ar7hur

A good reminder that an "acquisition proposal" without any indication of price is not an acquisition proposal. Asking so much information without any such indication is just crazy.

I strongly recommend this excellent post from Justin Kan: http://justinkan.com/the-founders-guide-to-selling-your-comp...

FYI in Clojure (with a live demo): http://duckling-lib.org

I appreciate you sharing your experience, this is something we founders are not enough prepared to go through.

Two comments IMHO:

- Would you have said yes and looked into the details of their offer, you would have realized that the $12M are not upfront. Through retention packages and revesting, most of the price is typically paid over 3 or 4 years. So the offer was probably less attractive than it initially looks.

- $12M is never a good outcome for investors (at least big angels and VCs). They make their returns only on big exits.

I love this kind of service, but extensibility is probably the most important feature and unless I'm wrong, your platform is not open (yet). We developers all use different services.

Hubot is extremely simple (based on regular expressions to match commands) and quite successful because it's so easy to write and use your own module [1]. Hubot being free and open source, what is your differentiation plan? I'm sure there is room for different players.

[1] http://hubot-script-catalog.herokuapp.com/

OK, now I'm curious. Please Show HN your photo-sharing app!

Fascinating. And 45 years later, here we are building photo-sharing apps.

Thanks for these excellent remarks.

We differentiate ourselves from the Android Speech API in several ways:

1) As a developer, with Google you have no way to customize the speech engine by providing your language model. Wit.ai builds a specific, customized configuration for each app. If your app is specific and you cannot tell Google what kind of thing it should expect, accuracy will be bad especially in noisy environments. Wit.ai builds a specific language model for each app automatically and in real time (each time Wit.ai learns more examples about your app, it's updated), and queries several speech engines in parallel. To do that it uses not only your data, but also relevant data from the community. This is the core of our value proposition and not something Google does provide today.

2) Google keep their Natural Language Understanding layer (what translates text to structured, actionable data) for themselves. Developers cannot access this. They're left with free text, but they often need actionable data.

3) Wit.ai is cross-platform. We have SDKs for iOS, Android, Linux, etc. [1] or you can just stream raw audio to our API. Android Speech API is just available on Android (well, you could hack it and use it from elsewhere but you're not supposed to, and you can be shut down anytime). More and more wearables and smart devices will run Linux. For instance hundreds of developers use Wit.ai on Raspberry Pi.

As for the Apple doc you linked, it's Mac only (no iOS) + it just recognizes a few phrases you provide in advance. I think it's a very old API that's still here :)

Regarding listen.ai, yes please we would love to have Google (especially Now) there. We have the Siri founder, the top Cortana guy, the former CEO of Nuance... but nobody from Google yet.

[1] https://wit.ai/docs

My personal definition of a seed round is a round where you don't give up any board seat or special power to investors. After a seed round you should basically work as usual (product, users, product, users, ... nothing else). By this definition, our round qualifies as seed. Managing a board takes time and energy and the more you can delay this, the better (from my experience).

That being said, this is a very subjective notion and everybody is free to have their own.

Yes open instances are free and unlimited. This is the cornerstone of our approach: we want developers to work together and share their training data. Natural Language is very hard and we need to join forces to crack it.

Thanks for the feedback, we'll try to make this easier to understand on the pricing page (yeah, natural language generation is hard for humans, too!)

Time is diffmatic.

Most commenters here find these demos very underwhelming and I must agree with them. Well known open source NLP packages like Stanford NLP or NLTK do much better on these (very academic) tasks.

I think the problem is that Watson's demos have been trained for very specific and narrow domains (like "cloud services" in the case of Resonance). What we really want to know is how easy (or difficult) it is to train Watson for our own domain.

If Watson has to make assumptions on my domain, then it will be really hard to build something useful with it.

OK, maybe unique was a strong word. But banks are starting to use speaker ID to log in to mobile apps, so the signature is, let's say, unique enough for practical applications.

Voices do have a unique signature. The technology do identify it is called Speaker Identification. See for instance http://research.microsoft.com/en-us/projects/whisperid

> Each person's voice is different. Some sounds, like "s", sound about the same no matter who says them, but other sounds, like vowels, tend to differ a lot from person to person. We use a special way of representing sound, the cepstrum, that captures lots of information, including the characteristic way you pronounce your vowels. Of course, someone could imitate the way you talk; fortunately, the cepstrum also captures certain fundamental characteristics of voices that are impossible to change. For instance, the length of your vocal tract -- the place where sound is produced in your body -- cannot be changed, and different length vocal tracts tend to produce cepstra with different characteristics. By identifying both the way you talk, and the way your body produces sound, WhisperID can do a great job of figuring out who you are.

The PIA Chrome extension enables you to open and manage some sites with speech. It was done at a hackathon using wit.ai

http://getpia.com

Thanks for sharing, nice work!

Quick question the Julius website says there is no English acoustic model available [1], how did you solve this? Do you provide a default acoustic model?

[1] http://julius.sourceforge.jp/en_index.php?q=en_grammar.html

All my Palo Alto neighbors who over water their lawns everyday surely don't look very desperate... (which makes me desperate)

No, PG built that not by giving advice to founders, but by selecting them.

Haha, nice example! If you want to do this you'd better use our full platform Wit.ai [1], that will arbitrate competition between the datetime entity (`tomorrow at three`) and the number_of_person entity!

[1] https://wit.ai

Hey sorry little downtime on the demo site :) it's back now

This is our first experiment with Rust on Raspberry Pi: https://wit.ai/blog/2014/09/12/office-automation-with-raspbe...

We'll share more as we progress.

Demo site updated, now shows partial parses.

We just updated the demo. Now you can try "see you tomorrow evening at 6" and see the partial parse.

SUTime is very good, like all the StanfordNLP stuff. We chose to do Duckling because:

- To my knowledge SUTime only supports English

- We wanted something that's easy to extend. SUTime is somewhat hard to extend, especially if you are not into Java

- We needed not just temporal expressions, but also monetary data, temperatures, quantities...

That being said, Duckling is still young and certainly not as proven as SUTime yet.

Absolutely. We are working on something called "assumptions" where Duckling informs you about what assumptions it made to produce the result (like: time was ambiguous, I chose PM), and then you can change these and get a new result. Coming soon.

I'm very excited about supporting Scottish colloquial expressions! Working on it now, thanks :)

You're right, fixing it now.

Ha! I was waiting for somebody to ask :)

The original name was Picsou (Uncle Scrooge's name in French) because the parsing strategy is super greedy. We liked the name, but when we decided to open source it we thought it may be hard to pronounce, so we switched to Duckling (keeping the duck link...).

Indeed, that's not supported yet, good catch. Will do.

Actually we are planning to port it to a language suitable for embedded use. Maybe Rust?

This is just a demo website, but in a real use case you would pass a context (for instance with the user timezone, country...) with your library calls.

You can also try the same sentence in French, you'll get February first.