Typeless Killer - VoiceInk!!
Directing AI to write code often hits a bottleneck in typing speed. The commonly used software is Typeless, but I find it too expensive, so I created an alternative version with comparable performance that is also easier to use.
Typeless Killer - VoiceInk!!
According to Typeless's official website, the average person types 45 words in a minute; however, by speaking, one can produce around 220 words within the same timeframe. Using voice to generate text can be up to four times faster than traditional typing. I personally tested it and found it indeed useful. After reaching the free quota, I discovered that even with a Christmas discount of $20, the annual fee is still $124, which is approximately NT$4000. I thought this was just too expensive.
So, why not create one myself? With this thought in mind, I began developing a substitute version of Typeless. The function I implemented includes word recognition, with matching speed and accuracy. My unique feature is statistical tracking of the number of words converted from voice and the duration of the audio uploaded.
<LinkPreview url="https://github.com/tienenwu/VoiceInk" />Above is VoiceInk, the software I created as an alternative to Typeless, and I've open-sourced it. In fact, developing this functionality isn’t difficult; the core ideas are twofold:
- Capture audio and convert it into text.
- Hand the converted text over to AI for refinement.
<span className="text-[1.35rem] font-semibold leading-[1.65]">My unique approach lies in how to refine the text converted from speech, which you can customize.</span>
It can automatically recognize and convert speech into text in over 100 languages.
The only downside is that my personal computer is macOS, so I’ve only developed a version for that platform.
However, the fundamental core concepts on how to achieve this are in the code. If others want to expand it to multiple platforms, they can refer to it.
Of course, this usage method may be somewhat challenging for those unfamiliar with coding. First, you need to register for an OpenAI API Key and set a credit card for quota registration; otherwise, it will not work.
I must say, not having to type on a keyboard is a fantastic feeling, because by using voice, I can quickly generate what I want to say to AI by simply speaking, which is incredibly fast. If you're interested, give it a try. If you find it useful, please remember to buy me a coffee via the BuyMeACoffee link.
Lastly, I should mention that it took me about two hours to implement the core functionality from scratch, including debugging.