WellSaid Labs Unveils Next-Gen AI Voice Platform, Emphasizing Enterprise Focus and Ethical Approach

With latest AI voice features, WellSaid Labs bets on a more cautious approach in a buzzy industry

AI voice startup WellSaid Labs is doubling down on its niche of enterprise customers and regulated industries -- hoping that a more judicious, behind-the-scenes approach will pay off for its business in the long run even as flashier rivals draw widespread attention and controversy. The company, based in Bellevue, Wash., launched a new version of its text-to-speech AI voice platform Monday with redesigned Studio software and its next-generation Caruso voice model, promising better workflows, improved audio quality, and fine-tuned controls, among other features. Unlike open voice-generation models that scrape public data, WellSaid's system is trained exclusively on licensed voice actor recordings, a closed-model approach that it says respects intellectual property and appeals to sectors such as healthcare, legal, and finance. WellSaid's latest release is a pivotal moment for the company -- the result of years of internal research now coming to market in a form that refines its focus on business and institutional users, said Chris Johnson, WellSaid's chief product and technology officer, in an interview. "We put our stake in the ground in being the best solution for enterprises in the market," Johnson said. "A lot of these innovations accrue to making that a reality for us." WellSaid, which spun out of Seattle's AI2 Incubator in 2019, works with large enterprise customers including LinkedIn, T-Mobile, ServiceNow, and Accenture. The company made an impression on the public in 2023 when NPR's Planet Money used WellSaid's technology to create a synthetic version of former host Robert Smith's voice -- a near-perfect replica that surprised listeners and showed both the promise and potential challenges of realistic AI audio. But WellSaid has struggled at times to break into the larger industry conversation. The challenge was underscored by its absence from a CB Insights market map of leading voice-AI startups -- topped by buzzy ElevenLabs, which has been at the center of controversy over the use of its technology to make fake AI voices of public figures and others. WellSaid executives say they're hoping to correct that specific oversight, but the challenge reflects a broader pattern among enterprise AI companies, particularly those in the Seattle region -- which often emphasize trust, governance, and regulatory scenarios in a tech culture still captivated by headline-generating Silicon Valley experiments and consumer apps. The company employs about 70 people, down slightly from a year ago, according to LinkedIn data. WellSaid has seen some turnover in its executive suite, including three CEOs in less than two years -- starting with founder Matt Hocking, then Brian Cook, and now Benjamin Dorr, who succeeded Cook earlier this year after serving as chief financial officer. "Every voice is connected to a real person, and that person receives royalties from the revenue that's generated on WellSaid," Dorr said on a recent episode of the Master Move podcast. "I think the things we do right by our voice actors allow us to do right by the enterprises that choose us, and I don't think everyone else can say that."

WellSaid pushes AI speech forward with faster, more natural voice production - SiliconANGLE

WellSaid pushes AI speech forward with faster, more natural voice production WellSaid Labs Inc., a developer of lifelike artificial intelligence voice technology, today unveiled the next generation of its enterprise platform, introducing major upgrades to its audio studio that make producing natural, customizable speech faster and more intuitive. WellSaid's core technology synthesizes voices that sound natural and human-like using proprietary AI model named Caruso. It was trained on exclusive licensed audio from professional voice actors, not on public sources. It uses its technology to produce natural speech in various styles that can account for accent, timbre, pronunciation and other expressive elements. "Today, enterprises require an AI voice solution that ships faster, sounds better and enables the ability to scale quickly while meeting compliance standards," said Chief Technology Officer Chris Johnson. The company upgraded its Studio product to provide easier production of speech with instant previews and fewer clicks to deliver speech audio. Users can fine-tune pitch, pace and loudness of voices at the word-level, while also adding multiple voices in one script or dialogue. To bring high quality audio to customers, WellSaid made audio up to 96 kilohertz the new standard, which produces natural clarity and more readily captures intonation and stress across synthesized voices. That includes the capacity for designing phonetic spellings for acronyms, brand names and borrowed terms. The platform also provides smart suggestions on potential phonetic spelling to assist users with getting pronunciation right. The platform covers a wide variety of potential words from numerous industries, including coverage for more than 9,000 medical terms, 500 legal terms and thousands more used across healthcare, aviation and industrial. All of them are backed by Oxford Dictionary guidance. The company also added 36 new voices covering global languages, including Arabic, Turkish, Persian and 18 dialects to enable localization of content. WellSaid also includes a large library of English-speaking voices that include accents such as Australian, British, Canadian, Irish and various regional accents from the United States. AI-powered voice generation has seen rapid growth as lifelike and expressive custom speech becomes increasingly sought after across industries. The market was valued at $3.5 billion in 2023 and is projected to reach $21.8 billion by 2030, according to market analysis firm Grand View Research. Companies such as Eleven Labs Inc. and Hume AI Inc. have raised significant capital this year, $180 million and $50 million respectively, for voice cloning and generation technologies. AI-generated voices now power intelligent voice agents that can understand natural speech, respond conversationally and take action on users' behalf. This trend has fueled a surge in voice-enabled technologies that sound strikingly human across devices, over the phone and in customer interactions. With human-like AI voices, companies can handle routine inquiries around the clock, schedule appointments and personalize outreach. Voice-capable AI agents can also contact customers directly with updates and tailor interactions to individual preferences and histories. Looking ahead, WellSaid plans to introduce additional upgrades, including a usage insights dashboard, and enhanced pronunciation and performance tools for managing emphasis, cues, variability and breath control.