Some time ago I did a "Show HN" of the same project mentioned in the story. It turned into an interesting and controversial discussion pretty quickly. I reference the HN thread in the article as well, if you are curious.
Hope this article is helpful for your entrepreneurial journey!
And, of course, I'm happy to discuss this project further.
The speed of production was definitely a consideration. The idea was to start with communities that read same/similar content. So that a lot of people would be requesting same articles for narration. Certainly, when the article is requested for the very first time — it could take up to 24 hours. At scale, even with human narrators, the gap could be minimized to an hour or less.
When an already recorded article is being requested — it's instant obviously.
Also, considering use-case, most users should be fine with audio not being available right away. Imagine: you request an article to get narrated, and by the time you hit the gym, start driving home/to work — the audio will be ready!
True. When I first read the article, the first use case that popped into my head was "I need to go to work, I'll request the audio version of this article and listen during my commute". While I was totally wrong about your intended use case, so will some other people. I'd suggest looking into machine-learning-based TTS, it might give you a faster turnaround with similar quality of sound.
6 comments
[ 3.4 ms ] story [ 14.6 ms ] threadSome time ago I did a "Show HN" of the same project mentioned in the story. It turned into an interesting and controversial discussion pretty quickly. I reference the HN thread in the article as well, if you are curious.
Hope this article is helpful for your entrepreneurial journey!
And, of course, I'm happy to discuss this project further.
The speed of production was definitely a consideration. The idea was to start with communities that read same/similar content. So that a lot of people would be requesting same articles for narration. Certainly, when the article is requested for the very first time — it could take up to 24 hours. At scale, even with human narrators, the gap could be minimized to an hour or less.
When an already recorded article is being requested — it's instant obviously.
Also, considering use-case, most users should be fine with audio not being available right away. Imagine: you request an article to get narrated, and by the time you hit the gym, start driving home/to work — the audio will be ready!
I'd assume there might be a gap between the time you request and actually hit the road.