What if you could speak and understand every language in the world? This convenient notion has been a flight of fancy for science fiction writers and fans for generations. Star Trek fans would recognize this technology as a universal translator (or “UT” for short), and we’ve seen it in Doctor Who’s Tardis Translation Circuit as well as in the form of a parasitic Babel Fish in The Hitchhiker’s Guide to the Galaxy.
But fortunately for any of us who are a bit squeamish at the thought of letting an alien fish swim around in our brains, real-time speech translation is growing more accessible with every advancement of AI-powered speech services and Baseet.ai’s easy-to-use AI app building interface.
AI Translates Speech in Real Time
Real-time speech translation has a wide range of useful applications. It can enable real-life conversations between businesses and foreign audiences and enable multinational participants to engage with each other at conferences. It makes international travel easier and more convenient for all of us, and this wide variety of uses has made real-time speech translation a hot topic in big tech.
Google offers end-to-end speech-to-speech translation through its Translotron application, which was released in 2019. At the time, Translotron was among the first models that used AI to directly translate speech from one language to another. Since then, there have been several new APIs built for this purpose, notably the technology behind Apple’s iTranslate Voice and Azure’s Speech Service.
Building Translation Apps With Baseet
Instant translation applications bridge communication divides between travellers and local communities and help people connect with each other from around the world. And with Baseet, adding translation services to your existing applications – or building your own easy-to-use AI app – is free and easy.
Baseet.ai provides pre-built nodes that help developers to build applications that utilize Azure’s Speech service without writing a lot of code. This cloud-based service application enables real-time translation of incoming speech from dozens of languages spoken around the world. With Baseet, developers can fork, tweak, and generate an API that can be copied-and-pasted into existing code. The API allows developers to add end-to-end speech translation to their existing applications, expanding their audience reach and making their services accessible to a global audience. And thanks to Baseet, this can all be accomplished with a simple low-code drag-and-drop with our unique approach to integrating customized, out-of-the-box solutions into existing applications. Baseet also gives developers a visual training app that they can use to train models on their own datasets – opening up a world of infinite possibilities!
How to Add Azure Speech Service to Your Applications
Baseet has pre-built nodes that can be linked together through our easy-to-use visual interface. A “node” is a reusable, customizable component that can be connected with other nodes in an easy drag-and-drop format to make increasingly complex and capable apps. An app is a series of nodes that can be linked together to make a functional application. If you’re looking to add Azure’s Speech Service to your existing code, simply take our auto-generated API and embed it into your app. With this approach, developers can build solutions around existing services with ease. Here’s how:
1. Setup API
To start with, you need an Azure subscription. To get a key for the speech service on Azure, just follow the steps mentioned in this link. Baseet offers this service as a node for you to build a customized application around and generate an API with the complete solution that can be applied across your enterprise platform.
- From the overview, you can select the region.
Don’t have an Azure subscription? Start with a free trial subscription key available here.
2. Access the Azure – Speech Translation Node
This node helps users to connect directly with speech translation service from Azure.
3. Select Inputs
Either use our drag-and-drop editor or code inputs yourself:
- audioPath: audio source which comes from Audio input node
- Service region: select from the menu the region that your service is registered on Azure service
- speechKey: credential key you got from Azure API for speech service
- Current language: select the current language of the current speech (audio)
- Target languages: select multiple target languages to translate the source speech
4. Generate Outputs
Connect Baseet’s pre-built nodes with your inputs, run the application, and see your outputs:
- speechText: the text of the audio source
- speechTranslation: dictionary of the target language as key associated with the translated text
The output is a URL for a PDF file contains the translation of the video speech with the languages you chose:
Check out Baseet’s Sample App
Baseet.ai provides sample apps for beginners to show how to use specific nodes. For Azure speech translation service, follow Azure – Speech Service APP to show the usage for each node.
Baseet’s unique user interface has eliminated the need for complex coding, which saves time and greatly reduces the likelihood of error in building sophisticated AI functions like real-time speech translation into your existing services. Try it for yourself by signing up for a free account at https://console.baseet.ai/signup or visit us on the web at https://baseet.ai/. Want to stay up-to-date on what our dynamic and user-friendly platform has to offer? Following us on social media at @BaseetAI on Twitter, Facebook, or LinkedIn.