Walking, Talking, Writing: A Blog Post Experiment

So, I’m going to attempt to write a blog post, actually by dictating it using IdeaShell. It’s my dream to write a whole blog post using voice. It’s not done using any Text-generating LLMs, but this is done using Speech to Text (STT). Just to try it out. I’m doing this while I’m taking a stroll at Madras, an evening stroll. It’s a good way to test it because I just wanted to test it in different environments. How does it perform?

I’ve made voice notes to myself before, like this, where I walk and talk to myself. This is a little bit more than that. Here, I am not recording for sound. I’m not interested in a voice recording, but I’m more interested in the text part of it. So, I will edit this later once it gets converted into text. Most probably, it’s going to be because there’s lots of nonsense when I’m talking. This is similar to how I edit audio recordings before sending them to someone or making them into a podcast. I will also likely need to edit the text part of this to correct the conversion errors. But I’m guessing 80% of the text would remain there. If this works, I think it will be a great way for me to write things. I think it’s like writing the first draft, and then I can go back and edit it later when I’m free before I publish.

I’m using an app called Idea Stories. I think it’s called IdeaShell, Voice Notes kind of a thing. However, it has several interesting features, including auto-tagging, good speech-to-text (STT), and summarization—modern features for a notebook. I’m not using it as a knowledge management system. I’m using it just to input content easily.

Why didn’t I use any standard text-to-speech conversion? I wanted it to do some corrections after it converts into STT. So, I didn’t want a simple STT. I wanted something smarter that could correct grammar, spelling, and other issues after it performs speech-to-text (STT). IdeaShell is a little more intelligent than your regular text-to-speech software, and it also performs auto-tagging and other functions, making it easy to find everything.

Essentially, I’ll be part of a pipeline that directs data from here to my blog or any other destination, such as the knowledge system (ZimWiki), where I’m using it as input.

That way, it’s much better to input knowledge or notes than actually typing it out because the focus is on reducing typing. Additionally, it reduces the need to be in front of the computer just for typing. It’s somewhat annoying that you want to write a blog post and have all the content in your head, but you still have to sit in front of a computer to type it. This, I think, is a better option.

I know people walking and typing have done that. I mean, there is much more. I know someone who does most of their programming, also walking and typing with a portable keyboard. This is in the same vein but not the same. I don’t want to type; I want to speak while I’m walking around taking a stroll and then get back to it later, finish it, and publish it. That way, my time in front of the screen is reduced by at least 10% of what it used to be before if I have to produce the same amount of writing.

Just to remind you, this is not generated content. The content is still by me. It’s probably like an assisted writing tool that converts speech to text, so I don’t have to sit and type physically. Let’s see how this experiment goes. You will read it online if it’s decent, at the very least. Yeah. Bye.

Notes:

1. This post was dictated to the IdeaShell App, and then the text was copyedited, linked, and formatted on a computer before being posted.
2. This post got edited, and many things got removed, but the process was much faster and easier than I expected. About 80% of the words remained. So, it’s not a bad conversion rate.
3. “This” in this post mostly means this post or this process of posting.
4. I may or may not use the IdeaShell App in the future. However, until now, that’s the best app I’ve found on mobile for this purpose. On my desktop, I use SpeechNote.
5. I speak a lot, I ramble a lot, I frequently repeat myself, and I switch contexts. That works as a podcast or in audio format, but when it gets converted into a text post. It feels like I am all over the place. It could also be because I am thinking as I speak. Something for me to keep in mind
6. I think it works well for first-person stories or blogs. It has a voice. It might work well for some kinds of posts than others. So, I will use this tool.

---