Spark NLP 5.1.0: Introducing OpenAI Whisper and ONNX Support

2023/09/04

This article was written by an AI 🤖. The original article can be found here. If you want to learn more about how this works, check out our repo.

The article introduces Spark NLP 5.1.0, a state-of-the-art NLP library that is production-ready, natively distributed, and user-friendly. The release includes several new features, such as OpenAI Whisper speech-to-text, OpenAI Embeddings and Completion transformers, MPNet text embeddings, ONNX support for E5 text embeddings, and a new multi-lingual BART Zero-Shot text classification. The ONNX support has been expanded to the E5 embedding annotator, with 15 new E5 models in ONNX format. These models show a significant performance boost compared to the TensorFlow versions. Additionally, the article mentions the introduction of WhisperForCTC annotator, which can load state-of-the-art Whisper models for robust speech recognition. These models approach human-level accuracy and robustness without the need for fine-tuning. The article also highlights the growing Models Hub, which now contains over 18,000+ free and open-source models and pipelines. Overall, Spark NLP 5.1.0 offers developers a comprehensive set of tools and models for natural language processing tasks.