Golden Speak (http://www.goldenspeak.com) helps you speak better by analyzing your voice for pitch, rhythm, vocab, fillers, and clarity. We'd like to talk about launching minimum features quickly vs. high user expectations that people have from apps like facebook, etc. Does it make sense to focus more on polish instead of launching within an almost embarrassing feature set?

I believe to be one of your target demographics (international students, thick accent) and I would definitely pay for a service like yours.

Focus on delivering a polished products that can deliver, I will try your product only once and if your product's quality convince me that there is an opportuntity to improve my speaking skills I will literally throw money at you.

Could Golden Speak be used to correct non native speakers with their accents? I have a (very) strong French accent when speaking English and I could use a tool like that.

My View: The smallest amount of features that will add value.

Oh, just saw that the post title converted the ampersand between Hackers-Painters to the word AND. This was supposed to be a subtle reference to Pg's book, which uses the ampersand. It is a great read if anyone is interested in either subject.

The essay uses "and" (http://www.paulgraham.com/hp.html), but the book uses the ampersand. Both seem correct.

We built Golden Draw to hack the drawing / painting process. With GD, anyone can draw or paint any picture - even if you have no artistic ability. The app works by breaking a picture into tiny boxes, where you draw each box one by one. It's sort of like a computer monitor breaking the screen into pixels, rendering each pixel one at a time.

We built this using the latest speech recognition technology. The major use case is for people who want to talk better. For example, if you use ums, ers, or ahs in your speech. Or perhaps you talk in a monotone. The app can be installed on ios or android.

Interesting. Why does it only analyze the first 60 seconds of speech? Is that a technical limitation or a theoretical one?

I'm guessing this isn't a technical limitation. It's probably related to getting the results back in a reasonable amount of time. When I did a speakoff on android, 60 seconds of speech took about 2 minutes to give me results. So I'm guessing the larger the payload, the greater the wait-time. But I may be wrong.

No technical limitation. It is that most people find it difficult to speak on a random topic for a minute.