MLRD-20: Marathi Lip Reading Dataset

MLRD-20: Marathi Lip Reading Dataset

MH Specific

Marathi lip reading dataset containing video recordings of speakers pronouncing Marathi words and phrases, designed for visual speech recognition and lip-reading AI systems. One of the few lip-reading datasets for any Indian language.

Build a Marathi lip reading model for silent speech recognition in noisy environments or for hearing-impaired users.
HomepageDownload

Quick Start

# MLRD-20 Marathi Lip Reading Dataset
import cv2
print('MLRD-20: Marathi Lip Reading Dataset')
print('20 Marathi words with video recordings')
Modality
video
Size
20 classes; video recordings of Marathi speech
License
Format
Video files
Language
mr
Update Frequency
static
Organization
Independent researcher (Kaggle)
Category
vision-ocr-multimodal

Schema

FieldTypeDescription
videovideoVideo clip of speaker's face
textstringSpoken text transcription in Marathi

Build With This

Create an assistive communication tool for hearing-impaired Marathi speakers using visual speech recognition
Develop a multimodal Marathi ASR system combining audio and visual cues for robust speech recognition
Build a Marathi speech privacy tool that recognizes speech from video without audio for security applications

AI Use Cases

Marathi visual speech recognitionLip reading for hearing-impaired accessibilityMultimodal speech recognition combining audio and videoSilent speech interfaces
Last verified: 2026-03-09