IntermediateSpeech⏱ 30 min

Speech Recognition Integration

Implement ASR in web applications — audio capture, streaming transcription, file processing, and error handling for all 50+ supported languages.

Prerequisites

◉LocaleNLP API key — get one from the Developer Hub
◉Basic JavaScript / TypeScript knowledge
◉Browser with getUserMedia support (Chrome, Firefox, Safari 14+)

Install the SDK

Install the LocaleNLP JavaScript SDK. It handles auth, audio codec negotiation, and streaming for you.

# npm
npm install @localenlp/sdk

# yarn
yarn add @localenlp/sdk

Python, Go, and Rust SDKs also available — see API Reference.

Initialize the client

Create a client instance with your API key. Keys are scoped per project and rotatable from the dashboard.

import { LocaleNLP } from "@localenlp/sdk";

const client = new LocaleNLP({
  apiKey: process.env.LOCALENLP_KEY,
});

Streaming transcription

Capture microphone audio and stream to the ASR endpoint. Results arrive as partial transcripts during speaking, final on silence.

const stream = await navigator.mediaDevices
  .getUserMedia({ audio: true });

const session = await client.asr.stream({
  lang: "sw-KE", // Swahili (Kenya)
  onPartial: (t) => console.log("partial:", t),
  onFinal: (t) => console.log("final:", t),
});

session.pipe(stream);

File transcription

For batch processing of audio files. Supports WAV, MP3, OGG, FLAC. Files up to 4 hours — larger files are chunked automatically.

const result = await client.asr.transcribeFile({
  file: audioFile, // File | Blob | Buffer
  lang: "yo-NG", // Yoruba (Nigeria)
  diarization: true,
});

console.log(result.transcript);
console.log(result.speakers); // if diarization: true

Error handling

All SDK errors include a machine-readable code. Handle gracefully with fallback to text input when appropriate.

try {
  const result = await client.asr.transcribeFile({ file, lang: "sw-KE" });
} catch (err) {
  if (err.code === "AUDIO_QUALITY") {
    showTextFallback();
  } else if (err.code === "RATE_LIMIT_EXCEEDED") {
    retryWithBackoff();
  }
}

Supported Language Codes

East Africa

sw-KESwahili (Kenya)

sw-TZSwahili (Tanzania)

am-ETAmharic (Ethiopia)

rw-RWKinyarwanda (Rwanda)

West Africa

yo-NGYoruba (Nigeria)

ha-NGHausa (Nigeria)

ig-NGIgbo (Nigeria)

wo-SNWolof (Senegal)

Next steps

Ready to go deeper?

Explore the full ASR endpoint reference for advanced parameters — speaker diarization, dialect hints, streaming vs. batch mode selection.

API Reference →More tutorials