SoundHound Launches Generative Chat AI Voice Assistant Platform and Mobile App
SoundHound has released a new generative AI voice assistant platform and mobile app called SoundHound Chat AI. The new platform links SoundHound’s voice AI technology to third-party large language models like OpenAI’s GPT-4 for commercial uses, while the Android and imminent iOS apps offer consumers a voice assistant with conversational abilities similar to ChatGPT and other generative AI chatbots.
SoundHound Chat AI
SoundHound is aiming the Chat AI platform toward brands interested in creating or augmenting a multi-modal voice assistant with generative AI models. The platform can integrate broad models like ChatGPT alongside domain-specific databases with current news, weather, or personal information. Both the platform and the mobile apps use SoundHound’s new Conversational AI Language (CaiLAN) and Conversational AI Network (CaiNet) systems. CaiLAN creates the knowledge domains of local and changeable information used by the voice assistant, such as nearby businesses, traffic, and weather. Meanwhile, CaiNET judges what databases should be accessed to answer user queries and sets up the pipeline to provide the answer when it’s generated. SoundHound had already begun embedding these systems into its restaurant and automotive platforms earlier this year, adding generative AI options to its Dynamic Interaction voice AI for business services.
“SoundHound Chat AI is a combination of software engineering and machine learning that can power any experience, from navigation to restaurant orders. Speaking is much faster as an input than typing. Reading may be better than listening, but [SoundHound’s platform] can do it all hands-free,” SoundHound COO Michael Zagorsek told Voicebot in an interview. “Voice interaction is going to be the best way to tap into some of the LLMs and generative AI, but in a voice environment, you need multiple data sources. The platform is ideal for routing [queries] to the appropriate data source to generate an answer. Our natural language understanding and real-time speech-to-meaning let [the AI] knows pretty early on during the query which database to pick. SoundHound uses parallel tracking – so we are confident our system can handle queries and make sure that [the assistant] doesn’t have a nonsensical result.”
Generative Voice Assistant App
The mobile app provides a portal to SoundHound Chat AI by subscription, costing $7 a week or $20 a year. The app, as seen above, responds to “Hey SoundHound” as a wake word, then operates like ChatGPT with a voice. There are a half dozen female and male voice options, with American, British, Australian, and Indian accents, or you can give it the same voice as the built-in voice assistant on your phone.
“The automatic speech recognition and natural language processing are ours, as is the wake word and text-to-speech technology. The TTS voices are ones available on the market, naturally optimized for common words and phrases,” Zagorsek said. “The app is a whole assistant that can answer questions in real time, and there’s no more “I didn’t get that” as a response. “A lot of voice assistant apps have relied pretty heavily on web search results, but they don’t have to do that anymore. There’s elegance in how [the AI] ties it all together. It’s useful for everyone.”
The rapid release of the mobile apps and platform continues SoundHound’s already hectic year, beginning by predicting it would reach profitability this year. The company then raised $25 million in equity financing in January from undisclosed investors after reducing staff from around 450 to 200 people in two rounds of layoffs, but has continually beat stock analyst expectations each quarter.
SoundHound has gone down this road before releasing a stand-alone voice assistant mobile app named Hound to the general public six years ago this month. According to Zagorsek, the new app is “everything that made Hound great and more, an evolution of what was originally launched solely as a consumer app.” That’s quite literal as the Android version of SoundHound Chat AI has supplanted the same spot in the Google Play Store, with all of the reviews and update history for Hound intact back to its beta stage. The iOS version is likely to maintain the same history as SoundHound’s webpage for the Chat AI app currently opens Hound’s Apple App Store page. Nonetheless, Zagorsek is clear that he sees the SoundHound Chat AI app and platform as an entirely new chapter.
“The evolution of LLMs is really the next evolutionary step in AI. Before that, you had more localized machine learning, but the ability to aggregate developed patterns and machine learning is so powerful, I don’t think any of us has seen technology take off like this since the web,” Zagorsek said. “The virality of [generative AI] has influenced everybody. Everyone has heard of it. It’s a new superpower.”