Facilitating the diffusion of knowledge and innovation in professional software development
Domain-specific languages allow domain experts to participate in the software development process. Few DSLs stand the test of time. Important success factors for long-term DSLs seem to be the user-centered design and compliance with the open-closed principle. Markdown, TeX and CSS have been popular and relevant for two decades, even if their original target audience has evolved.
We prepared this eMag for you with content created by professional software developers who have been working with microservices for some time . If you are considering migrating to a microservices approach, take notes on the teachings, mistakes, and recommendations from these experts.
In this article, author Yang Li explains the importance of the pre-computation technique in databases, OLAP, and data cubes as well some of the trends in using pre-computation in big data analysis.
Holly Cummins discusses some of the tradeoffs affecting climate change and provides a roadmap to find the right thing.
Laura Bell examines how biases affect the safety of a development lifecycle and examines three common biases that lead to major problems in this area.
Microsoft announces limited access to its text-to-speech neural AI
Recently Microsoft announced limited access to its text-to-speech neural AI called Custom Neural Voice. This service allows developers to create custom synthetic voices.
Custom neural voice is a text-to-speech (TTS) feature of Speech in Azure Cognitive Services that allows users to create a unique customized synthetic voice for their brand . Since the preview in September last year, the feature has helped several customers such as AT&T, Duolingo, Progressive and Swisscom to develop branded language solutions for their customers. The function is generally available (GA). However, customer access to Custom Neural Voice includes technical controls to prevent misuse of the service. You need to apply for it.
Microsoft’s underlying neural TTS technology for Custom Neural Voice consists of three main components: Text Analyzer, Neural Acoustic Model, and Neural Vocoder. The first component, Text Analyzer, is responsible for generating natural, synthetic speech from text. The text is first entered into the Text Analyzer, which provides output in the form of a phoneme (a unit of keynote that distinguishes one word in one language from another). Next, the phoneme sequence defines the pronunciations of the words contained in the text that go into the neural acoustic model to predict acoustic features that define speech signals, e.g. B. timbre, speaking style, speed, intonations and stress patterns. Finally, the Neural Vocoder converts the acoustic features into audible waves to generate synthetic speech.
TTS neural language models are trained using deep neural networks based on real-world speech recordings. With the customization function of Custom Neural Voice, customers can adapt the Neural TTS engine to their user scenarios. To use custom neural language, customers need an Azure account and subscription. Once approved to use the feature, they can then start a custom language project, upload data, train, test, and deploy the language model.
There are several use cases where customers can benefit from the custom neural voice, such as: B. Customer service chatbots, voice assistants, online learning, audiobooks, public service announcements and real-time translations. A previous user, Swiss.com, wanted to create a more engaging customer experience by building a voice assistant that clearly represented their brand. In a message from Microsoft Switzerland, the author wrote:
With the voice service, Swisscom has given its customers access to an intelligent, multilingual voice assistant that helps improve the customer experience and accelerate their own digital transformation.
Qinying Liao, chief program manager at Microsoft, described the benefits of using Custom Neural Voice in an Azure AI blog post:
This technology enables users to create highly realistic voices with Custom Neural Voice with just a few training audio. With this new technology, companies can spend a tenth of the effort traditionally required for the preparation of training data and at the same time significantly increase the naturalness of the synthetic speech output compared to conventional training methods.
To make computers more human, speech is a crucial component, and In 2020, companies will have to deviate from the robot-controlled and standardized voices that have accentuated synthetic speech in the past. The cloud enables this level of personalized creation of a personalized language experience – with availability, low-cost computing and operating capacity. It is therefore a widespread use case for all IaaS / PaaS players – and suitable for companies and their customers as well as employees, as they have a more human experience.
In addition to the ability to customize TTS language models, Microsoft offers over 200 neural and standard voices in 54 languages and locales.
Every Tuesday a summary of the InfoQ content from the previous week is sent out. Join a community of over 250,000 experienced developers.
Look at an example
You must register an InfoQ account or sign in or sign in to post comments. But there is much more to being registered.
Every Tuesday a summary of the InfoQ content from the last week is sent out. Join a community of over 250,000 experienced developers.
Look at an example
Learn about the trends, best practices, and solutions from the world’s most innovative software experts to validate your software roadmap. Join us for an online experience for senior software developers and architects every 2 weeks.