Marathi Speech-to-Text with Lowest WER

Built for every Marathi speaker across India at sub-70ms latency

Click anywhere to start transcribing

Experience pulse speech to text

Marathi Speech-to-Text with Lowest WER

Built for every Marathi speaker across India at sub-70ms latency

Click anywhere to start transcribing

Experience pulse speech to text

Marathi Speech-to-Text with Lowest WER

Built for every Marathi speaker across India at sub-70ms latency

Click anywhere to start transcribing

Experience pulse speech to text

Marathi Transcription Benchmarks

Marathi Transcription Benchmarks

Marathi Transcription Benchmarks

Model
FLEURS
Streaming
FLEURS
Batch
Smallest
7.5% WER
6.6% WER
Scribe v2
14.5% WER
12.8% WER
Deepgram
20.5% WER
18.0% WER
Whisper large v3
12.5% WER
11.0% WER

World’s Most Advanced Speech Intelligence

Go beyond text with automated speaker labeling, real-time sentiment analysis, and intelligent language identification for global production workloads.

World’s Most Advanced Speech Intelligence

Go beyond text with automated speaker labeling, real-time sentiment analysis, and intelligent language identification for global production workloads.

World’s Most Advanced Speech Intelligence

Go beyond text with automated speaker labeling, real-time sentiment analysis, and intelligent language identification for global production workloads.

Language overview

Marathi language overview

About

Indo-Aryan language of Maharashtra, India. Uses the Devanagari script shared with Hindi. The third most spoken language in India with deep Prakrit and Sanskrit roots.

Speakers

95 million

Official language

Maharashtra (India)

Accents

Standard Marathi, Varhadi, Konkani-inflected, Nagpuri

Spoken language demographic

Maharashtra, India, and Marathi diaspora worldwide

For Developers

Automate.Orchesrate. Dominate with code

Build real-time speech pipelines with Node and Python SDKs. Audio in, transcription out — no middleware, no complexity.

For Developers

Automate.Orchesrate. Dominate with code

Build real-time speech pipelines with Node and Python SDKs. Audio in, transcription out — no middleware, no complexity.

For Developers

Automate.Orchesrate. Dominate with code

Build real-time speech pipelines with Node and Python SDKs. Audio in, transcription out — no middleware, no complexity.

01const url = 'https://api.smallest.ai/waves/v1/pulse/get_text';02const options = {03 method: 'POST',04 headers: {05 Authorization: 'Bearer <BearerAuth>',06 'Content-Type': 'application/octet-stream'07 }08};09 10try {11 const response = await fetch(url, options);12 const data = await response.json();13 console.log(data);14} catch (error) {15 console.error(error);16}

Certified & Compliant

Guarding your data with enterprise security

Certified & Compliant

Guarding your data with enterprise security

Certified & Compliant

Guarding your data with enterprise security

Proactive Defense

Anticipating threats before they emerge, thanks to our advanced monitoring.

Proactive Defense

Anticipating threats before they emerge, thanks to our advanced monitoring.

Proactive Defense

Anticipating threats before they emerge, thanks to our advanced monitoring.

Frequently
asked questions

What formats are supported?

Does Pulse handle Marathi-English code-switching?

What's the latency?

Is speaker diarization built in?

What makes Pulse different from Deepgram or AssemblyAI?

Can I use Pulse for real-time voice agents?

Languages

From your city to Timbuktu,
we hear you.

Languages

From your city to Timbuktu,
we hear you.