April 2022 Summaries

Named Entity Recognition (NER), also known as Entity Detection, is a valuable tool in the Natural Language Processing (NLP) field. It identifies and categorizes key information such as names of people, organizations, addresses, phone numbers, social security numbers, locations, and more within a text. NER typically involves two steps: entity identification and entity classification. Entity Detection works by detecting various entities in the transcription text and then returning them in a JSON response when fetching a completed transcription from an API. Some currently supported entity types include person names, phone numbers, and occupations. There are several Entity Detection APIs available on the market today that support Named Entity Recognition. These include AssemblyAI, Dandelion, Google Natural Language, Azure Cognitive Services, TextRazor, and Allganize. Each of these platforms offers different features and pricing plans to cater to a wide range of industries and use cases. Using entity detection can provide invaluable customer or employee information across various industries such as telephony, CRM platforms, hiring platforms, virtual meeting platforms, voice bots, and medical sectors. By collecting this entity information, companies can perform analytics for better understanding of customers, adjust marketing campaigns, modify products, and much more.

In the AssemblyAI project, students at ASU HACKML 2022 utilized the Core Transcription API to create real-time speech-to-image generation. They reproduced elements of DALL-E 2's zero-shot capabilities with a simpler model. The project integrates Machine Learning models and web interface framework with AssemblyAI API, enabling corrective language modeling. The inspiration came from Open AI's paper on Zero-Shot Text-to-Image Generation. The build consists of real-time audio transcription using the AssemblyAI API, HTML/CSS for client-side interface, Node.js and Express for server hosting, pretrained models running in parallel with client and server, and Selenium to pass messages between components. The main takeaways include impressive improvements in audio transcription tools, the potential of less data for similar results as larger models, and the impact of changing pretraining paradigms. Future directions involve incorporating knowledge graphs for semantic correctness checks, associating natural language with image sequences, and generating videos from resultant image vectors.

This article provides a step-by-step guide on how to create a JavaScript Audio Transcript app using the AssemblyAI Speech-to-Text API. It covers setting up an AssemblyAI account, installing Axios and sending POST requests with it. Additionally, the article demonstrates automating the process of fetching audio transcripts by checking their completion status using setInterval() function.

To improve customer engagement through Conversation Intelligence Platforms powered by AI technology, companies should follow these three steps: 1) Automate Transcription; 2) Make Voice Conversations Searchable and Indexable; and 3) Gain Actionable Insights. By incorporating key technologies like Speech-to-Text transcription and Audio Intelligence into their platforms, businesses can gain valuable insights from customer interactions that drive process and behavior changes, leading to increased customer engagement and faster ROI.

MediaPipe is Google's open-source framework for customizable multi-modal processing of data streams (e.g., video, audio, depth). It provides easy deployment on mobile and edge devices, with capabilities like pose estimation, face detection, tracking and augmented reality, among others. While it does not provide any new models per se, it simplifies the process of using state-of-the-art models in real time applications by providing a unified interface for various tasks.

In this tutorial, we created a simple JavaScript application that utilizes the Web Speech API to convert text to speech. Here is a summary of what we accomplished: 1. We explained how to use the Web Speech API for Text-to-Speech in JavaScript and introduced the required objects and methods. 2. We wrote static code to test if our browser supports the Web Speech API and made it speak a predefined sentence. 3. We made our application dynamic by adding an HTML form and allowing users to input their text that will be converted into speech. 4. Additional features of the Web Speech API, such as language, pitch, rate, voice, volume were briefly mentioned for further exploration. 5. The final code example demonstrates how to use the Text-to-Speech feature in a simple yet functional web application. Note: This application is best tested on a device that supports speech synthesis and has built-in speakers.