Recently I developed voice_convertor application which can work fully offline.
The client asked me to build it was musician (my thought) from Los Angels. He was a fan of Michael, a well-known singer in his region and had been using Voqul to convert othe similar singer's voice into that Michael's style. But, he faced several limitations, which the tool was cloud based, had unstable connectivity, and the output quality was inconsistent.

He asked me to build a fully offline solution with higher-quality results and real-time capabilities.

To train the model, he provided around 20 audio tracks (each ~3 minutes long). Based on this dataset, I developed a voice conversion system that runs entirely offline. The application includes the following features (actually this is a simple app).

1. File-to-file voice conversion

2. Real-time voice streaming

3. Pitch control for fine-tuning output

4. Optimized inference for low-latency performance

We shipped Noir Discover last week. A community prompt feed built into the Noir Prompt dashboard.

The idea: instead of every AI creator starting from zero, you can browse prompts shared by other users, save any of them to your personal library in one click.

We built iQuelo Lead Prospector to help field sales teams prospect smarter, prioritize better, and reduce the cost of selling.

Instead of using disconnected tools for lead discovery, routing, and follow-up, we wanted one workflow that actually helps teams execute.

Would love your feedback from the Product Hunt community.

https://leadsprospector.iquelo.com

Are you solving hundreds of LeetCode problems but still struggling to land that Big Tech offer?

Companies like Google and Meta aren't just checking if your code compiles. They're looking for "Signal" your communication, edge-case handling, and architectural trade-offs.

Most people think lead-gen is just about asking for an email. I built Interactify to prove it s about making the decision for them.

In this demo, I show how to build a "Tech Stack Architect" recommender. Instead of just a generic quiz, this engine uses a specialized logic layer to weigh user answers and match them with specific solutions.

I designed this to be "One Brain." You manage your products, set logic weights for each answer, and the engine calculates the winner in real-time. No code, just logic.

Watch the build here:

Hey Product Hunt!

I'm Calvin, co-founder of ClarityAI, and today is Day 1 of getting us ready for GTM.

The honest version of why we built this:

I read 16 IRS publications from Korea. Then I found several government audit reports from the GAO. Here's the number that stopped me cold.

I am not American. I've never filed a Schedule C. I've never spoken to the IRS.

But I was already building a product for 31 million U.S. self-employed people. So I needed to understand their tax reality.

16 IRS publications. Multiple audit reports. Hundreds of pages. One month.

Hey Product Hunt!

Ever spent ages transcribing a podcast, interview, or lecture? I have too, and it s no fun. Early on in the STT and TTS tech wave, many great products appeared - shout out to @Otter, my longtime favorite.
But now, having all tools in one place is essential, since transcription is just one step in solving bigger problems or building projects

So we built Tila - an AI tool that turns hours of transcription into minutes. It handles multiple languages, speaker identification, and cleans up audio. And Tila is more than transcription - it s a multi-agent platform for audio, text, and visuals all in one.. transcribing, generating voiceovers, or processing images - Tila does it in seconds.

We re launching soon and we plan to actively enhance our product's capabilities. We NEED your feedback! What s your biggest transcription/voice-over pain? Any specific features you wish for?

Focus Quadrant is a task manager that forces you to prioritize, not just list.

Most task managers give you an infinite list and leave the hard work to you. Focus Quadrant takes a different approach: every task starts in a queue, and you decide where it belongs. Do (urgent and important), Plan (not urgent, important), Delegate (urgent, not important), or Drop (neither). That single drag-and-drop decision clarifies what actually deserves your time.