January 2022 Summaries

In this tutorial, learn how to set up serverless functions with Netlify and use Deepgram's Node.js SDK for transcriptions of hosted files and local files. The process involves installing necessary software, setting up a Netlify project, creating serverless functions, adding your Deepgram API key as an environment variable, testing the functions locally, and deploying them to production. This approach is suitable for scale, but not recommended for live transcription due to its nature of spinning up and down quickly.

This article delves into the differences between Vue 2 and Vue 3, specifically focusing on project setup using the Vue CLI. The author begins by explaining how to create a new Vue project in both versions of Vue. They then examine three key files: index.html, main.js, and App.vue. The index.html file remains unchanged between Vue 2 and Vue 3, serving as the default starting page for web applications. The main.js file, however, undergoes a significant transformation in Vue 3. Instead of using new Vue() to create a Vue instance, Vue 3 introduces the createApp function, which allows for separate configurations and instances of Vue. In the App.vue file, the most notable change is that Vue 3 no longer requires a single root element in the template block. This means developers can now have multiple direct child elements without needing an extra div. The author concludes by stating they are not yet intimidated by these changes and plans to explore more differences between Vue 2 and Vue 3, such as the setup() function and Composition API versus Options API.

In this project, the author demonstrates how to add translations to transcription projects using Deepgram and iTranslate APIs. The required packages are installed, and environment variables for Deepgram and iTranslate API keys are set up. An introduction to iTranslate's text translation capabilities is provided, along with a sample API request and response. A function to translate text between two languages is created using the iTranslate API. The author then explains how to use the translation function for pre-recorded transcript translation by first obtaining a transcript from Deepgram and then translating it using the function. Live transcript translation is also demonstrated, utilizing Deepgram's live transcription feature and the translate function to display translated text in real-time. In summary, pairing iTranslate with Deepgram's speech recognition API allows for fast and accurate translation services, making them a suitable combination for transcription projects.

The State of Voice Technology report reveals that companies are shifting their focus from using voice technology as a cost-saving tool to a revenue-generating strategy. Customer experience is the top motivator for speech technology adoption, with 73% of respondents noting it as the most impactful use for speech technology. Companies see significant productivity and revenue gains through speech technology, with 83% reporting 25-75% productivity gains and 59% seeing revenue gains of 25-75%. Concerns around bias in speech technology are top of mind, particularly regarding gender, race, and non-native speaker biases. The majority of respondents believe widespread adoption of the technology will take place within the next one to five years.

In a recent video, Kevin Lewis demonstrates how he created a wearable live transcription badge using a Raspberry Pi and Deepgram technology. The device captions the user's voice to help others understand them while wearing masks. The project gained significant popularity on social media due to its practical solution for a common problem. Built with JavaScript, Vue.js, Node.js, Deepgram, and iTranslate, the badge is hosted for free on Glitch. Viewers interested in creating their own can follow along with the video tutorial and access important links provided by Kevin.

In this guide, Kevin Lewis demonstrates how to automate the startup process of a web application running on Raspberry Pi OS using Kiosk Mode. By editing the autostart file and adding specific commands, he ensures that Chromium opens in fullscreen mode at a specified URL upon device launch. This eliminates the need for manual touchscreen interactions and effectively locks the user into the browser. The guide is aimed at helping developers build web-based Raspberry Pi projects more efficiently.

Companies are increasingly switching from Amazon and Google's speech recognition systems to Deepgram due to its superior speed, accuracy, model tailoring capabilities, implementation support, and cost savings. Deepgram's end-to-end, hardware-accelerated deep learning ASR system returns results up to 60x faster than Big Tech competitors while processing multiple audio streams. Its out-of-the-box accuracy consistently outperforms Amazon, Google, and other providers, with the option for users to quickly and easily train tailored models that understand specific jargon or languages. Deepgram also offers a full-fledged set of developer tools, a developer community, and live expert human support to help users optimize their voice technology implementations. Additionally, Deepgram charges customers only for the length of their audio, eliminating unnecessary fees associated with Big Tech's 15-second increment pricing model.

This article discusses three approaches to protect your API key while using live transcription services from a web browser with Deepgram. The first approach involves creating and deleting keys on-demand, the second approach uses automatically-expiring keys, and the third approach sets up a server proxy to communicate with Deepgram instead of sending keys directly to the client. These methods help ensure that your Deepgram account remains secure while using live transcription services in web applications.

This tutorial demonstrates how to create an executable script using Node Package Manager (npm) and its utility npx. The process involves creating a file specifically for the logic, specifying the executable file in package.json, starting the file with a shebang, and ensuring the code will run whenever the file is executed. Additionally, handling arguments and executing logic from the main package are discussed. Finally, validating command inputs and publishing and using the package are covered.

In this tutorial, a live transcription iOS app is built using Deepgram's Speech Recognition API and the WebSocket library Starscream. The final project code can be found at https://github.com/deepgram-devs/deepgram-live-transcripts-ios. To create the app, users need a Deepgram API Key and Xcode installed on their machine. The app uses AVAudioEngine to access audio data from the microphone and sends it to Deepgram for real-time transcription. The resulting transcriptions are displayed in a UITextView within the iOS app.