adalat-ai/koyal-kn-120m-1.0
Automatic Speech Recognition • Updated
Kumarmanas Nethil
janaab
AI & ML interests
speech, dialog systems and low-resource ML
Recent Activity
published a model about 7 hours ago
adalat-ai/koyal-kn-120m-1.0 published a model about 7 hours ago
adalat-ai/koyal-indic-600m-1.0 published a model about 7 hours ago
adalat-ai/koyal-ml-120m-1.0Organizations
published 5
models about 7 hours ago
published an article 1 day ago
Article
Diagnostic ASR evaluation for Indic and domain-specific speech
adalat-ai
• • 1 upvoted an article 16 days ago
Article
One sandbox per rollout, or how labs run RL for agents in 2026
sergiopaniego
• • 10 liked a
model 16 days ago
upvoted an article about 1 month ago
Article
SraVaani: How Vision Helps Hearing
ARTPARK-IISc
• • 6 liked a
model about 2 months ago
reacted to kavyamanohar's post with 🚀 2 months ago
Post
170
Some bugs teach you more than they cost you. It's a small bug with a big lesson I keep running into: speech tools are built and benchmarked on English, so their limits are quietly calibrated for English. You only find the edges when you work in the languages they weren't tested on.
Learnt things the hard way, wrote it down so you don't have to.
https://huggingface.co/blog/adalat-ai/whisper-token-limit
Learnt things the hard way, wrote it down so you don't have to.
https://huggingface.co/blog/adalat-ai/whisper-token-limit
upvoted an article 3 months ago
Article
Why Whisper cuts off Indic transcripts after six seconds
adalat-ai
• • 5 updated a
collection 4 months ago
liked a
model 4 months ago
upvoted a paper 4 months ago
upvoted an article 4 months ago
Article
How to build scalable web apps with OpenAI's Privacy Filter
- +1
ysharma, freddyaboulton, abidlabs
• • 16 authored a
paper 4 months ago
Reformat dataset for Dataset Viewer + add Dataset Card
2
#1 opened 5 months ago
by janaab
reacted to kavyamanohar's post with 🚀 4 months ago
Post
4603
Releasing Vividh-ASR — an open benchmark and models for Hindi and Malayalam ASR.
Vividh-ASR is built from public data, stratified by complexity:
→ Clean recordings
→ Noisy and accented speech
→ Spontaneous, conversational audio
Alongside the benchmark, we release:
→ Open models for Hindi and Malayalam
→ A training recipe with two counterintuitive choices that moved the needle
→ What failed, not just what worked
The stratified evaluation methodology transfers directly to any low-resource language setup — beyond Hindi and Malayalam.
Built at @adalatai , where we build speech tech for Indian courts. This is our first open contribution back to the community. @janaab @Kush0610 @orgh0
Link: https://huggingface.co/blog/adalat-ai/vividh-benchmark
Vividh-ASR is built from public data, stratified by complexity:
→ Clean recordings
→ Noisy and accented speech
→ Spontaneous, conversational audio
Alongside the benchmark, we release:
→ Open models for Hindi and Malayalam
→ A training recipe with two counterintuitive choices that moved the needle
→ What failed, not just what worked
The stratified evaluation methodology transfers directly to any low-resource language setup — beyond Hindi and Malayalam.
Built at @adalatai , where we build speech tech for Indian courts. This is our first open contribution back to the community. @janaab @Kush0610 @orgh0
Link: https://huggingface.co/blog/adalat-ai/vividh-benchmark