Batch Transcription

client.asr.transcribe() sends an audio file and returns the complete transcript once processing is done. Best for pre-recorded files where you don't need real-time output.

---

Signature

client.asr.transcribe(
  file: FileInput,
  options?: TranscribeOptions,
  requestOptions?: RequestOptions
): Promise<TranscriptionResult>

---

Parameters

file — FileInput

The audio to transcribe. Accepts any of:

Type	Example
Buffer	readFileSync('./audio.wav')
Uint8Array	Raw byte array
ArrayBuffer	From fetch response
Blob	Browser File API
File	<input type="file">
ReadableStream	Node.js / Web stream

options — TranscribeOptions

Property	Type	Default	Description
language	Language	auto-detect	ISO code ("en") or full name ("English")
includeScoring	boolean	false	Include per-segment quality score. Requires Professional plan.

requestOptions — RequestOptions

Property	Type	Default	Description
timeout	number	300000	Request timeout in ms (5 min default)
maxRetries	number	2	Override retry count for this request
signal	AbortSignal	—	Cancellation signal

---

Response — TranscriptionResult

{
  jobId: string;
  status: 'completed';
  transcriptionText: string;
  detectedLanguage: string | undefined;
  qualityScore: number | undefined;       // 0–5 MOS score (if includeScoring: true)
  audioMetadata: {
    durationSeconds: number;
    sampleRate: number | undefined;
    channels: number | undefined;
    format: string | undefined;
    bitrate: number | undefined;
  } | undefined;
  processingTimeMs: number | undefined;
  inputUrl: string | undefined;           // Presigned URL to the original file (24h TTL)
  inputUrlExpiresAt: string | undefined;
  inputFileSize: number | undefined;
  createdAt: string;
  completedAt: string | undefined;
}

---

Examples

Basic transcription

import { readFileSync } from 'fs';
import { Decrackle } from '@decrackle/sdk';

const client = new Decrackle({ apiKey: process.env.DECRACKLE_API_KEY });
const audio  = readFileSync('./interview.wav');

const result = await client.asr.transcribe(audio, { language: 'en' });

console.log(result.transcriptionText);
console.log('Duration:', result.audioMetadata?.durationSeconds, 's');

With quality scoring

const result = await client.asr.transcribe(audio, {
  language: 'en',
  includeScoring: true,
});

console.log(`Transcript: ${result.transcriptionText}`);
console.log(`Quality score: ${result.qualityScore}/5`);

From a URL (fetch first)

const response = await fetch('https://example.com/audio.mp3');
const buffer   = Buffer.from(await response.arrayBuffer());

const result = await client.asr.transcribe(buffer);
console.log(result.transcriptionText);

With cancellation

const controller = new AbortController();
setTimeout(() => controller.abort(), 60_000); // cancel after 60s

const result = await client.asr.transcribe(audio, {}, { signal: controller.signal });

---

Errors

Error class	When
DecrackleValidationError	Invalid format, language code, or file too large
DecracklePermissionError	includeScoring requires Professional plan
DecrackleServiceUnavailableError	ASR model temporarily unavailable
DecrackleTimeoutError	Request exceeded the timeout