---
title: "ARTPARK (IISc) models: release history and timeline"
description: "Not-for-profit AI and robotics foundation at the Indian Institute of Science that runs Project Vaani with IISc and publishes Indian-language speech models."
type: "release-archive"
canonical_url: "https://theaiengineering.com/releases/artpark-iisc"
markdown_url: "https://theaiengineering.com/releases/artpark-iisc.md"
author: "The AI Engineering Editors"
publisher: "The AI Engineering"
updated: "2026-08-08"
citation: "The AI Engineering Editors. \"ARTPARK (IISc) models: release history and timeline.\" The AI Engineering, updated Aug 08, 2026. https://theaiengineering.com/releases/artpark-iisc"
license: "CC BY 4.0"
license_url: "https://creativecommons.org/licenses/by/4.0/"
data_csv: "https://theaiengineering.com/releases.csv"
data_json: "https://theaiengineering.com/releases.json"
---

# ARTPARK (IISc) models: release history and timeline

Not-for-profit AI and robotics foundation at the Indian Institute of Science that runs Project Vaani with IISc and publishes Indian-language speech models.

- Headquarters: Bengaluru, India
- Website: https://artpark.in

1 release from 1 lab, 1 with open weights. Dates are the labs' official announcement dates.

## August 2026

### SraVaani-1.0

SraVaani-1.0 is a speech recognition model from ARTPARK and IISc's SPIRE Lab that transcribes dozens of Indian languages and dialects, including many low-resource and tribal languages.

- Date: 2026-08-08
- Lab: ARTPARK (IISc)
- Type: Speech
- Weights: Open weights
- Modalities: audio → text
- Parameters: 430M
- License: MIT
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2026/08#sravaani-1-0

Highlights:

- The team trained it from scratch in three stages: self-supervised pretraining on about 31,255 hours of Project Vaani speech across 105 languages, audio-image alignment on 11 million pairs, then fine-tuning on about 31,270 hours of transcribed speech
- It uses a FastConformer encoder with a hybrid TDT-CTC decoder at roughly 430M parameters
- The paper reports coverage of 65 Indian languages and dialects; the model card noted Urdu and Kashmiri were not supported in the released checkpoint
- Weights were released on Hugging Face under MIT, behind an access-conditions form

Source: [ARTPARK-IISc/SraVaani-1.0 · Hugging Face](https://huggingface.co/ARTPARK-IISc/SraVaani-1.0)

## Cite and reuse this data

These release records are free to reuse under CC BY 4.0, including commercially. Credit The AI Engineering and link to this page's URL. Dates are the labs' official announcement dates, and every record links to its source. List fields in the CSV are separated by " | ".

The AI Engineering Editors. "ARTPARK (IISc) models: release history and timeline." The AI Engineering, updated Aug 08, 2026. https://theaiengineering.com/releases/artpark-iisc

HTML link:

```html
<a href="https://theaiengineering.com/releases/artpark-iisc">ARTPARK (IISc) models: release history and timeline</a> (The AI Engineering)
```

- Title: ARTPARK (IISc) models: release history and timeline
- Author: The AI Engineering Editors
- Publisher: The AI Engineering
- Updated: 2026-08-08
- URL: https://theaiengineering.com/releases/artpark-iisc
- Markdown: https://theaiengineering.com/releases/artpark-iisc.md
- License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)
- Full dataset as CSV: https://theaiengineering.com/releases.csv
- Full dataset as JSON: https://theaiengineering.com/releases.json
