Amazon Polly has introduced a generative engine with three new voices: Ruth and Matthew (American English), and Amy (British English). Trained with diverse data, the engine offers high precision. Amazon Polly, a text-to-speech service, now includes high-quality, human-like voices in various languages. Users can choose from different voice options, adjust speech parameters, and deliver realistic voices quickly. The new generative voices are available in the US East (N. Virginia) Region.
Amazon's new Titan Text Embeddings V2, now available in Amazon Bedrock, is optimized for Retrieval-Augmented Generation (RAG) and pre-trained on 100+ languages and code. The model allows output vector size selection, balancing detail and computational time. It also introduces improved unit vector normalization for better vector similarity accuracy. It can be used for semantic searches, label classification, and improving search result relevance.
Amazon's Titan Image Generator, a generative AI model, is now available in Amazon Bedrock. It creates realistic images using English prompts. All images have an invisible watermark to identify AI-generated content. Watermark detection is available in the Amazon Bedrock console and a new DetectGeneratedContent API. The model is also accessible in PartyRock, an Amazon Bedrock playground.
Meta's Llama 3 models, designed for building and scaling generative AI applications, are now available in Amazon Bedrock. The models, trained on a dataset seven times larger than Llama 2 models, come in 8B and 70B parameter sizes. Two new Llama 3 models increase model choice within Amazon Bedrock, each suitable for different use cases. Meta is also developing additional Llama 3 models over 400B parameters in size, offering new capabilities.
Upload all Meta ad types in bulk directly from Google Sheets, single image, video, carousel, and flexible ads. Control placements, multiple headlines, primary texts, descriptions, and creatives from one spreadsheet. Built for agencies and teams managing dozens of ads across multiple accounts, helping you launch faster, stay consistent, and avoid costly manual errors.
