Voximplant: Turning up the volume on customers served using Google Cloud Speech-to-Text API

About Voximplant

Voximplant is a serverless communications platform that helps companies develop and implement real-time voice, video, and messaging solutions. Founded in 2013, Voximplant is now based in the US and handles hundreds of millions of calls for its customers worldwide.

Industries: Technology
Location: EMEA

Cloud communications platform Voximplant uses Cloud Speech-to-Text API to help businesses build simple speech recognition tools, boosting the number of calls they can handle.

Google Cloud results

  • Enables Voximplant clients to build speech recognition tools easily thanks to Cloud Speech-to-Text API and Dialogflow
  • Reduces time needed to set up additional servers during traffic peaks from two days to under 30 minutes
  • Leverages the global Google Cloud network to easily serve international clients

Supports processing of 1M minutes of speech every month

While playing pop music to customers on hold might make them less likely to hang up, businesses need to answer calls as efficiently as possible to keep customers happy. Communications software platform Voximplant helps businesses answer calls faster by providing the infrastructure, applications, and software development kits they need to implement cutting-edge technologies. Founded in 2013, it has grown significantly over the past five years and its international clients include Hyundai, Burger King, and Sberbank, one of the largest banks in Europe.

“If we want to stand out, we need to offer features powered by the best new technologies available on the market, such as natural language understanding and other AI models. It’s these innovative tools that enable our clients to transform the call center experience for their own customers.”

Alexey Aylarov, CEO and co-founder, Voximplant

Supporting voice, video, and messaging communication, the platform is fully programmable, and takes a serverless approach so businesses can focus on development. “Our most popular services are voice-based,” explains Alexey Aylarov, CEO and Co-founder of Voximplant. “Using Voximplant, you can build a whole range of services, from simple callback operations to automated phone surveys and interactive voice response functionality.” In 2018, more than 500 million successful calls passed through Voximplant, serving businesses based in Latin America, Asia, and Europe.

In order to offer cutting-edge technologies to its clients and sustain its growth across the globe, the company looked for a platform with excellent voice recognition capabilities that could also help it scale up. With Google Cloud, Voximplant has found the right fit for its needs.

“If we want to stand out, we need to offer features powered by the best new technologies available on the market, such as natural language understanding and other AI models,” says Alexey. “It’s these innovative tools that enable our clients to transform the call center experience for their own customers.”

Choosing the right speech recognition technology

While using automated technologies can help ease call center workloads, it’s important to also focus on the quality of the service to keep customer satisfaction high. “To build new services, we looked for the best possible product for speech recognition that would work in as many languages as possible,” explains Andrey Kovalenko, CTO and co-founder of Voximplant. The company also wanted to move some of its services from dedicated servers to a cloud infrastructure that would make it easier to serve businesses globally.

“The great thing about using Google Cloud SQL is that we don’t have to worry about backups, recovery, or maintaining high availability. It’s all built into the service, which really saves time for our developers and system administrators, so we can focus on improving our platform.”

Andrey Kovalenko, CTO and co-founder, Voximplant

It chose Cloud Speech-to-Text to provide speech recognition and transcription functionality in more than 120 languages, along with Dialogflow to build and deploy voice and text-based bots. “Our most popular languages are English and Russian, followed by Spanish, Portuguese, and Italian,” says Andrey. Using Dialogflow, clients can build voice bots on Voximplant that speak with customers using natural language, enabling them to handle inbound calls or launch outbound campaigns. In up to 60% of cases, automated agents can solve simple customer inquiries, meaning that live agents have more time to deal with complex scenarios.

As part of a multi-cloud strategy, Voximplant also created a low-maintenance, scalable architecture for its international workloads on Google Cloud. It built this infrastructure using virtual machines on Compute Engine, and using Cloud Storage and Cloud SQL for its data storage needs. It also uses App Engine to develop and deploy additional web services for its clients, and hosts one of its messaging services using Google Kubernetes Engine.

“The great thing about using Google Cloud SQL is we don’t have to worry about backups, recovery, or maintaining high availability,” says Andrey. “It’s all built into the service, which really saves time for our developers and system administrators, so we can focus on improving our platform.”

Becoming a Google Machine Learning Partner

Voximplant was an early adopter when Cloud Speech-to-Text was first launched in 2016. As part of the early access program, the Voximplant team was in touch with the product team at Google Cloud to provide feedback on the tool. “Because we’re always on the lookout for innovative products, we contacted Google Cloud straight away when Cloud Speech-to-Text API was launched,” says Alexey. “We’ve developed a great relationship with the product team.”

Then, in 2018, Voximplant became a Google Cloud Machine Learning Partner. As part of its software offering, it designed a connector that enables businesses to easily connect Voximplant’s telephone services with Dialogflow. Using the connector, clients can program Dialogflow to execute actions such as terminating a call. More complex call logic scenarios can be written in JavaScript using Voximplant’s platform.

“We’ve gained quite a few leads through the partnership,” Alexey shares. “We offer something extra for Google Cloud customers, and make it easier to build more complex services.”

“Voice recognition and conversational interface functionality is available out-of-the-box on Google Cloud, and this has been key. Without it, we wouldn’t have been able to launch so many services for our clients. One example is freeform speech recognition, which is really hard to achieve.”

Sergey Poroshin, Chief Development Officer and co-founder, Voximplant

Leveraging the Google Cloud network for growth

Using Cloud Speech-to-Text and Dialogflow, Voximplant processes more than 1 million minutes of speech every month, helping businesses build tools that are tailored to their needs, whether that be an automated phone survey or a robot for raffling tickets. “Voice recognition and conversational interface functionality is available out-of-the-box on Google Cloud, and this has been key,” says Sergey. “Without it, we wouldn’t have been able to launch so many services for our clients. One example is freeform speech recognition, which is really hard to achieve.”

Since migrating to Google Cloud, Voximplant can also easily scale to meet any peaks in traffic experienced by its customers, whether it’s an additional thousand or ten thousand additional simultaneous calls. “Before we migrated to Google Cloud, it took up to two days to install and configure a new server to cope with additional demand,” says Andrey. “With Google Cloud, we can do it in under half an hour, which really helps us handle our traffic.” It’s also able to serve businesses all over the globe, as it can locate its infrastructure in an international data center close to them, supporting compliance with regulations on data storage.

Voximplant is now looking to expand the volume of clients it serves, as well as extend its use of Google Cloud machine learning solutions. The company is currently exploring services such as sentiment analysis, offered on Dialogflow and powered by Cloud Natural Language.

“We currently handle 2 million calls a day, and our main focus is expanding to new regions,” says Alexey. “Having the global Google Cloud network of data centers available to us will make it much easier to serve clients in new locations, and we’re confident our Google Cloud infrastructure can easily scale to handle millions more calls.”

About Voximplant

Voximplant is a serverless communications platform that helps companies develop and implement real-time voice, video, and messaging solutions. Founded in 2013, Voximplant is now based in the US and handles hundreds of millions of calls for its customers worldwide.

Industries: Technology
Location: EMEA