Now there are three. Whisper for speech-to-text, Stable Diffusion for creating AI-generated images, and GPT-J for chatting with a large language model. These are the first three tools made available in our new public-facing AI Lab. Everything is open source and privacy-respecting.
Our AI Lab is a playground for people like you to experiment with some of the cool projects published by the open source community as part of the remarkable explosion of Artificial Intelligence capability happening in research labs throughout the world.
As we're tinkering with these cool gadgets, we're impressed by how the State of the Art in AI has, over the past year, made an enormous leap. As long-time researchers in machine intelligence, we've been familiar with obscure and cumbersome, unwieldy open source projects that take months to understand, weeks to install and configure correctly, and expert knowledge to tweak and keep running.
Now we're seeing slick and sexy, nimble Docker bundles which install in less than an hour and expose a gorgeous web interface that is well-designed and easy to run. Ordinary people now have access to software capabilities which are better than the best machine intelligence hidden in the largest research labs for decades.
We have a strong philosophical position on developing AI at the edge, so in keeping with that focus, we now have these three endpoints: Voice, Images, and Intelligence, available. They're free and come with TensorVoice's longtime commitment to security and privacy.
This is part of our ongoing pivot away from our core product. To be honest, our whole business model has been eclipsed by the latest voice technology which allows ordinary people to record a few seconds of audio and use it to imitate another person's voice. Although our systems are still secure because we use Multi Factor Authentication, clearly our best factor -- voice biometrics -- is sunsetting. We see the handwriting on the wall: the age of "your name is your password" is ending1. We do not yet foresee a way out of the arriving era of deepfakes. Businesses like ours must adapt or die. Pretty much all businesses are facing this in one form or another, but only a few of us are hit right in the solar plexus. We're fine. We're adapting.
We've been marketing secure voiceprint technology for years, which we honestly thought would be around for years to come. Our home page currently boasts of how voice biometrics is a reliable long-term approach to security. At the time we started marketing our solution, nobody expected what is happening today: that we'd soon have handheld devices which could easily hack such systems. At the time, such technology was enormously complex, obscure, and expensive. The future arrived more quickly than we expected, so we're pivoting to keep up.
We expect to be around for a long time yet. There is still plenty we can do with our skills in voice tensoring. We began our pivot by establishing this weblog in March. Now in April we're making a free, private, and secure AI Lab available, starting with these three open source projects. We're currently working internally to develop a new core product. Over the next few weeks, we'll be rebuilding our home page to reflect these changes, as well as developing a presence on social media.
Please explore and enjoy our new AI Lab. We're open for ideas on improvement like never before, so if you have suggestions for where to go next, drop us a line at ailab [@] tensorvoice [dot] org.
Footnotes:
1. The era of "Your name is your password" is ending, and apparently the era of "Roabobiog WinG Taing" is beginning. We generated this image using Stable Diffusion with the prompt: "a robot writing on a chalkboard." We're particularly pleased with this image because Stable Diffusion usually sees words as image patterns and therefore has a hard time generating readable text. This is readable. Maybe this has something to do with the fact that the robot has three eyes? Whatever the case, we feel this may be somehow meaningful: A machine intelligence created an image of a machine intelligence writing on a wall: "Roabobiog WinG Taing."
Image created by Stable Diffusion at sd.tensorvoice.org.