Speak Install & Auth Setup Examples
TypeScript Setup with Speech Recognition
import { SpeakClient, SpeechRecognizer } from '@speak/language-sdk';
const client = new SpeakClient({
apiKey: process.env.SPEAK_API_KEY!,
appId: process.env.SPEAK_APP_ID!,
language: 'es',
});
const recognizer = new SpeechRecognizer(client, {
onSpeechResult: (result) => {
console.log('User said:', result.transcript);
console.log('Pronunciation score:', result.pronunciationScore);
},
onError: (error) => console.error('Speech error:', error),
});Python Setup
import os
from speak_sdk import SpeakClient, LessonSession
client = SpeakClient(
api_key=os.environ.get('SPEAK_API_KEY'),
app_id=os.environ.get('SPEAK_APP_ID'),
language='ja' # Japanese
)
# Verify connection
status = client.health.check()
print(f"Connected: {status.healthy}")Alternative: OpenAI-Compatible API
Speak uses the OpenAI real-time API under the hood. Install the OpenAI SDK as an alternative:
npm install openaiSupported Languages
| Language | Code | Status |
|---|---|---|
| English | en | Available |
| Spanish | es | Available |
| French | fr | Available |
| German | de | Available |
| Portuguese (BR) | pt-BR | Available |
| Korean | ko | Available |
| Japanese | ja | Available |
| Mandarin (Traditional) | zh-TW | Available |
| Mandarin (Simplified) | zh-CN | Available |
| Indonesian | id | Available |
Token Expiration
- Default token validity: depends on API key type
- Implement token refresh for long-running sessions
- Store credentials securely using environment variables or a secrets manager