Imported from jeremylongshore/tons-of-skills-marketplace (plugins/saas-packs/assemblyai-pack/skills/assemblyai-sdk-patterns/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill assemblyai-sdk-patterns. Copyright stays with the author (MIT).

AssemblyAI SDK Patterns

Overview

Production-ready patterns for the assemblyai npm package covering client initialization, type-safe wrappers, error handling, and multi-tenant architectures.

Prerequisites

• assemblyai package installed (npm install assemblyai)
• Familiarity with async/await and TypeScript generics

Instructions

Step 1: Type-Safe Singleton Client

// src/assemblyai/client.ts
import { AssemblyAI } from 'assemblyai';

let instance: AssemblyAI | null = null;

export function getAssemblyAI(): AssemblyAI {
  if (!instance) {
    const apiKey = process.env.ASSEMBLYAI_API_KEY;
    if (!apiKey) throw new Error('ASSEMBLYAI_API_KEY is required');
    instance = new AssemblyAI({ apiKey });
  }
  return instance;
}

Step 2: Transcription Service Wrapper

// src/assemblyai/transcription-service.ts
import { AssemblyAI, type Transcript, type TranscriptParams } from 'assemblyai';

export interface TranscriptionResult {
  id: string;
  text: string;
  duration: number;
  words: Array<{ text: string; start: number; end: number; confidence: number }>;
  speakers?: Array<{ speaker: string; text: string }>;
}

export class TranscriptionService {
  constructor(private client: AssemblyAI) {}

  async transcribe(
    audio: string,
    options: Partial<TranscriptParams> = {}
  ): Promise<TranscriptionResult> {
    const transcript = await this.client.transcripts.transcribe({
      audio,
      ...options,
    });

    if (transcript.status === 'error') {
      throw new Error(`Transcription failed: ${transcript.error}`);
    }

    return {
      id: transcript.id,
      text: transcript.text ?? '',
      duration: transcript.audio_duration ?? 0,
      words: (transcript.words ?? []).map(w => ({
        text: w.text,
        start: w.start,
        end: w.end,
        confidence: w.confidence,
      })),
      speakers: transcript.utterances?.map(u => ({
        speaker: u.speaker,
        text: u.text,
      })),
    };
  }

  async getTranscript(id: string): Promise<Transcript> {
    return this.client.transcripts.get(id);
  }

  async listTranscripts(params?: { limit?: number; status?: string }) {
    const page = await this.client.transcripts.list(params);
    return page.transcripts;
  }

  async deleteTranscript(id: string): Promise<void> {
    await this.client.transcripts.delete(id);
  }
}

Step 3: Error Handling Wrapper

// src/assemblyai/errors.ts

export class AssemblyAIServiceError extends Error {
  constructor(
    message: string,
    public readonly statusCode?: number,
    public readonly retryable: boolean = false,
    public readonly transcriptId?: string
  ) {
    super(message);
    this.name = 'AssemblyAIServiceError';
  }
}

export async function safeTranscribe<T>(
  operation: () => Promise<T>
): Promise<{ data: T | null; error: AssemblyAIServiceError | null }> {
  try {
    const data = await operation();
    return { data, error: null };
  } catch (err: any) {
    const statusCode = err.status ?? err.statusCode;
    const retryable = statusCode === 429 || (statusCode >= 500 && statusCode < 600);

    return {
      data: null,
      error: new AssemblyAIServiceError(
        err.message ?? 'Unknown AssemblyAI error',
        statusCode,
        retryable
      ),
    };
  }
}

// Usage
const { data, error } = await safeTranscribe(() =>
  client.transcripts.transcribe({ audio: audioUrl })
);
if (error?.retryable) {
  // Implement retry logic
}

Step 4: Retry with Exponential Backoff

export async function withRetry<T>(
  operation: () => Promise<T>,
  maxRetries = 3,
  baseDelayMs = 1000
): Promise<T> {
  for (let attempt = 0; attempt <= maxRetries; attempt++) {
    try {
      return await operation();
    } catch (err: any) {
      if (attempt === maxRetries) throw err;

      const status = err.status ?? err.statusCode;
      // Only retry on rate limits (429) and server errors (5xx)
      if (status && status !== 429 && (status < 500 || status >= 600)) throw err;

      const delay = baseDelayMs * Math.pow(2, attempt) + Math.random() * 500;
      console.warn(`Retry ${attempt + 1}/${maxRetries} in ${delay.toFixed(0)}ms`);
      await new Promise(r => setTimeout(r, delay));
    }
  }
  throw new Error('Unreachable');
}

// Usage
const transcript = await withRetry(() =>
  client.transcripts.transcribe({ audio: audioUrl })
);

Step 5: Multi-Tenant Client Factory

// For apps serving multiple customers with their own API keys
const clients = new Map<string, AssemblyAI>();

export function getClientForTenant(tenantId: string): AssemblyAI {
  if (!clients.has(tenantId)) {
    const apiKey = getTenantApiKey(tenantId); // from your secrets store
    clients.set(tenantId, new AssemblyAI({ apiKey }));
  }
  return clients.get(tenantId)!;
}

Python Patterns

import assemblyai as aai
from dataclasses import dataclass
from typing import Optional

aai.settings.api_key = os.environ["ASSEMBLYAI_API_KEY"]

@dataclass
class TranscriptionResult:
    id: str
    text: str
    duration: float
    status: str

def transcribe_audio(audio_url: str, speaker_labels: bool = False) -> TranscriptionResult:
    config = aai.TranscriptionConfig(speaker_labels=speaker_labels)
    transcriber = aai.Transcriber()
    transcript = transcriber.transcribe(audio_url, config=config)

    if transcript.status == aai.TranscriptStatus.error:
        raise RuntimeError(f"Transcription failed: {transcript.error}")

    return TranscriptionResult(
        id=transcript.id,
        text=transcript.text,
        duration=transcript.audio_duration,
        status=transcript.status.value,
    )

Output

• Type-safe singleton client with environment validation
• Transcription service wrapper with clean return types
• Error handling wrapper that classifies retryable vs non-retryable errors
• Exponential backoff with jitter for rate limits
• Multi-tenant client factory pattern

Examples

For a tenant request, resolve the tenant from the authenticated server-side identity, obtain its key only from the approved secret store, submit a consented audio object, and retain the opaque transcript ID plus a redacted status. Do not cache credentials or transcript text in a process-global client without tenant lifecycle and revocation controls.

Error Handling

Resources

• AssemblyAI Node SDK Reference
• AssemblyAI TypeScript Types
• AssemblyAI API Reference

Next Steps

Apply patterns in assemblyai-core-workflow-a (async transcription) and assemblyai-core-workflow-b (real-time streaming).