Speechmorphing: Enhancing Customer Engagement with Natural, Expressive Voices

With Apple’s Siri, Google’s Now, and Amazon’s Alexa already being a big hit among users, what speech synthesis and voice technology, in general, can achieve in the future is limitless. Speech synthesis—the computer-generated simulation of human speech—has been applicable at the enterprise level for quite a long time now, and forms the basis of computational linguistics along with voice workflow, speech recognition, and natural language processing. California-based Speechmorphing is a personalized speech technology company primarily focusing on the speech synthesis aspect with a mission to enhance human-machine communication. Speechmorphing offers products and services that help brands build smart, expressive, and emotionally-intelligent voices unique to their businesses, in turn, broadening the practical applications of synthesized speech. “By introducing a voice technology that supports both personas and emotional intelligence, two key elements that are missing in traditional synthesized voices, Speechmorphing is helping to improve human-machine communications. We aim to speed up the proliferation of synthesized speech for real world applications,” remarks Dr. Fathy Yassa, Founder, President, and CEO of the company.

Speechmorphing doesn’t merely offer a text-to-speech technology but rather aims to transform the traditional practice of providing a generic, flat tone for conversational interfaces by instilling emotions and expressiveness into the voice. The continuous evolution of conversational platforms coupled with a promise to provide a more human-like experience eliminates the relevance of one voice for all kinds of chatbots and voice assistants. Today, most organizations are inclined toward having their own unique branded voice. By redefining methods of speech synthesis and focusing on the quality of voice, Speechmorphing is on the right track, enabling computers to not only follow the syntax of a language but also understand the context precisely. The company can customize as well as personalize any voice by adding emotions and expressiveness to it along with an ability to change tones dynamically, which is one of the biggest advantages of Speechmorphing’s technology.

As Speechmorphing’s voice has high emotional quotient (EQ), the tone can be automatically adjusted based on how the user at the other end talks. For instance, in the case of an agitated user, the computer can respond in a calm or apologetic tone while when a consumer is angry, the tone can automatically be modulated to sound relaxing and soothing. “Although our goal is to make human-machine communication more sentimental and human-like, we are aware that customers don’t prefer a voice that doesn’t even fractionally sound like a computer,” adds Shing Pan, VP of business development at Speechmorphing. The company successfully maintains a thin line between the two by offering a Voice-on-Demand service—Smorph®— and delivers top-quality voice at a fairly lower cost than its competitors with a shorter turnaround time (TAT). As opposed to the prevalent practice that involves hours of recording and TAT of up to several months, Speechmorphing requires only ten to thirty minutes of recording and only a few weeks of TAT to deliver the final voice.
User-friendly ‘Smorph’ for a Natural Branded Voice

Speechmorphing’s Smorph Voice-on-Demand service is bolstered by an advanced AI and deep learning-enabled voice technology. While quick implementation speeds up the time to market, the affordability and quality of Speechmorphing’s voice broaden the practical applications of synthesized speech. As the voice is nothing but an extension of the brand, organizations have started to realize the need to understand their customers better, which has resulted in an increased time spent on sentiment analysis. However, the findings of this research might fail to create the desired impact if the chatbots that these organizations use have a monotonous voice that doesn’t quite communicate emotions.

Our technology enables the personalization of voice, adding emotions and expressiveness, so we can change the tone of the voice dynamically through the conversation

Speechmorphing is very well positioned to assist such organizations in making the most of their sentiment analysis investment with regard to natural language understanding. The company’s user-friendly Smorph Voice-on-Demand service helps organizations create a voice that’s unique to their business and allows them to start either with existing or new recording of their brand ambassador. “We believe that the voice must have its own persona and expressiveness to be effective enough and hence, we model it with a plethora of styles and domains, from tones and demeanor to pronunciation and lingo,” states Shing. Speechmorphing provides several tools and granular controls within the text to enable organizations to further adjust the voice’s mood, volume, pitch, speed, intonation, and more. By equipping voice with personality, interest, and emotional intelligence, the company ensures optimal customer engagement for its clients.

"Speechmorphing’s unique ability to create high-quality and expressive synthesized voices with a small amount of speech data makes custom voices a true option for companies and applications. It particularly addresses the growing need for companies to have personalized and branded voices for their digital agents"

Easy Integration with Clients’ Existing Systems

Clients from various industries including banking, finance, contact centers, healthcare, retail, and more approach Speechmorphing to modernize their existing interactive voice response (IVR) systems as well as enable bots with a distinct branded voice.
An ability to seamlessly integrate with organizations’ existing conversational platforms allows the company to serve all industries, although with a predominant focus on the customer care sector. From providing neutral tones for the banking industry and energetic sound for sports announcements to a variety of animated styles for book reading, Speechmorphing does it all, based on the clients’ specific needs.

Requirements from some of these clients include pronouncing the names of companies correctly in adherence to the associated lingo rules. “In the case of companies with foreign names, for example, in Czechoslovakia, the sound of the letter ‘c’ is synonymous to ‘ch’. Our technology makes such alternations automatically and accurately based on the desired context,” claims Meir Friedlander, EVP of product, operations and business development at Speechmorphing. For customers of a bank wanting to check their account balance or details of recent transactions through a mobile application, the company provides the capability to repeat the numbers or even mention the currency if asked by the user. In addition, Speechmorphing also helps clients develop a localized or personalized marketing campaign for audiences of all kinds. The company has also been assisting call centers in resolving issues with regard to the lengthy waiting period that a customer has to experience when their call is transferred from one agent to another.

One of Speechmorphing’s clients from the banking industry faced a similar issue with customers calling to check their account balance or the details of previous transactions. The client receives around a million such calls every month. To improve the efficiency and quality of their service, the client planned to support such calls with a bot or a combination of the existing IVR and a bot. “They required us to incorporate the same voice as their IVR into the bot, which we did successfully. This ability to offer the same voice across multiple distinct systems makes us unique,” points Friedlander.

As Speechmorphing continues to upgrade its technology, it also faces the hurdle of curtailing the price while delivering the same quality with less than ten minutes of recording and less than a week of preparation and training. In spite of these challenges, the company believes it is on the right track to fully automate its platform by 2020, thus bringing down the TAT to a few hours and making the technology affordable for a larger audience. “With a team of experts in artificial intelligence (AI), we are constantly taking a step further. Currently, we are considering a product that can read text messages and emails in the sender’s voice,” says Friedlander. Speechmorphing is also looking forward to selling professional tools directly to clients in the future, instead of merely providing voice-on-demand services. The company plans to realize this transformation by encapsulating all the voice-on-demand service-related rules and methods to define personas relevant to a business, which will help organizations build their branded voice by themselves.


San Jose, CA

Fathy Yassa, President & CEO, Meir Friedlander, EVP of Product, Operations and Business Development and Shing Pan, VP of Business Development

A personalized speech technology company with a mission to improve human-machine communications by delivering natural and emotionally-intelligent voices