Ajay Juneja : Speech

So what have we been up to lately at Speak With Me?

ajayjuneja — Wed, 19 Nov 2008 18:08:00 GMT

I noticed that this week I got five emails from friends asking "So what's been going on?" or something to that effect! I figured this would be the easiest way to tell everyone!

Our economy has been a mess...

We're still raising and have been able to raise money.

We're on track for a really awesome product ready towards the tail end of next year. I hope it doesn't slip!

If you know me or attended the United Cerbral Palsy Foundation Charity event in Philadelphia, you might have an inkling of what the first product might be.

If you don't know me then know that the first product is something for your car.

But Americans aren't buying more cars. That's okay, they still have their current cars, they can put this into them!

The current system has been in my car for 3.5 years and 60,000 miles. Tested in the rain in Seattle while going through a tunnel, tested on the highways of California, Nevada, Oregon, and Washington.

I did distraction testing during an aggressive driving course with some friends & some of our investors. Prior to this, the usability tests showed a <1% difference in lap times between using the system while driving and not.

We demonstrated for real, that using Speak With Me's system while driving causes next to no distraction. How? Because the car behind me had trouble keeping up with me. Imagine his surprise when discovering I always had the music I wanted playing, and that my car did not have two turbos. (It does have a clutch-type 100% locking LSD which really helped my corner exit speeds, and the suspension is aligned and corner balanced).

What about those swanky iPhones, the new Android phones, the new Blackberry Bold & Storm, and the Windows Mobile smartphones? And Google came out with it's voice app this week? Isn't that a competitor?

The speech apps that are out there for mobile phones are using speech recognizers with complex language models. They work - kinda sorta - well they think about working. Are they competitors? Kinda, sorta, but not entirely.

As you can see from this review of Google's voice search added to their mobile app by John Markoff, the NY Times reporter and author of the fabulous book, What The Dormouse Said,

speech recognizers on their own, still aren't up to the task.

Speak With Me is different. We help the recognizer understand context. Think of it as a debriefing session from your assistant before you go on stage. That's how we're more accurate.

But gosh darn it, why haven't you shipped anything yet so we can try it out?

Because half baked cookies are kinda mushy. We could get a phone app out faster, but it would sacrifice the time to market of our other key product which is in the GPS market. The two products share some similarities, but have some differences too. One has the data on the device, one has the data in a cloud. So we made a tough decision this week, to stay on track with the GPS product, and hold off on the smartphone related products till we're a bit further along with the GPS product. Back in 2005 when you first read about us from Robert Scoble's Blog or Techcrunch, there were very few devices capable of running our software in the mobile and embedded space. Today there are tons, and everyone - not just Speak With Me, is scrambling to make the best use of the computing power we have today to give everyone fabulous software for their mobile devices. The phone today is like the PC world in the early eighties, I agree with Steve Ballmer on that point.

It is interesting how the internet has really led to much faster demand creation. What if the New York Times had an article about Xerox PARC in the seventies - would we have all had the GUI before the early eighties? I'm curious and I think I'll ask John Markoff and Doug Englebart that question directly in a couple weeks. SRI's doing a 40th anniversary event of the mother of all demos - on December 9th. Doug as you may or may not know, was the inventor of the mouse. Just think, he demoed in 1968 - what we as the public didn't get to use until 1982. Wow.

So how do you make a good cookie?

By using the highest quality ingredients, like Ghiradelli Chocolate Chips, Organic Cane Sugar, organic eggs, pure vanilla extract and not imitation. Then bake them for the right amount of time. You'll end up with something spectacular, just ask anyone who's eaten my cookies :) When they're ready, you'll know...

You didn't expect me to tell you the ingredients within Speak With Me, now did you ;)

Speak With Me - Video of our Navigation Application!

ajayjuneja — Fri, 04 Jan 2008 22:37:00 GMT

Hey everyone,

We now have a fully functional navigation system + Media Player + Cell Phone integration with StreetDeck + Microsoft MapPoint.

Please check out http://www.speakwithme.com for more info!

Special thanks to our friends at StreetDeck and the whole Speak With Me, Inc. team!

We are fast working on now porting our finished platform to Windows CE, QNX, Linux, Mac OS X, and Symbian!

Happy CES and MacWorld everybody - we'll be at CES showing some private demos this year!

Know any good T-shirt / Silkscreening vendors

ajayjuneja — Sun, 08 Apr 2007 06:40:00 GMT

...We're going to be making some Speak With Me T-shirts soon!

Please refer us some high quality and well-priced places!

Speak With Me is hiring!

ajjuneja — Sat, 16 Sep 2006 23:55:00 GMT

Well, this blog has been quiet for a long time. There have been lots of things in progress, most of which I unfortunately haven't been able to publicly talk about.

One thing I can talk about though, is that Speak With Me is hiring!

About Speak With Me:

Speak With Me is a company that is changing the way we interact with our electronic devices to access information. We plan to be commonplace in your home, in your car, and on the go. We flatten the user interface and make previously hard to access information easy to find. Mike Arrington of TechCrunch reviewed Speak With Me's technology a bit less than a year ago, and was quite enthusiastic about having our future products in his own car (rather, he seemed to want to steal mine). At this time, we've got most of the executive team in place, and we need the rest of the staff to be hired in the upcoming months.

We are looking for:

1. Product and Project managers

2. Software engineers and Senior Software engineers who like working with embedded systems You do not need direct experience with Speech and Dialogue systems for these roles, just be a great software engineer(C++, embedded systems)

3. People to make dev. tools and enhance our API -- if you've worked on Visual Studio, Eclipse, or other software - please contact me.

4. User Interface (UI) designers

5. Tech Writers.

6. Marketing and Sales people

7. Sales and Application engineers.

Speak With Me is located in the San Francisco Bay Area, and will be expanding over the next several months.
If you are interested, please contact speakwithmejobs AT gmail DOT com

Using the dialogue manager for 13 hours straight was as addicting as crack...

ajjuneja — Tue, 28 Feb 2006 15:14:00 GMT

Nothing like changing your tunes at 90 mph while going through the tunnel on I-90 in the rain... while keeping all my concentration still on driving. How addicting? Almost as addicting as track day at Laguna Seca is for a speed freak and car nut like myself...

Oh, also while in Seattle, I attended the Podcast Hotel, caught up with Chris Pirillo and made a few new friends, including Bre Pettis and Cielito Pascual. Bre and Cielito so graciously helped me film a video of the dialogue manager in Seattle, which will be broadcast on Bre's Blog in about 2 weeks as a Podcast.

What in the world was my car and me doing up in Seattle for a week? Well, a roadtrip to visit friends of course! Oh and speaking of friends, I do believe I owe a thank you to an old friend who tried to set me up two and a half years ago with a group of people in Seattle I happened to finally meet with this time around...

1835 miles of driving with the dialogue manager under constant use...

ajjuneja — Tue, 28 Feb 2006 14:52:00 GMT

So... I finally worked out my vibration issues with the hard drives -- first by switching to laptop hard drives, then by adding some creative engineering:

and special foam and at first things were better than before, but not quite good enough for all the various road conditions out there.

The problem?

The computer needs to be rigidly mounted. Bungee cords weren't enough, and because I was tinkering with things so much, I had removed the wooden panel framing the computer as seen in this pic:

I put that panel back in, and with the computer now firmly held in place, ALL my vibration problems were solved. I tested this on my drive back from Seattle to San Francisco and throughout the Seattle and SF Bay areas as well, and everything worked great! I wonder if this setup would hold up on Laguna Seca...

Review of Speak With Me prototype system by Mike Arrington

ajjuneja — Mon, 31 Oct 2005 00:48:00 GMT

Well, as I said, we officially started Speak With Me. Mike Arrington of Tech Crunch just did a review of a very early beta of our system, which he saw at TagCamp

The review is here:

http://www.techcrunch.com/2005/10/30/speakwithme-control-your-car-by-voice/

We fully plan to exceed your expectations. As for who our customers are, we can't say yet, other than that we are targeting both aftermarket car stereos and navigation systems, as well as those sold as part of a car you buy off the lot as well.

Special thanks to everyone who's been a great support to us so far, you know who you are!

Oh, and to anyone who even thinks about stealing my car the way Mike suggests, there is a very sophisticated alarm as part of the car that pages me if you so much as brush by it when you shouldn't.
You can however, see my car at the Toys for Tots car show in Los Angeles Dec. 11th, 2005. I am part of the Team Transport show squad.

The Trials and Tribulations of a Startup

ajjuneja — Thu, 01 Sep 2005 06:28:00 GMT

So, as some of you know or may have been able to figure out, I'm starting a speech company.
No big secret there. What I found out along this path is that there are two types of people out there, the selfish type
and the ones who consistently lend a hand without thinking "What's in this for me?"

What I've also found is that the people at the top of their game in this world, the true "masters of the universe,"
tend to be the type to help others out, no strings attached.

I'd like to take a moment and publically thank everyone who has consistently been very helpful and also when needed,
gone out on a limb for me. Sometimes that limb is only a twig!

Alex Rudnicky
Anand Manikutty
Andre Vanier
Arthur Chan
Ashleigh Anderson Felber
Ashu Kumra (Ash Gianni)
Catherine Copetas
Christian Chu
Ed Lin
Itamar Drexler
Jim Bai
Jon Anderson
Kimberly Lutjen
Manjul Bose
Mark Stehlik
Melodie McBride
Mike Shafer
Mom
Renee Lo
Ricky Houghton
Robert Scoble
Sarah Revi Sterling
Saureen Shah
Sean Patterson
Susan Trainer
Ted Havill
Vinod Krishnan
Yan Ke
These types of people are rare, and to be cherished dearly.
THANK YOU!!!

If you have a story to share about someone remarkable, feel free to either link it in the comments or mention them there.
People on this list are welcome to make fun of me, make jokes about me, or do whatever they feel like in the comments :)
If one or more of these people happens to also be one of your friends, you are a very lucky person, and consider yourself blessed!

Got scobleized...

ajjuneja — Thu, 18 Aug 2005 16:43:00 GMT

I got scobleized this morning!

So, yesterday on the way to the Geek Dinner, I decided to give scoble a ride so we could get into the carpool lane. As what I'm doing is now visible to people taking a ride with me, he got an inside scoop on some little things I am up to...

This is also why my blog has been so quiet for over a year. Want to see a sneak peek of the future? Come to the BarCamp.

For now that's all I'll say. When the products get made and are closer to release, there will be more noise. Consider this the teaser.

Sphinx 3.4 coming out soon...

ajjuneja — Thu, 04 Mar 2004 20:11:00 GMT

So I just attended a talk by Arthur Chan, formerly of SpeechWorks (now ScanSoft), who is now staff at Carnegie Mellon, and he's working on a new version of the Sphinx 3 recognizer (Sphinx 3.4). It's a very sophisticated approach using a tree-based lexicon (which improves speed, and marginally reduces accuracy of words that are slurred or cut off, since the language model cannot boost confidence scores with this approach), and a continuous HMM (Hidden Markov Model).

Results seem promising -- it is significantly faster than the original Sphinx 3 (which had trouble running in real time except on very very fast machines), for a marginal word-error-rate increase.

We'll need some testers and developers on this soon -- ping me if you'd like to play around with what will be, the best open-sourced (BSD license) recognizer around. I'll be extending a hand to help with this effort some too, though more so after I graduate and have more time to do so.

Kudos to Arthur, Ravi, and Jahanzeb for this effort! And yes, on my agenda in the future (6-12 months from now) will be Ariadne + Sphinx 3.x support. Too many things to do right now like graduating, going to Germany, then off to the Bay Area again (assuming I get offers from at least one of the two jobs I'm seriously looking at).