Indian Languages Audio Dataset

Indian Languages Audio Dataset

5-second MP3 audio samples across 10 Indian languages including Marathi, sourced from YouTube regional videos. Designed for spoken language identification and audio classification tasks rather than ASR transcription.

Build a multilingual Indian language identification system from audio that includes Marathi detection.
HomepageDownload

Quick Start

# Indian Languages Audio Dataset
import torchaudio
# Filter for Marathi subset
print("Access the Indian Languages Audio Dataset")
print("Filter for Marathi (mr) language code")
Modality
audio
Size
~748 MB; 5-second clips; 10 Indian languages including Marathi
License
Format
MP3
Language
mr, hi, en
Update Frequency
static
Organization
Independent researcher (Kaggle)
Category
speech-audio

Schema

FieldTypeDescription
audioaudioAudio recording in Indian language
textstringTranscription text
languagestringLanguage identifier

Build With This

Create a language-specific audio router for multilingual call centers serving Maharashtra's diverse population
Develop an Indian language ASR meta-model that leverages cross-lingual transfer from this multi-language dataset
Build a dialectal variation study comparing Marathi audio features against other Indo-Aryan languages in the dataset

AI Use Cases

Spoken language identification (Marathi vs other Indian languages)Audio classification and dialect detectionMultilingual call routing for contact centersLanguage detection in mixed-language broadcast media
Last verified: 2026-03-09