News 20th Jun 2019
Cloud computing image for new SaaS blog

Speechmatics has a new SaaS offering for cloud-based speech transcription which is hosted in Azure Kubernetes Service (AKS). The new service delivers extensive scalability and elasticity to support dynamic load whilst also providing feature parity across all of Speechmatics’ deployment options.

The new SaaS is intended for business partners on service contracts, focusing on a single API endpoint. The SaaS API simplifies our customers’ workflows and integrations with other systems, whether hosted in public clouds or private data centres, providing the latest JSON and plain TXT transcription output. To ensure the privacy of data for our customers, all jobs put through the updated SaaS are deleted 7 days after completion, however, customers can manually delete a job before this time if desired.*

Why did we choose Azure Kubernetes Service?

According to a Cisco survey, 73% of organisations say nearly all of their apps will be SaaS by 2020 and at Speechmatics, we want to offer our customers the best hosting services to meet their business needs.

Why Kubernetes?

Kubernetes was selected to power Speechmatics’ new SaaS as it offers many new capabilities for our customers. It has the ability to power those who have either outgrown their Virtual Appliances and wish to deploy in their own data centres and virtual private cloud environments, or others who wish to integrate Speechmatics’ automatic speech recognition (ASR) into their existing cloud strategy.

Kubernetes is the most effective platform for writing a cloud microservice architecture application which is not tied to any specific cloud provider (public or private). This microservices architecture is shared with Speechmatics’ Batch Virtual Appliance, so we’re able to build the product using the same architecture and APIs. It can also be deployed on different platforms depending on our customers’ scalability requirements.

Due to its wide adoption, Kubernetes has a growing community of companies and open source projects that help to extend its capabilities. Being part of this emerging community means that Speechmatics can evolve its product offering quicker than ever before and with less complexity. Kubernetes helps us to provide a cloud service that is built on modern principles with an elastically scalable architecture.

Speechmatics’ SaaS

Speechmatics always wants to provide the best possible transcription service to our customers. The new SaaS means that customers no longer have to worry about operational considerations such as capacity planning and maintaining service availability. Speechmatics removes the complexity and cost of deploying hardware in customers’ own or third-party infrastructures, by providing all of the maintenance, support and troubleshooting. The availability of Speechmatics’ SaaS and internal support accelerate customers’ time to market and removes the pain of having to manage their own deployment. Our dedicated Support team is always available to ensure our customers are successful.

Speechmatics’ cutting-edge deep neural network architecture means our transcripts have world-leading accuracy across all platforms. As well as unrivalled accuracy, Speechmatics’ SaaS has complete feature parity with our Batch Appliance and Batch Containers. This is possible as AKS allows the same core speech transcription component to be used in the cloud service as in our Virtual Appliances. This means all customers will now have immediate access to new features that are developed across each of our deployment options, including Advanced Punctuation, Custom Dictionary and Channel Diarisation. Customers will also have instant access to new language and release updates.

Why the cloud?

Businesses are shifting their operations to the cloud for a number of reasons, including:

  • To cut operational costs
  • To maintain a consistent working process
  • To collaborate in large and decentralised teams
  • To save the time needed for setup and training
  • To facilitate access to data
  • It is mobile-optimised and independent from the devices they’re using
  • To make it easier to connect applications with each other
  • To reduce the burden of upgrading, maintaining, and safeguarding data
  • To keep them in line with trends

Interested in speech recognition?

If you like the sound of Speechmatics’ new SaaS, get in touch – we’d love to hear about your unique use case and how we can help with the adoption of ASR into your business. 


*Where permitted by the customer contract, media files and transcription results may be retained by Speechmatics for service improvements.