Say2GO Offers Asynchronous Voice Chat For Windows Users
by Calley Nye on July 6, 2008

say2goInstant messaging, a technology that actually predates the Internet, hasn’t seen too many changes over the years. The AOL Instant Messenger of today is not that different from the AOL Instant Messenger from 1997. There has been some new technology with the integration of instant messaging, VoIP, and webcam capabilities with services such as Skype, but two things have remained the same. Voice chat has been real-time, and text chat has been near-real-time.

Say2GO, a new near-real-time voice messaging system, is attempting to change this. The new technology, they refer to as “voicing” is closer to voicemail than IM. The product, currently in limited beta, provides users the ability to send and receive voice messages that are transcribed using voice recognition and sent as both audio and text. The desktop client uses the Microsoft Speech API for the voice recognition element, so it is only available on Windows and XP users are required to download and install the client libraries from Microsoft. Without the voice recognition library, the client reverts to being a standard IM client with a neat clean interface such as the contact list being displayed in cover-flow style.

The idea and technology is still fairly new, and it shows. After spending 10 minutes to train both of our speech recognition libraries, Nik and I tested Say2GO and we saw very poor results. Nik said “Another test of the speech recognition software,” and it translated into “man another said the phone is the victim mission the way you read into.” I saw fairly better results but they were still off. I had said “Why did I spend ten minutes training this software,” and it translated into “why didn’t spend ten minutes training in software for.” Nik has a thick Australian accent and an absurdly loud fan on his computer, so that may have affected the transcription.

MSN Messenger has a similar capability called MS Voice Clips, that they added in 2005, which enables users to send 15 second or less voice notes to friends through MSN Messenger. Unlike Voice Clips, Say2GO enables you to playback your recording, cancel, or schedule it to be sent later.

Microsoft seems especially interested in this technology seeing as how Say2GO recently won the First Prize in ISV/Software Solutions in Microsoft’s 2008 Partner Contest, out of 47,000 member companies. They are heavily promoting the new speech recognition libraries, as once you train your profile in one application it can also be used in other apps that use the same library.

With voice recognition where it is at at the moment the feature on its own probably will not be a strong incentive for users to switch to a whole new IM client - but Say2Go intend to support existing IM networks rather than building out their own. Voice recognition is only the first part of planned ‘added’ services that can be used on top of these networks. Currently Say2Go supports the ICQ network, with support for other networks expected in the near future.

Will we ever bury voice recognition? Read more at Techcrunch IT >>

Trackback URL

Comments

haha maybe in another 10 years they will be able to get these voice recognition things working right..

 

As soon as the Microsoft Speech library is at a point where it is working better and users have built up their profiles/training, they will tie it into Office, MSN Messenger etc. etc. The way it is at the moment, its not very accesible or easy

 

I really don’t know how this could ever work. event individual don’t work .. I can’t see how multi user voice recognition could work … but who can tel anyway in this “ever-changing” technology world ..

 

I hate voice recognition. For some reason they never recognize my accent
I have to say the same word over and over again. well maybe in 10 year that thing will recognize my accent until then I rather stick with writting.

Paula
http://www.websiteduty.com

 

This might be good for people who are blind, but other that, I doubt this will take off.

 
 

thought voicemail was dead?

 

This is a step in the wrong direction IMO. Michael’s earlier posting on vmail drives it home: “think before you voicemail, more and more people just find it annoying.”

Who wants more voicemail?

 

ill just use my phone

 

I tried to implement the same exact thing two months ago but the microsoft speech api is totally useless. Good idea, but it’ll take many years before speech technology can handle something like this.

 

Does “limited beta” refer to their product or their logo. I’m guessing the logo.

 

very interesting indeed

 

Very interesting indeed as it is coming near to the discussions we had on voice chat technologies way back in 1999 when we were having low speed connections ( as low as 4 KB) . The idea is to train the software to create a profile and like IM environments loaded into instant chat windows the profile is transferred to the other members who are participating in chats. The type script is converted to voice like the regular TT Engine converts. Like Mike and Mary TT profiles the user profiles also start speaking on our behalf. But real time conversion is critical process. First talk, let the engine convert it to text, transfer the text to other end, convert it to voice and its transfers on very low speed connections. It may be clumsy but applications could be plenty as they can translate it to other languages where this technology works wonders.

 

There is one immediate application which appeals to those who are in virtual teaching. The text transferred will be displayed in those chat sessions which gives the clarity of the topic being discussed as accents from different regions of the world varies. Those who attend distant learning programs and webinars find this very useful.

 

I gave the app a quick run, actually sent a couple of voice messages to myself in say2go messenger , and I really liked the concepts. The interface might be too funny for business use. No surprise with SR - I having a well trained speech recognition profile (i’m using it often for dictating texts) and a GOOD headset was getting almost 100% accuracy. I’m sure though that on the present stage of SR technology those who tried using this function with a built-in laptop mic and a poorly trained profile would get MUCH worse results.

Anyway, as I’m also using ICQ a lot for business converations, I might even try to switch to say2go/ICQ for a test, to see how much time it would save me if I’d be saying my messages instead of typing and occasionally correcting the misrecognized parts. Hey, but where’s text search through the history? ALso it would be good to have forwarding of saved voice clips to another say2go user added sooner rather than later.

 

one important thing which might be easily overlooked is that this gizmo attempts implementing what i think is the best posible speech-to-text approach. Say2Go makes my PC recognize my part of the talk making a good use of my dual core processor and 4Gb RAM and my indivudual profile while the PC of the guy I’m talking to recognizes his or her part of the conversation. no superservers with profile re-loading needed any more! pretty smart.

 

I would disagree with people disregarding Say2GO “voicing” as another incarnation of useless voicemail. As I understand, here you can exchange quick voice messages almost instantly, and unlike voicemail (or, fr that matter, a regular phone talk) - you can cancel recording your message if you have mumbled something incomprehensible)) and start over. not a bad thing, especially if I can then send it to a regular phone as a prerecorded voicemail (can I with say2GO?)

It will probably take some time for people to get used to voicing, just as it once took for email users to get used to text IM. I also doubt that Say2GO can take off as a widespread IM network if not linking to other major networks smartly. But their approach seems so complementary to existing instant messengers that it’ll probably soon be on a shopping list of big guys like Microsoft or Yahoo, or Google…

 

Well that sounds like a cool idea, if the voice recognition software is good enough, it would be cool if it could also translate the text so you could international calls.

 

Struth mate, whaddiya mean he ad a strooong Ostralian Accent?????? Don’t those galah’s at Moircrosoft know howda translate an Aussie???

What would happen if you had Indian accent then?

 

Instant messaging, a technology that actually predates the Internet

Ah yes, I remember those teleportation pigeons, in whom rolled-up notes could be secreted before they’d disappear before your eyes and rematerialise at a colleague’s house. I wonder what happened to them? I never saw any since the early 70s…

 

Leave Comment

« Back to text comment

Commenting Options

Enter your personal information to the left, or sign in with your Facebook account by clicking the button below.

Alternatively, you can create an avatar that will appear whenever you leave a comment on a Gravatar-enabled blog.