Kathbath

by AI4Bharat (IIT Madras)

1684h read-speech ASR benchmark across 12 Indian languages.

TL;DR

1684h read-speech ASR benchmark across 12 Indian languages.

Best for indic ASR evaluation with consistent splits across 12 languages. Pricing: free.

Category
Open source
License
—
Stars
—
Last push
—
Pricing
free
Platforms
HuggingFace, GitHub

What it is

Kathbath is a 1684-hour benchmark across 12 Indian languages, crowd-sourced via smartphones. Evaluation companion to IndicSUPERB. License: CC BY 4.0.

Best for: Indic ASR evaluation with consistent splits across 12 languages.
Watch out for: CC BY 4.0 · read speech bias (smartphone-recorded). Cite: Javed et al., 2022.

Install / use

from datasets import load_dataset; ds = load_dataset('ai4bharat/kathbath')

Features

Speaker diarizationNo
Word-level timestampsNo
Streaming / real-timeNo
Languages supported12
HIPAA eligibleNo

Kathbath vs Whipscribe

FeatureKathbathWhipscribe
CategoryOpen sourceTranscription APIs
Pricingfreefree beta
Speaker diarizationNoYes
Word timestampsNoYes
StreamingNoNo
Languages1299
PlatformsHuggingFace, GitHubWeb, API, MCP

Alternatives to Kathbath

OpenAI Whisper
OpenAI

The reference open-source multilingual ASR model from OpenAI.

OSS · MIT ★ 98.1k
whisper.cpp
Georgi Gerganov

C/C++ port of Whisper — runs on anything, from a Raspberry Pi to Apple Silicon.

OSS · MIT ★ 48.8k
faster-whisper
SYSTRAN

4× faster than reference Whisper using CTranslate2 — production sweet spot.

OSS · MIT ★ 22.3k

Whipscribe is a managed faster-whisper + whisperX service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.