Microsoft has a lot of different speech services. There's Bing Speech, Custom Speech, Speech Translation, Office Dictation, and more. Utlilizing them all is a somewhat convoluted process, and it's one Microsoft is planning to fix.
The announcement came in a quiet moment on the last day of Build 2018 during a Cognitive Services Speech SDK session. The idea is to have the common speech API and SDK support all modern programming languages and run on all platforms. It will be positioned under the Microsoft Cognitive Services offering and will be named the Microsoft Speech Service.
The API and SDK will provide easier access to speech-to-text, speech translation, speech-to-intent, and custom keyword spotting. Support will launch for all 28 languages, to new and veteran developers alike.
Now in Preview
Microsoft has already launched a preview of the service with limited support. It's far from the catch-all that the company is planning, but it's a good foundation.
“This is a good first step,” said Rob Chambers during the Build 2018 session. “The preview supports Windows 10, Linux and Android, and works with C#, C++ and Java currently. Support for iOS and macOS X are coming soon.”
Some support comes from Speech Devices SDK, which Microsoft also announced at Build. Among other things, it allows for noise cancellation, far-field microphone capabilities, and echo cancellation. The idea is to prompt companies to create services like automatic drive-thru machines with relative ease.
It will also help propel Microsoft's AI for Acessibility program, which has pledged $25 million to developers who create disability-focused tools. It's not yet clear how the new service will gel with the legacy speech tecnology in Windows 10, but we expect more information soon.