#1 – You’re Paying Them To Promote Their Brand
The rise of Alexa and Google Speech as a service has been swift. Many of the leading brands have signed up to using their off-the-shelf speech systems, building voice skills for their customers or integrating them into their products as the only voice interaction method. Just like “Intel Inside” of years past, we now have, “Works With <Alexa/Google Assistant>.” However, this is not an easily removed sticker that tells your customers what type of microchip is inside their computer. It is a core interface of your product or service that two of the largest corporations on the planet use to gather your customer’s data as they interact with your products. Like it or not, you are paying for a service that promotes another company’s brand while at the same time feeding them data to make them smarter.
#2 – You Have Limited Access To Your Data
When deploying speech technologies, understanding the domain (how and what your customers are saying) and acoustic environment are the two key starting points to ensure the best accuracy of your system. Having access to data is critical to be able to tune and adapt your language and acoustic models to increase recognition accuracy and improve your customer’s experience when interacting with your service or product. For example, if you’re deploying voice command-and-control for manufacturing equipment, you need to tune your acoustic models to the noisy environment where interactions with the voice system occur. In such a case, a speech system that uses models tuned to standard acoustic environments will be an exercise in frustration for your users due to frequent misrecognition. Unfortunately, the double-edged sword of using platform speech service providers such as Amazon and Google is that you don’t control your data, and for the most part, you have no access to it. As a result, it will be challenging to deploy services for non-ideal environments, to create a domain-specific voice recognition system (such as in manufacturing, finance, legal, medical, government, or military), or obtain data to improve the service continuously.
#3 – Your Customer Data Is At Risk
Security and privacy are at the forefront of discussion these days. In the last several decades, there has been fantastic progress and benefits from technology, but it has also slowly eroded our privacy. Bugs in consumer voice-assistants accidentally share our private conversations at home with other users, Facebook knows everything about us, and our smartphones track where we go and places we frequent. We now see increasing awareness and concern about the loss of privacy from customers. Deploying voice services on Amazon or Google guarantees that your customer data is no longer in your control. They will use your audio and transcription data to increase the overall accuracy of their speech recognition platform, and if you handle data of a sensitive nature, such as banking, medical, etc. you should carefully consider the risks.
So How Do You Take Back Control?
Cobalt Speech & Language is a firm founded by luminaries in the speech recognition industry. Customers choose Cobalt to:
- Implement innovative voice services customized to ensure high accuracy. Some of our projects include:
- Custom voice assistants (voice bots and chatbots) for retail, finance, government, military, and medical use cases,
- Phone call transcription and word spotting for customer service call centers,
- Courtroom and legal transcription,
- Voice command-and-control for appliances, smart home, medical and manufacturing equipment,
- Disease detection by analyzing changes in voice patterns,
- Voice biometrics for security.
- Build robust models for noisy and challenging acoustic environments,
- Deploy Cobalt’s custom speech solutions on their premises to ensure their data stays secure (e.g., private cloud, embedded, etc.). We’ll enable you to leverage the data to continuously improve the service you provide your customers, even as you control all of the privacy decisions.
With Cobalt’s assistance, you no longer have to worry about your brand being diluted or overshadowed by the large platform companies, or your data being used to make your competition smarter. Let Cobalt help you take control over one of the most exciting and intuitive user interfaces that adds significant value to your customer experience, speech recognition. Get in touch to see how we can help you.
About the Author
Victor Jablokov is a serial entrepreneur and founder of Yap and Wallflower Labs. Yap created the world’s first carrier-grade scalable fully automated speech recognition cloud platform to convert voicemails-to-text, which in its era was the first platform to beat the accuracy and speed of human transcription services. Amazon acquired Yap in 2011; their team and technology helped give birth to Alexa. Victor now advises startups and assists Cobalt Speech & Language with strategy, business development, and sales.
