Getting Started

Shunyalabs Speech-to-Text

A high-accuracy, low-latency speech recognition API built specifically for Indian languages. Transcribe pre-recorded audio or stream live audio over WebSocket — with built-in NLP including diarization, sentiment, and intent detection.

Choose your path

Before you start, decide which transcription mode fits your use case.

Pre-recorded

Batch transcription

Send an audio file, buffer, or remote URL via HTTP POST and receive a full structured transcript. Best for offline audio processing.

call recordingspodcastsvideo filesvoice notes

Get started →

Streaming

Real-time streaming

Open a persistent WebSocket connection and receive transcripts as audio is spoken. Partial results arrive within milliseconds.

voice agentslive captionsIVRreal-time assist

Get started →

Key differentiators

Indic-first accuracy

Purpose-built model (zero-indic) with significantly lower WER on Hindi, Telugu, Kannada, Bengali, and more.

Low-latency streaming

Partial transcripts emitted in real time. Configurable chunk size and silence threshold for latency tuning.

Built-in NLP

Diarization, sentiment, intent, emotion, summarization, translation — all in one API call, no separate pipeline.

Supported languages

All 55 languages run on the zero-indic model. Use language_code="auto" for automatic detection.

Language	language_code	Model	Status
Ahirani	ahr	zero-indic	Available
Assamese	as	zero-indic	Available
Awadhi	awa	zero-indic	Available
Bagheli	bfy	zero-indic	Available
Bagri	bgq	zero-indic	Available
Banjari	bwq	zero-indic	Available
Bengali	bn	zero-indic	Available
Bhili	bhb	zero-indic	Available
Bhojpuri	bho	zero-indic	Available
Bodo	brx	zero-indic	Available
Braj	bra	zero-indic	Available
Bundeli	bns	zero-indic	Available
Chhattisgarhi	hne	zero-indic	Available
Dogri	doi	zero-indic	Available
English	en	zero-indic	Available
Garhwali	gbm	zero-indic	Available
Garo	grt	zero-indic	Available
Gujarati	gu	zero-indic	Available
Harouti	hoj	zero-indic	Available
Haryanvi	bgc	zero-indic	Available
Hindi	hi	zero-indic	Available
Kachchhi	kfr	zero-indic	Available
Kangri	xnr	zero-indic	Available
Kannada	kn	zero-indic	Available
Kashmiri	ks	zero-indic	Available
Khortha	ktk	zero-indic	Available
Kodava	kfa	zero-indic	Available
Konkani	kok	zero-indic	Available
Kumaoni	kfy	zero-indic	Available
Kurukh	kru	zero-indic	Available
Lambadi	lmn	zero-indic	Available
Magahi	mag	zero-indic	Available
Maithili	mai	zero-indic	Available
Malayalam	ml	zero-indic	Available
Manipuri	mni	zero-indic	Available
Marathi	mr	zero-indic	Available
Marwadi	mwr	zero-indic	Available
Meitei	mni	zero-indic	Available
Mewari	mtr	zero-indic	Available
Nepali	ne	zero-indic	Available
Nimadi	noe	zero-indic	Available
Odia	or	zero-indic	Available
Pahari Mahasui	him	zero-indic	Available
Punjabi	pa	zero-indic	Available
Rajasthani	raj	zero-indic	Available
Sambalpuri	spv	zero-indic	Available
Sanskrit	sa	zero-indic	Available
Santali	sat	zero-indic	Available
Sindhi	sd	zero-indic	Available
Surgujia	sgj	zero-indic	Available
Tamil	ta	zero-indic	Available
Telugu	te	zero-indic	Available
Tulu	tcy	zero-indic	Available
Urdu	ur	zero-indic	Available
Wagdi	wbr	zero-indic	Available

Minimal example

python

from shunyalabs import AsyncShunyaClient
from shunyalabs.asr import TranscriptionConfig

async with AsyncShunyaClient() as client:
    result = await client.asr.transcribe(
        "audio.wav",
        config=TranscriptionConfig(model="zero-indic"),
    )
    print(result.text)
    # "नमस्ते, आज मौसम बहुत अच्छा है।"

Environment variable:

Set SHUNYALABS_API_KEY and omit api_key=. The SDK picks it up automatically.

How the API works

—	Pre-recorded (Batch)	Streaming
Transport	HTTP POST	WebSocket
Input	File, bytes, or URL	Raw PCM audio chunks
Response	Single structured response	Event stream (PARTIAL → FINAL)
Config object	TranscriptionConfig	StreamingConfig
NLP features	All features available	Core transcription only
Best for	Recordings, files, pipelines	Voice agents, live captions, IVR

Next steps

Setup

Authentication & API Keys →

Get your API key and configure the SDK.

Quickstart

Pre-recorded Audio →

Transcribe your first audio file end-to-end.

Quickstart

Streaming Audio →

Open your first WebSocket session.

Reference

Models & Languages →

Full language code table and model details.

Next Authentication & API Keys