Also, the fact that 44,100 can be factored as 2^2 * 3^2 * 5^2 * 7^2 makes it very efficient to do Fast Fourrier Transforms on moving windows. Which we do a lot in Speech Recognition and DSP in general.
2,210 karma · joined September 5, 2013
I respect what you've done so far, but I can't help wondering how investors can base their analysis on an experiment restricted to Palo Alto, Mountain View and San Jose. From what I see, the bulk of orders (in $) are from fellow startups ordering lunch and dinners. Where else is the world (except SF and maybe NYC) do so many companies, in such a small geography, order so much food from restaurants?
I'm not saying it's not a good business. I'm wondering if there is really room for growth (a growth big enough to justify the huge series A valuation) outside of very specific geographies.
(second degree comment)
How about:
(comp foo second list)Seems somewhat exaggerated to me. There's definitely a shift from the valley to the city (and Palantir eating up all available space in PA sure catalyzes this process), but I wouldn't say PA/MV are "no more part of SV".
That being said, I want to moderate my comment. Until now, the situation is still under control, but it could become worse and then start to be a PR problem for KV. It's still time for them to do something.
I understand that junior employees can sometimes go publicly against the company they just left, but that's really not something you expect from somebody who is supposed to be a seasoned, professional exec. Good luck for the rest of your career, Rocky, you'll need a lot of that -- which CEO is the world is willing to hire you again?
Actually our typical latency is less than 0.5s if you stream audio to the API (instead of waiting until a silence, then sending a WAV file). Also, we are working on an embeddable client (you would still use Wit.AI online to train your model, but then they can run locally on your Raspberry Pi).
An easiest way would have been to use Wit.AI API [1](disclaimer: Wit team member here). You can create any command in minutes and use any device capable to stream audio to the API (including your iPhone) as client. I'd be interested to have your feedback if you have time to check it out!
[1] https://wit.ai
We also have a free community plan: if you agree to give back to the community by sharing your training data, then Wit.AI is free for you.
That being said, we are just getting started and we'll adjust our pricing after we get more developers' feedback.
Now to make it still cooler and Jarvis-like, you can add voice commands with the Raspberry Pi. Somebody did it here [1] using the Wit Speech API [2]
[1] http://blog.fmpwizard.com/blog/go-cortex-using-wit-ai-using-...
That tells a lot about France's so called "successes" (I'm French btw)