README.md

---
title: Josh Talks AI
emoji: 🎙️
colorFrom: indigo
colorTo: pink
sdk: streamlit
sdk_version: 1.45.1
app_file: app.py
pinned: false
---

# Josh Talks AI 🎙️ 
*On a mission to make machines talk like humans.*

Welcome to the official Hugging Face organization for **Josh Talks AI** — a data-centric initiative focused on creating high-quality datasets.

🌐 **Learn more**: [https://ai.joshtalks.com](https://ai.joshtalks.com)

---

## 🧭 Problems with ASR
- *Conversational multi speaker speech*

- *Dialect and Accented speech*

- *Child Speech*

- *Small models for real time low latency*

## 🧭 Our Mission

We are building open datasets to power the next generation of Speech AI — across languages, domains, and communities.

## 🧭 Our Work

**Datasets**  - Highest quality scientifically designed datasets for training models.

**Benchmarks** - We can only fix what we can measure. Our benchmarks tell researchers exactly where their models fail.


## 📦 Coming Soon

We are currently preparing a series of open datasets, including:

- **Multilingual speech datasets**  
  Transcribed Indian language audio from real-world conversations and talks.

- **Cultural & regional text corpora**  
  Clean, annotated text datasets in Hindi, Tamil, Bengali, Marathi, and more.

- **Media-rich data from Indian contexts**  
  Video metadata, subtitles, and emotion-rich labels for AI-driven storytelling.

Stay tuned — our first releases are just around the corner!

## 🤝 Let's Collaborate

We welcome partnerships with researchers, universities, NGOs, and AI developers working on ASR challenges. If you're interested in contributing or using our datasets, reach out below.

---

## 📬 Contact

🌐 **Website**: [https://ai.joshtalks.com](https://ai.joshtalks.com)  
💼 **LinkedIn**: [Josh Talks](https://www.linkedin.com/company/joshtalks/)