Typing to communicate and reading text (as you are now) is certainly 'old school' and certainly not as pleasant or efficient as talking between ourselves. Speech interaction with humanoid figures ('avatars` of all kinds of shape, race, gender, age and appearance) is a `natural` in the interactive web as well as computer communications for education, medicine, banking, arts, entertainment and gaming. Whether it be for home, kiosk, or mobile apps, interactive speech recognition avatars are here to stay, at least for those who would like something more than 'typing'.

We are available for consultation, design or production on speech-interactive user interfaces and 3D avatars for demo, display or product development. We have designed 3D avatars for speech-interactive 3D scenarios for corporate and art clients, as proof-of-concept demo or as works of media art.

Speech-recognition software, provided by client or using our previously licensed demo SR software, is required for all speech-interactive projects.

Our design teams are constituted on a contract basis and tailored to the intended project. The Project Lead Al Razutis has years of production expertise which includes 3D animation, speech-interaction for demo and web, and award-winning creations in stereoscopic 3D, motion-pictures, and VR worlds.

Contact Al Razutis with your requirements.

All video links open separate windows - enable 'pop-up' functions to view demos:

Demo-only samples of speech animation; updated from previous client work for Mission Corp. - 'Sally': Avatar video clip 1 ('Welcome...') Avatar video clip 2 ('Portfolio...medical center') Avatar video clip 3 ('Concert seating') - < 30 sec. clips - < 1mB for Win Media (.WMV) player. Demo-only samples of speech animation; client work for Sonora Technologies Inc. - 'Sonora': Avatar video clip 4 ('Welcome...') Avatar video clip 5 ('I was created by...') - < 30 sec. clips - < 1mB for Win Media (.WMV) player. Model/animation by Andrew Hogan

---

About speech-interaction for the web:

Speech-interaction graphical interfaces (you are talking, not typing or employing pointers) are inevitable for game and other applications like web browsing, chat, entertainment, security, remote control, medicine, e-learning.

Issuing voice commands is more 'human interactive' than punching hotkeys to generate actions. Talking with a interactive avatar (humanoid) is more 'social' than typing (waiting, and waiting, for text or action response). As with stereoscopic 3D game and display applications, voice-interaction is coming to the near future. Competetive companies and design houses should take note that old bandwidths are being reinvested with new streaming solutions.

The old 2D web - chat room typing

3D web gamers want to converse, not type / read

We like to chat, meet people, engage in 'relationships'. Human interaction is typified by conversations with other humans, and speech-recognition (SR) makes web simulations with avatar-humanoids a two-way conversation.

For the http web, a stable browser environment is required to support the speaking user interacting with web content. The major choices out there (in the marketplace) are dominated by 'speech to text' functionalities which focus on 'document' or navigation of content. A 2D (eg. hyperlink, Flash action script, embedded media, video animation) or 3D scenario (VRML, Java Applet, API, full poly animation, game engine animation, QTVR) is developed to interact in real-time with the spoken commands via http servers.

Proprietary rendering engines utilize different I/O, rendering, streaming and back-end strategies and the battle.net is ample evidence of multi-user interactions and simulations.

About avatars - face mapping - speech interaction:

The avatar can be 'realistic' or 'fanciful' but the graphic artist not the technician gives it a personality and telling features (expressions, details, movement) that correspond to familiar 'cues' (eg. eye movements, mouth shapes) that are necessary for exciting human speech interaction. Texture mapping of photo-realistic human images is an economical way to 'humanize' a poly-mesh avatar.

Producing a humanoid form and animating this form in a designed architectural space is the first step to creating an interesting 'interactive environment'. Flat scrolling maps with 3D avatars moving about are on their way out.

The speech-interactive scenario is based on embedded speech to text (code) functions of the browser to activate specific avatar (humanoid) responses and delivery of web content (response). The avatar exists in a library (database) of available responses; the recognition of user input is based on a grammar - syntax determined by the speech-enabled browser and an intelligent (A.I.) back-end that 'recognizes' the user in terms of a unique history of personal parameters. In this capacity, the speech-recognition capabilities of the browser environment must be accurate, and the avatar - content response must be equally accurate and timely.

The avatar visual characteristics (skins) are subject to viewer choice input, from anything in the viewer's skins pile. It could be a 'self-portrait', an 'alter-ego' movie star or...? The avatar can be photo-realistic or stylized characature, photo-based or CGI synthetic.

In the marketplace the viewer sees the artistry, design, animation and the delicate work of software engineers is hidden. That's were art and technology are put to their test.

The market will discard a poorly performing (robotic) avatar, however intricate or complex the software design. Many projects that did not incorporate effective (artistic) graphic design have gone into the marketplace trash bin.

Today's focus on portable browsing devices, freedom from keyboard and mouse, make speech-interactive avatar development all the more relevant. It is rather astounding that major players such as Microsoft, and game publishers, have not refined and shipped functioning SR avatar packages along with their software. Perhaps the opportunity is there for the individual entrepreneur or start-up to realize this vision and its future.

PRAXIS: