Azure Cognitive Services capabilities are now in preview
Published date: November 04, 2019
The following new Azure Cognitive Services capabilities are now available in preview
Custom Neural Voice capability
The Custom Neural Voice capability enables you to build human-like, personalized voices using deep neural networks and your own training audio. It’s also possible to build long-form audio content with highly human-like digital voices.
Custom Speech with Office 365 data
Custom Speech allows you to optimize speech recognition with tailored models. Now, organizations that use Office 365 can optimize speech recognition accuracy for organization-specific terms by automatically generating custom models using their Office 365 data.
Text Analytics—Sentiment Analysis 3.0
Text Analytics, an Azure Cognitive Service, is releasing the latest version of its Sentiment Analysis capability. Languages have been expanded to include English, French, Italian, German, Spanish Korean, Chinese, and Japanese.
The new model has the following improvements:
- Improved accuracy— Significant improvement in detecting positive, neutral, negative, and mixed sentiment in text documents over previous versions.
- Document and sentence-level sentiment score—Detect the sentiment of both a document and its individual sentences. If the document includes multiple sentences, each sentence is also assigned a sentiment score.
- Sentiment category and score—The API now returns sentiment categories (positive, negative, neutral, and mixed) for text, in addition to a sentiment score.
- Improved output—Sentiment analysis now returns information for both an entire text document and its individual sentences.
Text Analytics—Expanded Entity Type Support in Entities 3.0 endpoint
Named entity recognition (NER) is the ability to take unstructured natural language text and identify the occurrences of entities such as people, locations, and organizations. The new version of NER will support more than 100 named entity types in 5 languages and more than 20 named entity types in 16 other languages.
Text Analytics—Personally Identifiable Information support in Entities 3.0 endpoint
We are also excited to announce the public preview of personally identifiable information (PII) recognition in Text Analytics’ Named Entity Recognition. With this addition, users will now be able to detect and extract more than 80 kinds of personally identifiable information in English language text documents.