February 2022 Summaries

Webhooks are a method of server-to-server communication that allow for real-time updates and event triggers between applications. Unlike traditional HTTP requests, webhooks involve a remote server sending an HTTP POST request to a public URL on another application's server when an event occurs. This enables the receiving application to trigger events or perform actions based on the received data. Webhooks are commonly used in scenarios such as payment processing and transcription services. They offer advantages over polling, which involves continuous requests from one server to check for updates on another server. However, webhooks require support from third-party services to function effectively.

Deepgram has launched a Diversity, Equity, and Inclusion (DE&I) program as part of their commitment to creating an inclusive work environment and fostering ongoing conversations about these issues. The company's new statement of values emphasizes the importance of valuing every voice and promoting diversity in background, thought process, and experience. Deepgram is particularly focused on supporting its Black team members and recognizing their contributions to the company's success. This initiative coincides with Black History Month, a time to reflect on the struggles and achievements of Black Americans throughout history. The company acknowledges that there is still much work to be done in addressing systemic racism and other forms of inequality, but remains committed to fostering an inclusive culture where every voice is heard and understood.

In the article "How to Build a Podcast Player with Transcriptions using Vue and Supabase" by Brian Barrow, he guides readers through creating a Vue application that enables users to log in, subscribe to podcast feeds, and transcribe any episode using Deepgram. The author has previously written about getting started with Supabase's database and implementing authentication into Vue 3 apps. The final code is available on GitHub for those who want to dive right in. For any questions or feedback, readers can reach out to the author via Twitter @DeepgramAI or through GitHub discussions.

In the final post of her 'Diving Into Vue 3' series, Sandra Rodgers demonstrates how to use the Composition API to write reusable code. She builds a single-file component with an image on the left and text on the right that changes size based on the image being resized. The post covers topics such as working with the DOM, using lifecycle methods, template refs, and creating composables for reusability. Rodgers also provides a link to her GitHub repository containing various branches showcasing the progression of the project's refactoring process.

The text discusses a partnership between Daily and Deepgram that enables developers to add accurate live transcriptions to video calls within their applications using just a few lines of code. A React.js demo and starter project are provided by the Daily team, while a video tutorial demonstrates how to set up live transcription without a framework in under 40 lines of code. The tutorial covers setting up a Daily video call, enabling live transcriptions for a Daily domain, and starting transcribing calls. Deepgram can be reached on Twitter at @DeepgramAI, and feedback is welcome via GitHub discussions.

Filip Grebowski, a member of the Deepgram community, has created an innovative demo that enables users to build websites by dictating elements and their attributes using voice commands. The project is divided into two parts. In Part 1, Filip expands on the "Get Live Speech Transcriptions In Your Browser" demo to detect keywords and construct a webpage. In Part 2, additional functionality is introduced to incorporate layout and styling for the webpage. Deepgram encourages feedback on this post or any other related topics through their GitHub discussions.

In the fourth part of her "Diving Into Vue 3" series, Sandra Rodgers discusses how to use Reactivity APIs such as ref and reactive in Vue.js. She explains that reactivity refers to a feature where changes to state automatically update what is displayed on screen. The new Reactivity API in Vue 3 uses the JavaScript Proxy API instead of Object.defineProperty, which was used in Vue 2. Rodgers also introduces related helpers such as toRef, toRefs, and isRef. She emphasizes that understanding both ref and reactive is crucial for using Vue 3 effectively.

End-to-End Deep Learning (E2EDL) Speech Recognition software is ready and in production now, with customers running millions of hours of new audio transcribed per month. E2EDL has moved from research into stable production and shown the world that E2EDL is not just a pipe dream but a reality. Deep learning approaches to ASR are more accurate than traditional approaches. E2EDL provides lower costs, higher accuracy, faster speed, easier scaling, and future-ready technology for businesses.

In this tutorial, the author introduces methods, watch, and computed properties in Vue 3. They demonstrate how to use these features with a real-world example of tracking employees and managers in a company. The author also provides a comparison between watch and the new watchEffect feature in Vue 3. Additionally, they explain how to write methods inside the setup function and how to create composables for organizing component logic.

This article provides a step-by-step guide on adding Supabase authentication to a Vue 3 application. It covers setting up Supabase, creating components for registration and login, gating content behind authentication, logging out, and completing the flow with a 'magic link' via email. The author encourages readers to reach out on Twitter or provide feedback through GitHub discussions if they have any questions or comments about the post.

Classroom Captioner is a project designed to enhance accessibility in educational environments by providing live captions for lectures using Deepgram's speech-to-text capabilities. It allows lecturers to easily set up a captioning system by opening a web application, creating a room, and validating themselves with a lecture key. Students can join the room using a code to receive real-time transcriptions of the lecture. The application utilizes Node.js, express, and socket.io for server management and real-time communication, while Deepgram API keys facilitate transcription. The lecturer's microphone captures audio, which is transcribed and broadcast to participants in specified rooms. The project aims to create an inclusive classroom experience by offering a text representation of spoken content, thereby supporting students who need or prefer written transcripts. The full implementation and deployment instructions are available on GitHub, enabling others to replicate and customize the setup for their own use.

Voice technology is transforming the landscape of customer experience by enabling businesses to better understand customer needs and sentiment through analyzing conversations. Four key ways that voice tech is improving customer experience include turning audio into text, identifying appropriate actions for improved customer experience, supporting employees in understanding customers, and creating a searchable database of conversations for insights extraction.

Bekah Hawrot Weigel has joined the Developer Relations team at Deepgram as a Technical Community Builder. With her background in education, technical training, and community organizing, she aims to create spaces where people feel comfortable sharing, learning, and mentoring together. Her personal experience with trauma led her into tech, and she founded Virtual Coffee, an organization that supports developers during the pandemic. As part of Deepgram's Developer Relations team, Bekah continues to focus on building community and fostering growth within the tech industry.

The article discusses the implementation of a collapsible navigation system using Nuxt Content for a documentation page at Deepgram. It highlights the use of recursive components and an array to manage the state of expanded sub-menus. Additionally, it introduces a custom transition component to ensure smooth expand/collapse animations. The author provides a working example in a code sandbox and encourages feedback on their work.

Deepgram has achieved Service Organization Control (SOC) 2 Type II certification from Cyberguard Compliance, confirming its security, availability, and confidentiality measures meet industry standards as defined by the American Institute of Certified Public Accountants (AICPA). The six-month audit found no exceptions to SOC 2 Type II requirements. Deepgram previously received SOC 2 Type I certification. The company's systems are using best practices for security, confidentiality, and privacy, ensuring customer data is protected.

In this article, the author discusses the new setup function introduced in Vue 3 and how it differs from the Options API used in Vue 2. The setup function allows for organizing code by logical concern rather than separating script into sections by options. This leads to a new pattern called Vue 'composables', which involves grouping code by logical concern and making it reusable. The author provides an example of how to use the setup function, highlighting important facts about its usage such as timing, lack of keyword this, requirement for returning an object, and reactivity with ref or reactive functions. Props and context are also discussed as arguments that can be passed to the setup function.

Tonya Sims, a former pharmaceutical sales representative and professional basketball player, has joined the Developer Relations Team at Deepgram as a Python Developer Advocate. Her passion for technology began when she was eight years old after her parents bought a computer. After taking a break from coding to play basketball professionally, Tonya rediscovered her love for programming and eventually got hooked on Python. She enjoys listening to music, playing the piano, reading, and spending time with her nieces and nephews. Connect with Tonya on Twitter or provide feedback about Deepgram through their GitHub discussions.

Deep Learning is considered the best approach for speech recognition due to its ability to process speech in an end-to-end manner, unlike traditional methods that rely on separate acoustic, pronunciation, and language models. Traditional methods are highly lossy to a computer as they involve making simplifying assumptions at each step of the process, resulting in performance limitations and suboptimal results. Deep learning, on the other hand, allows for simultaneous optimization across the entire problem domain, leading to more accurate speech recognition systems like Alexa and Siri.