Key Takeaways
- Seasalt is constructing customizable speech recognition tech for enterprise name facilities.
- The founders offered their final startup to Baidu in 2017.
- The corporate companions with cloud communications large Twilio.
After promoting their final startup to Baidu, a pair of tech vets are leaping again into the crowded house of voice and speech recognition with an organization known as Seasalt.AI.
The startup sells a software program platform to enterprise firms with contact facilities. Builders can use Seasalt to construct apps, units and companies that talk conversationally with customers.
The corporate was based by Guoguo Chen and Xuchen Yao, consultants within the discipline of voice and speech recognition software program. Chen created the “OK Google” hotword for Android and co-authored a speech recognition venture known as Kaldi, which Nvidia ultimately built-in into its graphics card. Yao in the meantime is a Johns Hopkins College PhD graduate who beforehand labored on the Allen Institute for AI (AI2) incubator in Seattle.
In 2015, Chen and Yao co-founded KITT.AI, a startup that spun out of AI2. One of many firm’s hottest merchandise was a customizable wake phrase engine known as Snowboy, a software program toolkit that allowed builders so as to add verbal hotwords to their very own {hardware}. The startup additionally launched ChatFlow, a framework for builders to construct chatbots.
Baidu acquired KITT.AI in 2017. Chen and Yao labored for the Chinese language tech large for 2 years, leaving in 2019.
Seasalt, which has 22 staff, supplies a customizable speech recognition engine. The startup describes its tech because the “subsequent era” of conversational AI. It raised a small seed spherical when it initially launched in January 2020.
The corporate sells six purposes, which work in tandem as a part of its full suite of companies, listed under, known as “SeaSuite.”
- SeaChat lets customers create a framework for automated chatbot responses.
- SeaCode is a software program improvement studio for conversational AI. Customers can use the platform to construct instruments corresponding to chatbots.
- SeaVoice is a speech-to-text (STT) transcription function that may be personalized to know completely different languages and nuanced speech, amongst different makes use of. This device additionally has a text-to-speech (TTT) function, which will be personalized to sound like Tom Hanks or David Attenborough.
- SeaMeet‘s secretary-like options can be utilized in conferences and conferences. It could actually establish as much as 12 distinctive audio system within the room. Customers can prepare the mannequin to offer computerized assembly minutes and follow-up notes, amongst different actions.
- SeaWord will be personalized to go over textual content to extract significant info. The device may also be used to focus on and redact phrases like identifiable info.
- SeaX is a device designed for contact facilities. It could actually automate responses to incoming messages, calls and social media, amongst others. The software program additionally incorporates a device that decision middle brokers can use to transcribe and categorize incoming calls from prospects.
Seasalt goals to supply instruments which are able to understanding nuance in each speech and textual content. Its primary use case is in enterprise contact facilities. These firms use the software program to not solely monitor and consider their brokers, however to additionally mixture voice information to extract insights.
International companies have to function name facilities in a whole lot of nations, which means they’ll inevitably encounter low-resource languages and accents in each place they function. In America, for instance, there are a minimum of 24 dialects of English.
Seasalt has about 50 prospects, together with massive enterprise firms in Southeast Asia.
“For any enterprise firm, in case you have some actually bizarre spelling or some technical jargon, we will maintain that,” Yao informed Startup.
Seattle has grow to be a hotbed for NLP-focused startups, a lot of which spun out of AI2, which focuses on such analysis. Its present roster consists of Xembly, Learn, Unwrap and Increase, amongst others.
Spoken Communication, a Seattle startup that offered speech recognition tech to name facilities, was acquired by Avaya in 2018.
Yao stated that because of the pandemic, there are lots of cross-border e-commerce websites popping up in north and Southeast Asia, promoting their merchandise abroad. That is making a tailwind for Seasalt, he stated. He added that the area is presently underserved by opponents.
The decision-center expertise market consists of many current gamers. Tech large Google sells its pure language capabilities in a packaged resolution known as Contact Heart Synthetic Intelligence. Amazon and Microsoft even have their very own companies: AWS Contact Heart Intelligence and Azure Cognitive Companies. Different notable gamers embody Deepgram, Five9, Avaya and eight×8.
Yao stated Seasalt wouldn’t have the “muscle” to compete towards Five9 or different publicly traded firms if it didn’t have its partnership with Twilio, which bolsters its gross sales chain. He defined {that a} lesson he realized from KITT.AI was that software program itself just isn’t what will create a moat for the startup. As a substitute, he added, it comes from its present distribution, commercialization and buyer base.