The Neon AI SDK is the first “full-source” software development kit for conversational AI developers.
“The only way to predict the future is to create it.” - Peter Drucker
“The future is voice, not keyboards.” - Richard Leeds, President, Neongecko.com Inc.
The pace of human-computer interactions is accelerating with the adoption of conversational artificial intelligence (CAI), speech processing, speech-to-text, and text-to-speech. Measured in words per minute (wpm), conveying information via speech is typically more than twice as fast as typing (speech = 140 to 200+ wpm, typing = 20 to 90 wpm). In addition, speaking is frequently more convenient and easier. By comparison, listening can be easier than reading — when driving, exercising, or doing chores, for example — but is only approximately 60% as fast (listening = 140 to >200 wpm, reading = <180 to >400 wpm). Also note that some audio players, including Neon AI, as discussed below, enable speed-up options that can narrow the gap.
CAI will soon be ubiquitous in human-computer interactions, with readers still at an advantage (especially with speech-to-text). Market data shows that voice input (and images) will be used for more searches than typing in 2020, and more than 50% of all human-computer interactions will be via voice by 2025. Consumers are waiting for hands-free ways to communicate with their devices, from digital assistants to lighting and appliance controls to automobiles to medical providers recording patient information. The price of adding speech processing has fallen to levels at which individual devices can be voice-enabled on the internet of things (IoT). The list of applications that will be converted from typing to voice, and from reading to hearing, is endless.
The evolution of voice processing
While speech recording and output have been available since Edison’s invention of the phonograph in 1877, the technology for speech processing was not possible until the digital electronics revolution, and advances in speech recognition have followed Moore’s Law (the density of integrated circuits doubles every two years). The following list offers a high-level summary of developments:
- 1952 — Bell Labs — Audrey — 10 digits
- 1962 — IBM — Shoebox — 16 English words
- 1976 — Carnegie Mellon — Harpy — 1,011 words
- 1985 — IBM — Tangora — 20,000 words
- 1990 — Dragon — Dictate — Consumer speech recognition
- 1993 — Carnegie Mellon — Sphynx-II continuous speech recognition
- 2002 — Microsoft — Office
- 2006 — IBM — Watson
- 2008 — Google — Voice Search for iPhones
- 2011 — Apple — Siri
- 2013 — China — Baidu, TenCent, Alibaba
- 2014 — Microsoft — Cortana
- 2015 — Amazon — Alexa
- 2016 — Mycroft.ai — Open source conversational AI
- 2018 — Amazon — Alexa application programming interface (API)
- 2019 — Neongecko — Neon AI Conversational Software Development Kit (SDK)
The AI revolution
Artificial intelligence (AI) is a modern term used for a combination of algorithms, data structures, and big data that statistically predict outcomes based on training corpuses (notably, neural networks). Prior to AI, the automation of human tasks was limited to mechanized Taylorism and the automation of menial tasks. The use of AI predictive algorithms has revolutionized many fields, from speech-to-text to translations to driving to advertising to medicine to politics, as well as the rapid automation of service industries, continuing what might be termed a Taylorism of service tasks.
The current technology required for user-friendly AI systems with speaker-independent speech-to-text — that is, conversational AI — is available only on specialized systems “in the cloud,” notably from Google, Amazon, Apple, and IBM. Within less than two years, the newest generation of AI chips will combine to make conversation-capable IoT affordable, potentially for a lower cost than a keyboard, and will increase privacy and reliability by removing requirements for network connections.
The next generation of CAI systems development requires more than an API
Most of the CAI development systems that developers and consumers use today are “closed” — they require the developers to use the services in exactly the way that the device manufacturers want, and they give all the user data to the CAI systems. The “walled garden” approach of Amazon, Google, Apple, Microsoft, and IBM, popularized by AOL in the 1990s, is making a comeback. By comparison, at Neongecko, we do not feel that “walled gardens” are appropriate for the future of this incredible technology.
We proudly offer an open-source solution — the Neon AI Software Development Kit (SDK) — with endless opportunities, and we encourage collaborative building on our code. We think that developers at thousands of companies, labs, and universities around the world should determine the course of the technology, not a few corporations. Unfortunately, the same cannot be said for Google, Amazon, Apple, and other AI companies.
While we are thankful for the contributions made by these major AI players, and although we use some of their software, we recognize that the time has come to take things to the next level. Now, you can choose a solution that works for your needs — not the needs of a few AI companies’ bottom lines — and not share your data with entities that go on to “share your data with trusted partners.” You can use our software completely independently of our servers and services. Our Neon AI SDK can provide the framework to customize your products, business, school, or home for all of your conversational AI needs.
The Neon AI SDK offers more options and better privacy
When using our service, you can use portions of other popular AI service providers like Amazon and Google. If this makes you nervous about your privacy, we have you covered. Our software acts like a proxy, so you can get industry-standard STT and TTS without having to give up all your private information. Furthermore, if you decide to go a completely custom route with your own server, you don’t need to use them at all.
The Neon AI SDK
The Neon AI SDK integrates advanced AI and natural-language understanding into a cohesive software engineering platform that is fully customizable and easy to use.
We offer Neon AI on a variety of platforms. Furthermore, we have created next-generation customized devices and offer Neon AI in a variety of off-the-shelf solutions. The Neon Nano installs Neon AI into any website in one line of code. Meanwhile, our Klat.com Adaptive Forums internet chat application, which is available on over 100 websites, hosts a full-featured artificially intelligent chat service based on Neon AI. The Neon AI app for Android phones combined our best features into the first open-source multi-platform CAI system.
Developers will be drawn to the limitless potential of our products. We offer lots of instructions in text and video on how to install the SDK, create skills, and develop your products with Neon AI.
The Neon AI SDK is a full-stack application with which developers can incorporate new AI skills and enhance core execution capabilities. Neon AI skills include real-time transcription, translation, conversations, device control, audible gestures (like claps), and much more.
The Neon AI SDK is available on our private GitHub repository. The Neongecko repository enables developers and users to install our technology on development systems and targeted devices (like the NeonU, generic Linux PCs, and Windows with a virtual environment). Devices and standard browser clients can connect to our server or yours.
Summary of differences between Neon AI and its competitors
Amazon Alexa, Microsoft Cortana, Apple Siri, and Google Home:
- Closed system
- Limited options
- Privacy concerns on their platforms with STT, TTS, location information, network data, etc.
- User license allows them to “share with trusted partners”
Neongecko Neon AI:
- Better privacy
- To the greatest extent possible, Neongecko doesn’t share information.
- Neon AI can use Amazon or Google services, and our server aggregates connections, which obfuscates individual users.
- Available for more platforms.
- Neon AI enables smart speakers, “view” devices, and PCs.
- Neon AI app is available for Android phones.
- Neon AI enables “standard” websites.
- Neon AI provides a conversation console in web browsers.
- Neon AI enables communications between platforms.
- Can use services wherever you want: our servers, shared servers (like Mycroft AI), or your own servers (like Deep Speech for STT and Mimic for TTS)
- More advanced skills (from database and website content integration to easy scripting)
- Open on all levels for customization and enhancements
- Enables users to work at many levels of programming
- White label system with license available to the Neon AI server
“Imagination is more important than knowledge. For knowledge is limited to all we now know and understand, while imagination embraces the entire world, and all there ever will be to know and understand." - Albert Einstein