Show HN: ShortLoop – Replay Audio Calls and Test Your VoiceAI https://bit.ly/3WoYr0O

Show HN: ShortLoop – Replay Audio Calls and Test Your VoiceAI VoiceAI builders can replay production calls (or) create voice test data set and replay those calls to your VoiceAI. With this( https://bit.ly/4ds3cxr ) you can now test end to end flows frequently and in an automated way. We have been tinkering with LLMs and building voice applications for the past few months. As we were trying to improve our voice bots performance (interruptions, latency of tts, llm etc) it became very tedious to test our bot over and over again speaking to it. With real users we would observe edge cases that we didn’t handle before. After making changes it was difficult to recreate that exact scenario (accent, noise, slow speech or gaps between sentences) and test changes. So we built a way to use any existing call recording and replay ‘user’ portion of the call to the bot. Since LLMs won’t repeat the exact text or follow same flow on each run, it became important to handle those deviations. So took sometime experimenting and came up with this approach: 1. Segment call in to audio snippets of dialogue 2. Create a transcript and understand flow, intent, details of the call. 3. In response to LLM, identify correct user audio snippet to replay based on the flow/context (lot of work here :D) 4. In cases where it is a new question or reconfirmation by bot, generate new text+audio. I think this is pretty cool and would love for you to try it over your VoiceAI bots and give feedback to improve it. Earlier we also built a visualisation for analysing audio calls (player, transcript and basic errors) and included it. Here is a quick demo of how both look: https://youtu.be/j3kRhSxD5P0 What’s next? Thinking of creating more voice based automated tests. Open to ideas and would love to know how you improve your VoiceBots. https://bit.ly/4ds3cxr August 2, 2024 at 08:03AM

Show HN: Zig Inspired Standalone OpenCL Compiler https://bit.ly/3SCxR39

Show HN: Zig Inspired Standalone OpenCL Compiler Hello HN, I’d like some feedback on a recent project: openclc. As the name suggests it’s an AOT compiler for OpenCL-C/C++ code targeting SPIR-V to be consumed by `clCreateProgramWithIL`. Coming from CUDA, I liked using the OpenCL language on a school project. That being said, I found the compile at runtime, put the kernels in a c string flow to be janky and off-putting. Thankfully, Khronos created an LLVM backend that converts LLVM IR to SPIR-V. Despite the good code in the SPIRV-LLVM-Translator, it leaves much to be desired when it comes to packaging. It requires a build from source against a system LLVM installation, doesn’t do SPIRV-Opt performance passes, and leaves you to figure out the inclusion of the SPIR-V into your program. Openclc bundles clang/llvm, the LLVM-SPIRV translator, and spirv-opt performance passes into one static executable. It can output the SPIR-V directly or as a C array so you can easily embed it into a binary. * I also included builds of Spirv-Tools for windows, linux, and macos. Future Idea: OpenCLC Runtime The biggest problem with OpenCL is the ardous and error prone device discovery and kernel scheduling. It would be a huge boost to OpenCL usability to offload device discovery and scheduling to a runtime library like CUDA does with the CUDA Runtime. Instead of just compiling cl sources to SPIR-V, it could offer a regular c symbol for each kernel where the `clEnqueueNDRangeKernel` ugliness is handled underneath the hood. With sufficient abstraction the OpenCL backend could be swapped for Vulkan, Level Zero, and maybe even Metal through SPIR-V cross. I'd love to answer any questions. https://bit.ly/3yqf6Js August 2, 2024 at 03:42AM

Show HN: Mongo Explorer – Open-source AI-powered MongoDB management tool https://bit.ly/4fx7m95

Show HN: Mongo Explorer – Open-source AI-powered MongoDB management tool Show HN: Mongo Explorer – Open-source AI-powered MongoDB management tool Mongo Explorer is an open-source project that integrates AI into MongoDB management, aiming to simplify database optimization and query generation. Key features include: 1. AI-assisted query generation: Convert natural language to MongoDB queries 2. Intelligent index suggestions with one-click creation 3. Visual query performance analysis and profiling 4. Schema exploration for collections and queries Tech stack: - Frontend: React - Backend: ASP.NET Core 8 - Deployment: Docker The AI integration uses OpenAI's GPT models to analyze database structures, generate optimized queries, and suggest performance improvements. GitHub: https://bit.ly/3Wy83qb Demo: https://bit.ly/3Wr8zGt I built this tool to address common pain points in MongoDB management, particularly for complex databases. I'm interested in the HN community's thoughts on AI-assisted database management and potential future developments in this space. https://bit.ly/3Wy83qb August 1, 2024 at 09:17PM

Show HN: Using AI to Generate Custom Sounds from Text https://bit.ly/4dp1RHM

Show HN: Using AI to Generate Custom Sounds from Text I was tired of finding videos with sound effects , extracting the audio then extracting the audio I want. So I made AI get to work for me :) https://bit.ly/4c4rpsI August 1, 2024 at 11:52PM

Show HN: Run Llama3.1 405B on a 8GB VRAM challenge [video] https://bit.ly/46BxQT2

Show HN: Run Llama3.1 405B on a 8GB VRAM challenge [video] How to run Llama3.1 405B on a 8GB VRAM https://www.youtube.com/shorts/fTM-48cu1Vg August 1, 2024 at 05:18AM

Show HN: I build a Python library for zaprite for Bitcoin payment processing https://bit.ly/3Wuvel7

Show HN: I build a Python library for zaprite for Bitcoin payment processing Hey HN! I am Jogi, a solo dev, and I made my first ever FOSS library because I needed to plug into bitcoin payments for a client project and I thought it might help others more easily do the same. My main motivation was to lower the barrier for developers who want to integrate Bitcoin payments into their flask or django projects. I'm not entirely sure if it's any good, which is why I wanted to share it here. Roast it, I can take it lol Anyway, I rarely get feedback on my work from devs, so I'm really looking forward to hearing your thoughts, critiques, and suggestions. If you have experience with Bitcoin integration or library development, I'd especially appreciate your insights. Thanks for checking it out! https://bit.ly/3WlASpu August 1, 2024 at 04:27AM

Show HN: Failure Hunt – failure stories from entrepreneurs, makers, and experts https://bit.ly/46w9Q3m

Show HN: Failure Hunt – failure stories from entrepreneurs, makers, and experts Successful entrepreneurs, creators, indie hackers, and professionals share their failures and how these experiences have helped them achieve new goals. https://bit.ly/46ygJRR July 31, 2024 at 10:31PM

Show HN: Non SaaS – Directory of Non SaaS Apps https://bit.ly/46ssWaS

Show HN: Non SaaS – Directory of Non SaaS Apps https://bit.ly/3AabwUs July 31, 2024 at 11:19PM

Show HN: I made an online journaling app focused on day overview using emojis https://bit.ly/4d0Y0kw

Show HN: I made an online journaling app focused on day overview using emojis Please let me know what you think! https://bit.ly/3WJ27vC July 31, 2024 at 01:00AM

Show HN: A Path-Based Data storage/retrieval web service to prevent crawling https://bit.ly/4fmTBd9

Show HN: A Path-Based Data storage/retrieval web service to prevent crawling I listed my email on GitHub, and it got collected and used without my consent... To prevent this, I created a simple service using Django and PostgreSQL. As you can see when you access the site, you can save data by specifying a path. Think of it as a global key-value storage designed simply to prevent crawling. The data is not stored in an encrypted form, so please do not save sensitive information. Thank you. P.S.: Yes, I am a Faker's fan. https://bit.ly/4ftTk8b July 31, 2024 at 12:10AM

Show HN: Merlin – Contextual Intelligence Agent for GTM https://bit.ly/3LHNtP0

Show HN: Merlin – Contextual Intelligence Agent for GTM Hi HN! We are building [Merlin]( https://bit.ly/3yfJI0f ), a GTM intelligence platform that gathers contextual insights about prospects and key stakeholders, so sellers don’t have to spend hours on tedious research tasks across hundreds of companies. Our thesis is that content generation and workflow/sequence automation providers are generating a lot of noise in social media (and inboxes) today and they are making the problem of outbound sales worse. We believe that every customer interaction (emails, cold calls or discovery calls) should be powered by intelligence. With the relevant insights, the user can make an informed decision whether to contact the prospect, understand what the prospect is interested in and what are their pain points that the user can solve. We’ve built two main features: Lite Intel - allows you to ask any number of questions across 100s companies. Please feel free to try out this feature for free - https://bit.ly/3WJ1zGc... Deep Intel - provides contextual insights about prospects and key buyers, by automatically understanding the user’s solution and service. This includes identifying new qualified leads based on contextual intelligence, instead of filtering based on # of employees, sectors and technology used. There are several use cases you can apply with our contextual insights: 1. Qualify leads 2. Identify new highly qualified accounts 3. Draft hyper targeted emails 4. Prepare for discovery and follow-up calls 5. We’d love to hear suggestions of other use cases. Could be non-GTM related too We believe the future is in Insights-Led Growth . We’re still pretty early in the development and want to keep on improving on our contextual insights. Any feedback is much appreciated! Thank you in advance! https://bit.ly/3yfJI0f July 30, 2024 at 10:17AM

Show HN: Bob-editor – create and edit ASCII art graphically https://bit.ly/3St5bti

Show HN: Bob-editor – create and edit ASCII art graphically https://bit.ly/3CvxlfZ July 30, 2024 at 04:30AM

Show HN: Turn any website into a knowledge base for LLMs https://bit.ly/3WHeyYR

Show HN: Turn any website into a knowledge base for LLMs https://bit.ly/3yfuFUj July 30, 2024 at 01:54AM

Show HN: ChainFactory – Run Structured LLM Inference with Easy Parallelism https://bit.ly/4c55MZ2

Show HN: ChainFactory – Run Structured LLM Inference with Easy Parallelism Hi HN! Disclaimer: I submitted another post about ChainFactory a few days ago. Here's what has changed since: - Added hash based caching of auto-generated prompts and masks. - Did some internal restructuring and cleanup. - Updated the order in which README doc introduces concepts and terminology. Posting this again because honestly, I am kinda puzzled about what to add/fix/change due to having 0 users and no genuine feedback. By genuine feedback, I mean feedback from strangers who do not have a social pressure to be polite and pull punches. Please take a look if you find this interesting and leave a comment. If you think it's an deranged or stupid idea not worth your time, please at least leave a 'no' - I'd still be delighted as it's an honest opinion. Thanks a lot! PS: Is it okay to post updates and changes at regular intervals? https://bit.ly/3Wo3crx July 30, 2024 at 02:48AM

Show HN: I built an AI tool to turn Excel data into 10-page reports in seconds https://bit.ly/4c1NqbA

Show HN: I built an AI tool to turn Excel data into 10-page reports in seconds Hey everyone! I'm Alex, the co-founder of Excel Dashboard Ai. I’m excited to share our AI tool that turns your Excel data into awesome dashboards and detailed reports. Think visualized charts, diverse analysis angles, and almost 10,000 words of AI-driven insights—all in one place. Say goodbye to endless hours of data analysis and report writing. ExcelDashboard is here to boost your productivity! Why did we build this? Back in my previous company, we were constantly asked for data-driven insights and long reports. This process was a huge time drain. I used to joke, "Wouldn't it be great if something could handle the entire report creation process?" Turns out, many felt the same way. Creating these reports was a big hassle for everyone. Then the AI revolution hit I thought, what if AI could help non-tech users generate analysis and whip up a 20-page report in seconds? That’s how ExcelDashboard.ai came to be—your personal data assistant. What can ExcelDashboard.ai do? AI Batch Analysis Let AI explore 10 different data analysis angles for you. Keep swapping until you're happy! AI Dashboard No more merging data and building charts! AI creates custom dashboards from your Excel files with one click. AI Data Report Forget spending hours writing reports! AI generates a 20-page analysis report with just one click. AI Data Insights No more puzzling over data! AI explains the numbers in simple, easy-to-understand language. AI Data Query Instant answers in plain English. Let AI handle the coding, formulas, and charts. Transform your Excel data into dashboards and reports with AI in just 2 clicks! Save hours and boost your productivity. Who is this for? Perfect for business professionals, marketers, operations folks, students, and anyone who needs quick and easy data analysis and reporting. Why try ExcelDashboard today? To celebrate our launch, we’re offering an exclusive 75% off for all hackers! Use code YXMTQ0NQ. We’re all about helping non-tech users simplify data analysis. Try ExcelDashboard.ai and let us know what you think! https://bit.ly/3LK7xAu July 29, 2024 at 03:15AM

Show HN: How I wrote a LaTeX paper without writing any LaTeX https://bit.ly/4c2zX39

Show HN: How I wrote a LaTeX paper without writing any LaTeX Stempad is an online scientific text-editor. I built it because I wanted a way to do fast scientific writing with just my keyboard, and LaTeX wasn't cutting it. I recently launched the ability to export Stempad documents to LaTex. I tested it by rewriting part of a paper I found online (Metabolic scaling in small life forms by Marc E. Ritchie & Christopher P. Kempes) and exporting it. You can try the editor and export yourself using the post url. The export button is on the top right of the page. In case you want to see the result directly, this was it: https://bit.ly/3LI6GjD Feedback is really appreciated! If anyone thinks they might find Stempad useful, let me know and I'd love to get in touch. https://bit.ly/4c0mbOx July 28, 2024 at 11:54PM

Show HN: Slide the daily word into the bottom row or: Caesar's 20-puzzle https://bit.ly/46l1R9v

Show HN: Slide the daily word into the bottom row or: Caesar's 20-puzzle https://bit.ly/3SqrcsF July 25, 2024 at 12:39AM

Show HN: WordPress Plugin Compatibility Checker https://bit.ly/3ygZN5K

Show HN: WordPress Plugin Compatibility Checker Use cases: * Verify compatibility of new plugin versions with your current setup. * Assess compatibility of your current WordPress version and installed plugins. * Discover the latest available versions of WordPress and your plugins. Available via API and WebUI. https://bit.ly/3YoBqxA July 27, 2024 at 07:25PM

Show HN: SaaS marketing that's out of this world https://bit.ly/3Wlyqj4

Show HN: SaaS marketing that's out of this world https://bit.ly/3YsNBt8 July 27, 2024 at 02:35PM

Show HN: Symbols > We are building Figma for developers https://bit.ly/3Yi0C8T

Show HN: Symbols > We are building Figma for developers What is Symbols? “The ultimate platform for developers & teams to build, test and document fully functional & reusable UI libraries on a canvas. Publish as a website or export to your existing tech stack with open-source development” Lifetime offers: We are currently running $100 one off offers for the first 100 users, which includes unlimited projects or users for 1 project. Eventually charging $49 per month per user/project, as design system platforms charge a similar amount. So grab a deal! Landing: Here is a quick landing put together, with a better one in the works :) https://bit.ly/4cXZo7s Any questions let me know! https://bit.ly/4cXZo7s July 26, 2024 at 09:17PM