README.md

metadata

title: Memoni Audio Labeling Tool
emoji: 🎙️
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 6.14.0
python_version: '3.13'
app_file: app.py
pinned: false
license: mit

Memoni Audio Labeling Tool

Help build the first Memoni language speech dataset by transcribing short audio clips.

How to Use

1. Enter your name and click Start Labeling
2. Press play and listen to the audio clip
3. Type what you hear in the text box
4. Choose your script: Roman, Urdu, or Mixed
5. Click Submit — the next clip loads automatically
6. Click Skip if the audio is unclear or silent

No account needed. No install. Just open and start.

---

Writing Guide

You can write in any of these — choose what feels natural:

Style	Example
Roman (English letters)	mein ghar aao bohat acha hai
Urdu script	میں گھر آؤ بہت اچھا ہے
Mixed	mein گھر aao بہت acha ہے

---

Common Words Reference

Roman	Urdu	Meaning
mein	میں	I / in
hai	ہے	is
hain	ہیں	are
aur	اور	and
nahi	نہیں	no / not
kya	کیا	what / did
aao	آؤ	come (plural)
jao	جاؤ	go
yeh	یہ	this
woh	وہ	that
ghar	گھر	house
bohat	بہت	very
acha	اچھا	good / okay
han	ہاں	yes
aaj	آج	today
ab	اب	now
sath	ساتھ	with
kab	کب	when
kahan	کہاں	where
kyun	کیوں	why

---

Roman Writing Rules

Use these canonical spellings so everyone writes consistently:

Write this	Not this
mein	main, me, mn
hai	hy, he, hay
aur	or, ur
nahi	nahin, nai, ni
kya	kia, kiya
aao	aavo, aaw
yeh	ye, ya
woh	wo, wu
ghar	gher
bohat	buhat, bhut
acha	accha, atcha

---

Privacy

Your name and transcripts are stored in a private HuggingFace dataset. They will only be used to train the Memoni language model.