---

name: speak-reference-architecture description: | Implement Speak reference architecture with best-practice project layout for language learning apps. Use when designing new Speak integrations, reviewing project structure, or establishing architecture standards for language learning applications. Trigger with phrases like "speak architecture", "speak best practices", "speak project structure", "how to organize speak", "speak layout". allowed-tools: Read, Grep version: 1.0.0 license: MIT author: Jeremy Longshore jeremy@intentsolutions.io

Speak Reference Architecture

Overview

Production-ready architecture patterns for Speak language learning integrations.

Prerequisites

• Understanding of layered architecture
• Speak SDK knowledge
• TypeScript project setup
• Testing framework configured

Project Structure

my-speak-app/
├── src/
│   ├── speak/
│   │   ├── client.ts           # Singleton client wrapper
│   │   ├── config.ts           # Environment configuration
│   │   ├── types.ts            # TypeScript types
│   │   ├── errors.ts           # Custom error classes
│   │   └── handlers/
│   │       ├── webhooks.ts     # Webhook handlers
│   │       └── events.ts       # Event processing
│   ├── lessons/
│   │   ├── session.ts          # Session management
│   │   ├── conversation.ts     # Conversation practice
│   │   ├── pronunciation.ts    # Pronunciation training
│   │   └── vocabulary.ts       # Vocabulary exercises
│   ├── speech/
│   │   ├── recognizer.ts       # Speech recognition wrapper
│   │   ├── scorer.ts           # Pronunciation scoring
│   │   └── audio.ts            # Audio processing utilities
│   ├── progress/
│   │   ├── tracker.ts          # Progress tracking
│   │   ├── streaks.ts          # Streak management
│   │   └── achievements.ts     # Gamification
│   ├── services/
│   │   └── speak/
│   │       ├── index.ts        # Service facade
│   │       ├── sync.ts         # Data synchronization
│   │       └── cache.ts        # Caching layer
│   ├── api/
│   │   └── speak/
│   │       ├── lessons.ts      # Lesson endpoints
│   │       ├── progress.ts     # Progress endpoints
│   │       └── webhook.ts      # Webhook endpoint
│   └── jobs/
│       └── speak/
│           ├── sync.ts         # Background sync job
│           └── notifications.ts # Reminder notifications
├── tests/
│   ├── unit/
│   │   └── speak/
│   ├── integration/
│   │   └── speak/
│   └── fixtures/
│       └── audio/              # Test audio files
├── config/
│   ├── speak.development.json
│   ├── speak.staging.json
│   └── speak.production.json
└── docs/
    └── speak/
        ├── SETUP.md
        └── RUNBOOK.md

Layer Architecture

┌─────────────────────────────────────────┐
│             API Layer                    │
│   (Controllers, Routes, Webhooks)        │
├─────────────────────────────────────────┤
│           Lesson Layer                   │
│  (Conversation, Pronunciation, Vocab)    │
├─────────────────────────────────────────┤
│          Speech Layer                    │
│   (Recognition, Scoring, Audio)          │
├─────────────────────────────────────────┤
│          Speak SDK Layer                 │
│   (Client, Types, Error Handling)        │
├─────────────────────────────────────────┤
│         Infrastructure Layer             │
│    (Cache, Queue, Storage, Monitoring)   │
└─────────────────────────────────────────┘

Key Components

Step 1: Client Wrapper

// src/speak/client.ts
import { SpeakClient as SDKClient } from '@speak/language-sdk';

export class SpeakService {
  private client: SDKClient;
  private cache: Cache;
  private monitor: Monitor;

  constructor(config: SpeakConfig) {
    this.client = new SDKClient({
      apiKey: config.apiKey,
      appId: config.appId,
    });
    this.cache = new Cache(config.cacheOptions);
    this.monitor = new Monitor('speak');
  }

  get tutor() {
    return {
      startSession: (config: SessionConfig) =>
        this.monitor.track('tutor.startSession', () =>
          this.client.tutor.startSession(config)
        ),
      getPrompt: (sessionId: string) =>
        this.monitor.track('tutor.getPrompt', () =>
          this.client.tutor.getPrompt(sessionId)
        ),
    };
  }

  get speech() {
    return {
      recognize: (audio: ArrayBuffer) =>
        this.monitor.track('speech.recognize', () =>
          this.client.speech.recognize(audio)
        ),
      score: (audio: ArrayBuffer, text: string) =>
        this.monitor.track('speech.score', () =>
          this.client.speech.score({ audio, expectedText: text })
        ),
    };
  }

  get vocabulary() {
    return {
      lookup: (word: string, language: string) =>
        this.cache.getOrFetch(`vocab:${language}:${word}`, () =>
          this.client.vocabulary.lookup(word, language)
        ),
    };
  }
}

Step 2: Error Boundary

// src/speak/errors.ts
export class SpeakServiceError extends Error {
  constructor(
    message: string,
    public readonly code: string,
    public readonly retryable: boolean,
    public readonly originalError?: Error
  ) {
    super(message);
    this.name = 'SpeakServiceError';
  }

  static fromSDKError(error: unknown): SpeakServiceError {
    if (error instanceof RateLimitError) {
      return new SpeakServiceError(
        'Rate limit exceeded',
        'RATE_LIMITED',
        true,
        error
      );
    }
    if (error instanceof AuthError) {
      return new SpeakServiceError(
        'Authentication failed',
        'AUTH_ERROR',
        false,
        error
      );
    }
    if (error instanceof AudioError) {
      return new SpeakServiceError(
        'Audio processing failed',
        'AUDIO_ERROR',
        true,
        error
      );
    }
    return new SpeakServiceError(
      error instanceof Error ? error.message : 'Unknown error',
      'UNKNOWN',
      true
    );
  }
}

Step 3: Lesson Session Manager

// src/lessons/session.ts
interface ActiveSession {
  id: string;
  userId: string;
  language: string;
  topic: string;
  startedAt: Date;
  exchanges: number;
}

export class LessonSessionManager {
  private sessions: Map<string, ActiveSession> = new Map();
  private speakService: SpeakService;

  constructor(speakService: SpeakService) {
    this.speakService = speakService;
  }

  async startSession(userId: string, config: LessonConfig): Promise<Session> {
    // End existing session if any
    await this.endSession(userId);

    const session = await this.speakService.tutor.startSession({
      language: config.language,
      topic: config.topic,
      difficulty: config.difficulty,
    });

    this.sessions.set(userId, {
      id: session.id,
      userId,
      language: config.language,
      topic: config.topic,
      startedAt: new Date(),
      exchanges: 0,
    });

    // Track session start
    await analytics.track('lesson_started', {
      userId,
      sessionId: session.id,
      ...config,
    });

    return session;
  }

  getActiveSession(userId: string): ActiveSession | null {
    return this.sessions.get(userId) || null;
  }

  async endSession(userId: string): Promise<SessionSummary | null> {
    const active = this.sessions.get(userId);
    if (!active) return null;

    const summary = await this.speakService.tutor.endSession(active.id);
    this.sessions.delete(userId);

    // Track session end
    await analytics.track('lesson_completed', {
      userId,
      sessionId: active.id,
      duration: Date.now() - active.startedAt.getTime(),
      exchanges: active.exchanges,
    });

    return summary;
  }
}

Step 4: Health Check

// src/speak/health.ts
interface SpeakHealthStatus {
  status: 'healthy' | 'degraded' | 'unhealthy';
  latencyMs: number;
  services: {
    api: boolean;
    speech: boolean;
    tutor: boolean;
  };
}

export async function checkSpeakHealth(): Promise<SpeakHealthStatus> {
  const start = Date.now();
  const services = { api: false, speech: false, tutor: false };

  try {
    // Check API
    await speakService.health.check();
    services.api = true;

    // Check speech recognition (with minimal audio)
    services.speech = await testSpeechRecognition();

    // Check tutor availability
    services.tutor = await testTutorAvailability();

    return {
      status: Object.values(services).every(Boolean) ? 'healthy' : 'degraded',
      latencyMs: Date.now() - start,
      services,
    };
  } catch (error) {
    return {
      status: 'unhealthy',
      latencyMs: Date.now() - start,
      services,
    };
  }
}

Data Flow Diagram

User Interaction
     │
     ▼
┌─────────────┐
│   Web/App   │
│   Client    │
└──────┬──────┘
       │ Audio + Text
       ▼
┌─────────────┐    ┌─────────────┐
│   API       │───▶│   Cache     │
│   Gateway   │    │   (Redis)   │
└──────┬──────┘    └─────────────┘
       │
       ▼
┌─────────────┐    ┌─────────────┐
│   Lesson    │───▶│   Audio     │
│   Service   │    │   Storage   │
└──────┬──────┘    └─────────────┘
       │
       ▼
┌─────────────┐
│   Speak     │
│   SDK       │
└──────┬──────┘
       │
       ▼
┌─────────────┐
│   Speak     │
│   API       │
└─────────────┘

Configuration Management

// config/speak.ts
export interface SpeakConfig {
  apiKey: string;
  appId: string;
  environment: 'development' | 'staging' | 'production';
  defaultLanguage: string;
  supportedLanguages: string[];
  cache: {
    enabled: boolean;
    ttlSeconds: number;
  };
  audio: {
    maxDuration: number;
    format: 'wav' | 'mp3';
    sampleRate: number;
  };
}

export function loadSpeakConfig(): SpeakConfig {
  const env = process.env.NODE_ENV || 'development';
  const baseConfig = require('./speak.base.json');
  const envConfig = require(`./speak.${env}.json`);

  return {
    ...baseConfig,
    ...envConfig,
    apiKey: process.env.SPEAK_API_KEY!,
    appId: process.env.SPEAK_APP_ID!,
    environment: env as SpeakConfig['environment'],
  };
}

Output

• Structured project layout
• Client wrapper with caching
• Error boundary implemented
• Health checks configured
• Session management

Error Handling

Issue	Cause	Solution
Circular dependencies	Wrong layering	Separate concerns by layer
Config not loading	Wrong paths	Verify config file locations
Type errors	Missing types	Add Speak types
Test isolation	Shared state	Use dependency injection

Examples

Quick Setup Script

# Create reference structure
mkdir -p src/speak/{handlers}
mkdir -p src/lessons src/speech src/progress
mkdir -p src/services/speak src/api/speak src/jobs/speak
mkdir -p tests/{unit,integration,fixtures}/speak
mkdir -p config docs/speak

touch src/speak/{client,config,types,errors}.ts
touch src/lessons/{session,conversation,pronunciation,vocabulary}.ts
touch src/speech/{recognizer,scorer,audio}.ts

Resources

• Speak SDK Documentation
• Speak Best Practices
• Clean Architecture

Flagship Skills

For multi-environment setup, see speak-multi-env-setup.