? How are these datasets connected? How are Amazon reviews connected versus the Twitter tweets? If they’re all looking at the same subject, or if you’re doing a bunch of interviews and focus groups, what are the connection there? So this graph that we’re working on looks to uncover that information.

In a really visual, interactive way, so that you can find patterns, find connections, pattern themes in a way that you haven’t been able to do before. The other part that I think is really important, and something that we’ve been working on for quite a while, is a CSV importer system, and so the way that I showed you that CSV we sort of had to custom build that to bring it together into the template of the import, which you know basically allows you to go right in to speak and then upload it. But what we want to be able to do is in a very intuitive. May be able to allow you to match the columns that you want, so if it was an Amazon, I’ll continue with Amazon, but it’s a five star review. You don’t want that as A tag. You want that in the actual description, then you are automatically able to port and route that to where you want. And So what? The goal here is that this opens up a huge level of imports and to speak with language data that allows you to analyze it and allows us to not even necessarily have to do these crazy native integration.

You can just export CSV from the data platform that you’re looking for, bring them to speak automatically, get the sentiment automatically, get the name, get recognition the data format that we do and all the other beautiful insights on the Explorer page and everything else that our system gives you. So I think that’s a big exciting piece. The other thing I had mentioned a bit was this new sort of standardized data input. So we want to enrich organization. We want to enrich analysis, and that could be things like age that could be things like gender, location, and we can go deeper into that city, state, province.

Etcetera and these are very common data inputs that are coming out of systems that add another layer of information to the language data that is often being collected and speak. And so as we build out those columns. Lose that inputs, you’ll be able to route your imports into that, and then we’ll be doing auto analysis of that. Seeing again with. With this relationship graph with better analysis, what is being unlocked here that might not have been seen before, so very excited about that overall and then something we were working on for a long time that might be coming quick. Might be coming not as quick. We’re we’re, you know, you know always doing your best is this idea of you know word by word timestamping. So this allows you to have better analysis, better navigation, better editing. If you’re editing a transcript, you don’t have to jump to right now. We do sentence by sentence, timestamping. You don’t have to jump to the very start of the sentence. You can go to that.

Backward we know exactly where an entity is, not the rough location of it in a transcript, when it’s just by sentence, and so there’s a lot of unlocking of power that comes from word by word, time, transcription, even something as simple as formatting captions and so very excited to bring that all together. And we’ll keep you updated as we go, and so speed continues to chug along. The application continues to grow. We have more customers than ever. We have more users than ever. We’re really excited about that. We’ve had to adjust to the climate.

Or lean where you know we’re doing everything that we can to make sure that we’ve got a great sustainable business here that allows us to continue to innovate and work with customers and grow. And we continue to see what is working and what is not working and then evolve and iterate on that. And then also, you know, listen to market signals all this other stuff to make sure that we’re adapting and doing the best we can. So we appreciate everyone who supported us in that journey. We’re going to continue on that and look forward to helping, hopefully having you know a lot of innovation.

Continue to come and some more opportunities to work with you who have been such big support of our supporters of us. So that is, I think, a big piece on speak. This is sort of a a way to reflect on that and pass the 15 minutes. I realize I’m not uploading this to LinkedIn so I don’t need to cap it at 15 minutes. I’ve been racing through this whole time and I could have been a little bit more calm, but that’s OK. I’m going to touch into sort of a couple of just personal notes here, which is one of them. Very excited to say this was a huge sort of catalyst for me in my life.

In 2019, and then there was the pandemic came and so Burning Man did not happen. I am returning in 2022 in at the end of August, and you know, I guess this is a, you know, that’s all from our team is also coming. So considering it a company retreat and so we will be a little bit offline from the end of August to in the start of September. But I think that’s going to be an amazing moment to see, you know, to process to have fun and then also just get inspired by the incredible. Innovation that the art, the talent, the you know, the the life that is created. The experience that is created at Burning Man. And so I’m completely grateful for this. I’m so excited. I have a picture that I’m looking at right now that you can’t see of some of my good friends from Burning Man in 2019. Can’t wait to go back to the desert. And if you’re going to send me a message, I’d love to hear from you. Would love to connect out there. I know it’s gonna be a little bit of chaos but it will also be a lot of fun and a great way to learn. And then lastly, I think the personal note that I am in touch, I mean.

Many, many things, but something I’ve been working on is sort of in the side here on the side, but every day I’ve been. Trying to post, you know, videos as much as I possibly can, and in the start, you know, really, that barely does anything. You barely get any views. You very get anything and I’m grateful to see that some of the stuff that I’ve done is now paying off a little bit and starting to see some growth. Almost 5000 views of the videos in July, almost 200 hours and then 40 plus subscribers, which is you know, at 80, basically doubled subscribers in a month. I had someone on LinkedIn after I posted this today, say, hey, that’s 80 subscribers is not very good. Let’s get you thousands of followers.

I’m building this organically. I want a real audience here and I’m just very excited about the growth with the goal to hit. 1000 subscribers, 4000 hours. 4000 hours of views. I think if I I need to do that to basically be eligible for basically eligible for monetization and I’m testing out a couple of different monetization things here. Right now you can start to see a little bit of a, you know, a spike in everything there and you can start to see that you know pretty consistently publishing video, so this has been a lot of fun. I’ve learned a lot people have been commenting on my on my videos I’ve published and actually informing me about things.

Which then has sent me down more rabbit holes, so I’m incredibly grateful for that. It’s it’s like a feedback loop of lessons and insights, and one that I’m, you know. It’s definitely a lot of work, but one I’m thankful to be a part of, and then I guess. Just lastly, August August, it’s Leo season, first of all, and it is also my birthday coming up and I’m turning 30 years old, and so there’s definitely a period of reflection that’s happening. I’m, you know, trying to put together some sort of lessons.

My my life lives have made it to 30 which is still young but not as young as it used to be. And you know, thinking a lot about you know what am I truly optimizing for in these monthly updates? I’ve put these sort of traction metrics and that’s been a big part of optimization and focus for for myself and my team. But, you know, as the environment changes, the world’s changed, I’m starting to be more critical about that and also seeing you know, improvements in my own personal life make me want to spend more time in my personal life. And, you know, I’m really.

Trying to understand this balance of what’s the best way to build a company? What’s the best way to honor your purpose? What’s the best way to be healthy and happy? All these things are sort of emerging and I think it comes as a macro theme. As you you know, hit these big checkpoints. I know this came from me at 25, I 30 and then I’m excited to have Burning Man here to spend some time reflecting on that disconnected and also connected. I think that’s a really good place to to have lessons and grow and understand more.

You know? With that being said, like. I’m going to continue to share these insights. I’m going to create videos going to continue to grow speak. Continue to evolve both professionally and personally, and I’m just honored that there are people who still care about this. You know it might not be a lot, but it means a lot that there are even some and I’m thankful to everyone who sends me messages or emails or comments, or anyone who’s watched these videos or likes and comments. Subscribe all that good stuff. It means so much to me and I hope to see you maybe a little bit later because it’s going to be quite the journey home from Burning Man. But sometime in September. And I’ll do a little review on August and keep this trend going.

As as as we continue to go, and you know, in the meantime, feel encouraged to send me a message anytime. Always love hearing from you. So this has been Tyler Braden. I did a little review on on July talking about PKI so Twitter, Twitter analysis, scraping and analysis, Amazon scraping analysis, Live stream, live stream, sort of transcription with capturing of clips. We’ve got automatic data enrichment with age and gender. What’s next with different CSV imports? Word by word timestamping. So much good stuff.

That it’s on its way and it always takes longer than you think. But it’s always exciting when some of the the stuff that comes together and I’m excited to share more demos, tutorials, all this good stuff. So thank you very much for tuning in tuning in. This has been Tyler Bryden from speak I and I will see you. Well see you soon, but in terms of this exact context, maybe I’ll see you next month. Have a great rest of the day. Bye bye.