OpenAI Whisper, GPT3, Codex & DALL-E 2 Hackathon Summary

Use AI to build powerful presentations. A good presentation is critically important, because it will form the impression of your product for your audience Making an impact is done by creating a sensory experiences Presentations with videos and accompanying music will leave your audience mesmorised. Using AI to assist in efficiently produced maximum impact presentations Making good presentations takes time Regardless of the context time is always precious Presenting ideas is always necessary from starting small projects to guiding important decisions Weather Vane will save vital time for professionals, students and collaborators everywhere Time saved is value added

Weather Vane

Our idea was about using GPT-3 and DALLE together by combining results from one neural network into prompt for the other one. After some brainstorming we came up with what we call project Infinite Gallery. It allows anyone to stroll through infinite amount of art pieces on the topic that they like. GPT-3 generates paintings descriptions and DALLE generate corresponding picture. Initially we were planning to make the gallery literally infinite by adding new pieces as the user moves but because time limitations we stopped at one "corridor of a museum".

Three dimensional beings

While applying for jobs one needs to create a perfect on-page resume with relevant skills to that specific job. While often individuals (like the ones participating here) have multiple experience. Filtering and creating a new resume for each job application can tedious and time consuming. Resume AI solves this in 3 simple steps 1. Upload your master resume. 2. Paste the link to Job application. 3. Generate the perfect resume in less then 10 seconds.

Resume AI

So i had to learn how to use it first so i followed a tutorial, but i couldnt fully learn how to do what i was trying to just yet. I used a youtube tutorial and the template that yall have. I'll make the brainstorming assistant over time

Beginners Luck

Voice to Entertainment - Music Objective: To provide music based on voice command. Functionalities: User goes to my website, clicks on a mic button and insructs what kind of music they want. Output is provided in mp3 form which can be listened to for enjoyment and and downloaded for use. Thanks: To the several Python APIs that I've leveraged for this, and equally important lablabai's much friendly staff and the developer tutorials. Concept, Programming and Integration: Muthukumaran Azhagesan, [email protected] (http://www.autoshields.website)

Voice to Entertainment

Why only English Speaking Guys have been all the fun ? The revolution of AI is something which every human should have access to, and hence we have built something like that. It's a Voice to image generator that allows the user to give an audio input in their native language and generate an amazing image. Tech Stack - React on Frontend and Flask on Backend. We have used APIs from Whisper, Dalle and GPT-3 making the best case use of every of every technology at its best.

renesis

As a HR agency or recruitment professional, you know how time-consuming the recruitment process can be. ScreenAIr is here to help! This innovative tool can save up to 60% of the time typically spent on the recruitment process. With its advanced GPT-3 powered algorithms, ScreenAIr can quickly filter through resumes to find the most qualified applicants. By automating the initial screening process, ScreenAIr can significantly reduce the amount of time spent searching for new hires.

VulcAInts

Newspapers are outdated business model.NiFTy News (NN) gameficies the newspaper, adds freshness and interest . NN reads news through API,getting difference from last run, generates imgs and then mints NFTs

LocaLgHosT

Distiller condenses information shared during meetings into bit-sized summaries and provides inspirations and actionable plans to drive projects forward productively. It transcribes long discussions into searchable transcript, summarizes content into easily consumable forms, provide action items and follow-up questions to push the project forward, and generate metaphors and images to promote more brainstorming.

Headjackers

Built with GPT-3, React, and Flask. As the job search becomes increasingly competitive and top companies are increasingly strict in their employee sourcing criteria, acing the behavioural interview is an often overlooked component of successfully rounding out your application. To provide a readily-available, continuously-improving, and convenient solution to preparing for behavioural interviews, we developed Interview.Me. With Interview.Me, users can generate behavioural interview questions pertaining to their companies and positions of interest with the click of a button. Users can simulate a real interview experience using the audio input feature, in addition to receiving feedback based on the questions and answers provided. Interview.Me makes the behavioural interview preparation more convenient than ever, so applicants can feel confident they're making the best impression.

BobaTalks

Some videos are too long but contain key information. There is no points in absorbing the entirety of them when you are looking for information. Our project takes in a Youtube URL, extracts the audio from the video, and generates a transcription. With the transcription, we display a results page that all summary, key points with clickable timestamps that take you to their place in an embedded youtube player of the video the user submitted, and shows the total transcript at the bottom as well.

The Neural Networkers

Our solution is a web app that generates PowerPoint presentations giving either a prompt about what the slides should cover or a summary of a specific topic. The purpose of this app is to make presentations using the power of AI quickly. Technology-wise, for the frontend client, we utilized Next.js, and for the backend server, we utilized Python’s Flask. For the AI handling, we utilized ChatGPT3 to generate slide text and titles. Additionally, we used Dall E 2 to create each image on the individual slides. Lastly, we used Vercel to host the front end and Heroku to host the backend.

buildrs

Learn from videos with AI! Check out the live demo: https://kiwi.video

kiwi.video

There is a market niche for interior design services for empty rooms, as many people struggle to visualize multiple design options without physically placing furniture and decor in the space. Using AI and stable diffusion techniques, it is possible to create multiple design alternatives that can help clients better understand the potential of their empty room. This can save time and resources, as clients will not have to physically set up and take down multiple design configurations. By offering this service, interior design companies can meet a unique and unmet need in the market.

AI Mavericks

To ignite creativity and learning with engaging experiences in the classroom, we designed an AI solution to creating content, activities, and ideas for dynamic lesson planning. A well-designed lesson plan helps students and teachers understand the goals of an instructional module. This allows the teacher to translate the curriculum into learning activities. Though lesson planning has its benefits, it is time-consuming. Our AI answer is to generate the content needed for a lesson plan without removing the flexibility for different teaching styles. Thank you for your time!

LA Hackers

We built a Interactive Children Story Book Generator using Whisper, Dall-E 2 and GPT3.

TaleTellers

Let your creative ideas grow with the smart AR app that helps you construct spatial mind maps. A system that inspires and fosters brainstorming. Through simple and intelligent text-based interactions, users’ input and choice will SPROUT, ROOT and BLOOM the seeds of their imagination. The 3 main STEMS serve as smart tools to expand, extend and envision your ideas. SPROUT will extract keywords based on the user’s text input, and list the relevant terms and concepts to expand on user’s ideation. ROOT will extend on the selected keyword and provide additional information. BLOOM will emerge imageries generated based on your idea. Creative Construct is an engine driven by AI (OpenAI’s GPT-3 and DALL·E 2), empowering the flow and growth of creative thinking and brainstorming, allowing the user to puzzle with new inspirations or unfamiliar ideas in any physical setting through the lenses of AR, for the user to cumulatively construct a spatial mind map natural to each user’s creative mind.

Textual Architecture

An AI powered food buddy which takes in your food preferences and creates a bespoke recipe for you in real time and not just that, it creates some delectable looking images of how your final dish could look. The preferences which it currently handles - list of ingredients, cuisine, flavor profile, allergies, time of meal, dietary restrictions, calories, preparation time

Pytorchbearers

The application is built on React Native + Python. It takes raw audio as an input and performs speaker diarization using pyannote.audio. Then by using Whisper it creates a transcription of the call. The transcribed text is summarized by GPT3 and analyzed by a blacklist algorithm that uses a list of words associated with popular scams. To improve algorithm performance we experimented with GPT, but only 3.5 version(chatGPT) was improving analysis quality. Since only GPT3 is available through API, we decided to wait with adding GPT to the algorithm. Summarized text with a calculated probability of scams is being sent to user's relative.

Young Bulls

An OpenAI powered tool that helps users generate interior designs effortlessly, with zero knowledge of prompt engineering. It eliminates the friction between discovery, consideration and purchase of inferior design products, specifically furniture as users can directly buy the products generated with Stable Diffusion from our partner manufacturers.

AIException

This Application is developed for social media, prominently for images. Idea behind this project is to have a "Generate" option beside the usual "Upload". This project focuses on generating profile pictures and banners depending on social media platform a user want to generate for. "Prompt Profile Picture" have utilized Dall.E2, Dalle-mini and Streamlit.

OraOraOraOra....

Our service/ application is an amalgamation of streamlit with DALL-E's API and gradio with Whisper. In the case of DALL-E the user needs to give the number of images he/ she wants and add the corresponding image prompt or the imager description. The API will then generate the closest possible images to the given image description when the use clicks on the generate image button. Talking about the Whisper part, we used gradio to implement it. Over here it is capable to translate the speech or the audio input by the user to the text. This text can further be used for several applications.

Geeks

Bibliotopia is a service that enables you to search books from your descriptions.

Impact

With the Technological Advancement and growing pool of Knowledge base. It is getting difficult for Learners to Understand Specific Topics, Videos, and Audios in Fast pace. specially when we have Sea of Information on Google. So we came up with the Idea of ezTutor, that can help learners to understand specific topics and Evaluate themselves with the help of an AI Generated Content and Quizzes. In The ezTutor App: If you want to learn any topic, you can enter the text, and get the results with examples, images and keywords. If you want to learn by Video then by pasting YouTube's video url and get the reading content as well the summary of the topic, if you are running out of time. Similarly if you want to learn from Audio, recorded in a class then simply upload the audio. ezTutor will transcribe and Summarize the content. You can also test your learning by attempting the quiz.

Deepfai

An AI learning companion that answers questions and provides explanations and clarifications related to supplied literature for people with learning difficulties.

Book And I

Urecipy is a personal recipe notebook that allows its users to cleanse the content of food recipe YouTube videos and provide them with just the recipe content i.e. the ingredients used and the steps performed. It displays this result in both textual format as a recipe card and audio format (in case someone finds it hard reading the details). The application provides users to add as many recipes from YouTube and search them efficiently if the need arises.

Lone Warrior

Personal Elf is an AI-driven application for recommending gifts. Have you ever struggled with choosing the most suitable gift for one of your close ones? Don't worry, from now on your Personal Elf has your back! This application is driven by OpenAI's models. We used GPT-like transformer-based DaVinci model for generating gifts proposals and then visualized proposed gifts with diffusion model called DALL-E 2. We created a demo using streamlit that was also deployed using their service. We highly value inclusivity, therefore we designed our solution to be convenient also for other occasions, such as birthdays, anniversaries or hanukkah. Such flexibility allows us to make our product more accessible to users and it can be advertised accordingly during the year. While building our userbase and being able to drift afloat thanks to adverts, we want to get into partnerships with various online shops. We believe we can provide them with a great advertisement by integrating their shops into our application and recommending the users to buy on our partners' sites.

OpenGolem

The Goal of this project is to make organizing information simpler and to minimize the amount of clicks/taps a user needs to save dates/todos/goals. Making it easier for a user to manage their life events and objectives using AI to categorize tasks and building UI based on tags.

GeekCafe

*Generate hyper-realistic forensic sketches* SaaS that allows forensic sketch artists to improve the quality and speed of their work, creating a baseline hyper-realistic of the criminal, based on the witness description.

EagleAI

DocuSumm: less words, more meaning Our mission is to help people quickly and easily understand complex information by providing accurate and concise summaries of any type of documents. We believe that by simplifying and organizing information, we can empower people and companies to make better decisions and achieve their goals. During the hackathon we created an website, and Chrome extension and an API to summarize Youtube videos, with the goal to add more documents formats in the future. Our tech stack includes OpenAI's Whisper and GPT-3, as well as Python, PHP, MySQL, Lighttpd, Javascript, HTML and CSS.

DocuSumm

Game developers often face the challenge of coming up with unique and engaging game ideas, as well as organizing their thoughts and design elements in a cohesive manner. We developed RavenAI GameDev Toolkit in order to tackle these challeges by harvesting the catabilities of OpenAI's GPT-3 and DALL-E 2. Brainstorming ideas for game design is now easy! Simply select what you want to get ideas for from the sidebar, and fill in the form with any ideas that you may have. Press the button and.. Done! Behind the scenes, your inputs get processed and then passed to the AI models to get the final results. The tools are designed to complement each other in order to create a cohesive vision in the end, but they can also be used individually

Raven AI

Our goal at OpenCode is to provide users with solutions to their programming questions according to their needs. Opencode was developed to assist students in learning. OpenCode works with the OpenAI CodeX API. Our explainable AI collaborates with CodeX AI to provide systematic explanations of every line of code. To generate this code, we used the model code-cushman-001.

Data Smashers

Avocado - a mobile application to guide beginners in gym, do it safe and get results. Avocado can help you do gym in several ways: - Based on your exercises, it suggests what to add or remove to have balanced training for all muscle groups - Plans the best training plan for you providing enough time to rest and recover - Tracks your progress and provides instants feedback on how to improve and prevent harm To process video and audio information I use Whisper to get transcript. And GPT to extract information and make recomendations. There are two tasks: - The first one is convert video to exercises. We extract all exercises from the video along with all required information including name, summary, steps, timecodes, involved muscles and movement types (pulling or pushing). - Then we use AI in similar way to process audio recordings, which user makes during the training session. We extract number of repeats, weight, feel and harm. Then we use it to adjust current session, prevent harm and make recomendations with help of GPT

TechDock

Our application allows users without coding experience to create webpages by using simple voice commands. We utilized the Whisper AI to convert voice to text, and Chat-GPT3 to generate HTML and render a webpage. Users can then iterate on their design by giving the AI follow-up commands until the desired webpage is developed. Users can then copy the HTML source code and use in their personal webpages. Our mission is to make webpage design more accessible for people without a technical background who are interested in creating their own webpages for personal or business needs.

AI on the Fly

The Opinion Miner crawls data from different videos on Youtube. I dedicate this project to informing enterprises of what their customers and reviewers say about their products on the Internet, with filters: popularity of video, positive or negative opinion. so they can have better marketing strategies, improve products and improve customer satisfaction… Customers now have more and more choices as they can choose their favorite sellers regardless of their location, thanks to digitalization. Marketing strategies have switched their focus from whether goods can be sold or not, to customers' satisfaction and opinion, otherwise, they will find other sellers.

tpn

By offering a user-friendly and adaptable framework for website building, this project leverages text-DaVinci-003 and Whisper to create websites for people. This may be especially useful for individuals and small businesses that lack the technical skills or resources to create their own websites from scratch. To use the platform, a user would likely interact with their keyboards or whisper through a natural language interface. The user would provide information about their websites, such as the type of content they want to include and the design elements they prefer, and our application would use this information to generate a website based on pre-designed templates and customizations.

Code Jugadus

TrenchesAI is an AI powered educational blog/forum. Users of the blog get access to quality AI powered tools, articles and content related to Tech and AI. It is a one stop shop for newbies or anyone in particular trying to break in the field of AI or harness the power of AI for themselves The blog gives all users access to written content, premium users get access to the AI powered tools

TRENCHES AI

An virtual AI Psychotherapist which Helps and consoles people who were depressed and became mentally unstable.Speaks with them revive them from their Suicidal Thoughts and gives tasks to keep them engaged and Joyful.Its an Voice to voice AI.

OpenThunder

Our project is a guessing game where our user will attempt to guess the brand names from the logos shown in a webapp. The project will be built using the Python programming language via the python framework Streamlit. OpenAI will be used to provide speech-to-text capabilities for our project through the replicate API. We hope to target young kids with our project to teach them proper pronunciation and speaking through the playing of a guessing game using their voices. This project is therefore a small step towards what we hope to be able do in the future, to make learning fun and convenient, as games like this can be played by kids wherever and whenever.

Massive L

Try it out: https://customer-support-gpt.vercel.app/ Founders and customer support representatives often get an influx of repetitive emails that could be easily automated using GPT-3. The challenge is that GPT-3 knowledge is general and doesn't have company-specific information. Enter Customer Support GPT, it uses the company's internal knowledge such as frequently asked questions, articles, etc. to give company-specific answers. If it doesn't know the answer, it'll flag to be answered by a human. How it works: To get company-specific responses, Customer Support GPT does the following: Parse and create an index of company data. When an incoming customer support query is presented, the index is searched for the most relevant result(s). If none are present, flag it to a human reviewer. This result(s) are passed into GPT-3 along with the incoming query, to return a personalized, relevant response. Future work is to integrate with Zendesk, Crisp, and other providers to pre-generate answers and use in real life. Improve the GPT-3 prompt to return results that are closer to how customer support agents communicate Built by: Abdellatif - Ex-twitter engineer and founder of Tarteel.ai Ahmed - 17-year-old hacker, creator of remail.ai We're happy that you've reached this far, if you'd like to use the tool or have any questions please don't hesitate to reach out [email protected]

Customer Support AI

AtYou, it is mostly created to extract transcription from youtube and summarize it. The summarization will help people to know more about the video in short span or time or say they dont have to spend their time watching videos. It saves time, helps to know main points , also help in SEO to optimize recommendation by modelling main points. The app will also help people with hearing loss and non-native english speakers.

Sentient

The bookworm helps you to generate the image of a person or place from the description of that person/place .then it helps you to find the concept of the book or key points from it without reading the whole book then it help you get key notes or concept from audio books and can you can listen to the result as audio or read the result and find if this book is the right one to start /learn about the book without reading everything. this mainly helps children understand the story with visuals and reduce the time reading books.

The AI

AIShout leverages Whisper and GPT-3 capabilities to complete your meeting experience. Indeed, during meetings and encounters, a reporter has the tedious job to write and summarize everything into an appropriate template. For meetings, we usually use a Minute Template provided by the company. Let AIShout be that reporter for you.

YodAI

We built an app that listens to you speaking, transcribes it to text using whisper, then generates a formal email based on what you said with the help of the gpt-3 model.

AIWiz

Introducing Gist - the ultimate study tool for students of all levels! Whether you're in middle school or college, GistGen has you covered with thousands of expertly-crafted questions tailored to your specific grade and subject. simply choose your grade and subject, and GistGen will provide you with a wealth of practice questions to help you succeed. With GistGen your understanding of the material, boost your test scores, and achieve your academic goals. Not only that but Gistgen is an application that also helps you understand the main points of an article quickly and easily. It uses advanced natural language processing techniques to automatically summarize the content of an article, giving you a concise and comprehensive overview of the material. Whether you're trying to keep up with the latest news or want to dive deeper into a particular subject, Gistgen is the perfect tool for anyone looking to efficiently learn about new topics. With Gistgen, you can stay informed and stay ahead of the curve, no matter how much information you have to process.

Code Crusaders

Image Generation from Text Input using Dalle2. Can Be used to generate arts or pictures. which can be used for printing or marketing purposes or even inspiration for artists.

Non Zero

This is a small web service project that allows you to upload a mp3 audio or provide a youtube link - the source audio then gets transcribed and summarized by openAi models. The project was realized as part of OpenAI Whisper, GPT3, Codex & DALL-E 2 Hackathon together with colonelWalterKurtz and PioSikorski. The app was realized using python 3.10 with libraries such as Flask, openai, moviepy and pytube. The audio transcript is fed into the GPT-3 model in several pieces to ensure that it does not shorten and erase too much information. The prototype allows to convert short videos efficiently however it takes significantly more time to process longer audio files due to slow working of the requests to each model. The project provides a proof of concept that could potentailly be useful to many people who often do not wish to spend much time listening to audio files such as podcasts and if improved could allow to deliver such service online.

Summar'z3 web service

Our idea for a Shopify Print on Demand app that uses DALL-E inpainting would be to create an app that allows users to add custom images to their products. The app would use DALL-E to personalize the images based on a description provided by the user. For example, a user might want to add a picture of their grandparent’s dog to a t-shirt. They could then add props like a “Santa hat” to the dog, and DALL-E would generate the image and add it to the t-shirt or any other POD product. This would allow users to easily customize their products with unique images without needing any design skills. Another use case for this could be using the DALL-E outpainting tool, where a user might upload a picture of their own dog and ask the app to put it in a certain environment, i.e. “space”. The app would significantly reduce the turnover time and processes of personalized POD (currently 1-3 days) and design costs (currently $1-$5/design). We would charge merchants just $1 for each actual purchase on their store that was made via PrintAi. Overall, the app would make it easy for personalized POD merchants to allow their customers to add custom, high-quality images to their products, making their online stores more unique and appealing to customers.

PrintAi

Children of heaven🌸🧒🏡is a non profit educational oriented solution that uses AI to generate beautiful multilingual poems and relevant images. With its powerful language GPT3 model, it can create unique and inspiring multilingual poems on a wide range of childrens’ topics, and its Dall E model creates images that perfectly complement the poem ,We have built Audio input using gradio and whisper for multilingual input for kids who have difficulty in typing. Give children of heaven a try and discover the magic of multilingual poetry and art. Whether you’re a professional or kid , this app is sure to spark your creativity and inspire you to create something beautiful.

Millennials

This open-source project was built to give bloggers flexible tooling for their content creation. It only takes 5-10 minutes to set up and is cheaper than using services that mark up the price of OpenAI. The tool is streamlined to create higher-quality content by guiding the user thru a series of prompts.

Peaches

A Drag-and-Drop configurable solution for implementing conversational use of GPT-3 in Unity.

Dark Polarity AI

Novela ink is your own personal AI assistant platform to create/modify/enchance your stories. With power of OpenAI it was possible to create a storywritter that can really fit into your needs. With minimal effort you can quickly create stories, books, and creations. You don't need expensive graphic designers, and copywriters. It also helps to get an inspiration! And everything is tamperproof, and immutable thanks to immudb, so you can't really lose your creations. Everything in easy markdown format - could be exported to pdf. Features: - Full books management - AI story completion with different setups - AI story completion in-place - Image generation for selected text - Image generation for summary - Inspirations - Time Travel - Immutability - All AI actions saved

ApproxTeam

During the hackathon, we fine-tuned GPT-3 and built a self-analysis tool that helps one objectively assess their problem and develop new ideas for solving it. It can be used by people who can't access mental health care because of high prices and stigma. It is based on CBT and should be highly effective in the following cases: 1. A person has a problem and doesn't know how to solve it. For example, "I can't keep up with deadlines," or "My parents are overprotective." 2. A person can't make a decision. "Should I move?", "Should I accept an offer from a new company?" etc. 3. A person can't sort out their thoughts. "I can't understand why I'm so uncomfortable being a dad," "Why have I become so irritable?" etc. 4. A person wants to improve their relationship. "I'm so jealous," "We fight all the time," "I'm not happy with my wife. I cheated, and I feel guilty". In therapy, people who are objective about their situation and able to set specific goals tend to achieve better results. This tool does exactly that. A typical session consists of three parts: 1. Analysis. This part includes questions that make the person analyze various aspects of the situation and draw an objective picture. The essence of this part is the transition from an emotional to a rational perception of reality. 2. Empathy. It consists of a comprehensive generalizing statement aimed at supporting the client emotionally. 3. Decision. It consists of questions that allow the person to analyze the availability of resources and ways to solve the problem. Questions force the person to move from emotions to concrete steps toward the goal.

Elomia Health

Galaxy invites AI agents into collaboration on real-time whiteboards. It will provide an open marketplace to share replica of your intelligence trained on your messages and blogposts and earning when being invited to assist others on the whiteboard collaboration. The very first Galaxy AI Agent is deployed and available for communication via my personal telegram account: @galaxygur

Galaxy