Cantonese Language Tutor

This example demonstrates Fliiq’s text-to-speech, Telegram delivery, and scheduled jobs — a daily Cantonese tutor that teaches phrases with native audio pronunciation.

The Setup

A user wanted to learn conversational Cantonese but couldn’t find good audio resources for the language. They asked Fliiq:

"Set up a daily Cantonese lesson job that runs at 8am. Teach me useful
conversational phrases — greetings, restaurant ordering, directions, small talk.
Send audio pronunciation to Telegram so I can practice listening."

What Fliiq Created

The agent called create_job and wrote this YAML:

name: cantonese-tutor
trigger:
  type: cron
  schedule: "0 12 * * *"       # 12pm UTC = 8am EDT
prompt: |
  You are my Cantonese language tutor. Read your job memory
  to see what phrases we've covered so far and my progress.

  Today, teach me 3-5 new conversational phrases in Cantonese.
  For each phrase, provide:
  - The Cantonese characters
  - Jyutping romanization
  - English meaning
  - A brief usage note

  Pick a theme for today (greetings, food, directions, shopping,
  small talk, etc.) and don't repeat themes from the last 3 days.

  After writing the lesson text, use text_to_speech to generate
  audio of each phrase spoken naturally, then send the lesson
  and audio to Telegram.

  Update your memory with today's phrases and theme.
skills:
  - memory_read
  - memory_write
  - text_to_speech
  - send_telegram
  - send_telegram_audio
  - get_current_time
delivery:
  type: telegram
  chat_id: "12345678"
enabled: true

How It Works

First Lesson

The agent starts with beginner greetings:

🗣️ Cantonese Lesson — Day 1: Greetings

1. 你好 (nei5 hou2) — Hello
   The most common greeting. Works in any situation.

2. 早晨 (zou2 san4) — Good morning
   Used before noon. More formal than 你好.

3. 多謝 (do1 ze6) — Thank you
   Use for gifts or favors received.

4. 唔該 (m4 goi1) — Thank you / Excuse me
   Use for services (waiter, taxi driver). Also means "excuse me."

5. 再見 (zoi3 gin3) — Goodbye
   Standard farewell.

🎧 Audio attached — listen and repeat each phrase!

The agent generates audio using text_to_speech with the Cantonese_GentleLady voice, then sends it via send_telegram_audio. The user gets both the text lesson and a native-sounding audio file in their Telegram chat.

Progressive Lessons

Each day, the agent reads its memory, avoids repeating recent themes, and builds on previous lessons:

🗣️ Cantonese Lesson — Day 4: At a Restaurant

1. 我想要... (ngo5 soeng2 jiu3...) — I want...
   Point at the menu and say this.

2. 幾多錢？ (gei2 do1 cin2?) — How much?
   Essential for markets and restaurants.

3. 好好食 (hou2 hou2 sik6) — Very delicious
   Compliment the chef!

4. 埋單 (maai4 daan1) — Bill please
   Wave to your waiter and say this.

💡 Yesterday's review: Try saying 唔該埋單 (m4 goi1 maai4 daan1)
   — "Excuse me, bill please." Combines Day 1 + Day 4!

Over Weeks

The memory accumulates a full vocabulary log. The agent tracks themes covered, introduces review days, and starts combining phrases into longer sentences:

## Progress Notes
- Week 1: Greetings, Numbers, Food, Directions (20 phrases)
- Week 2: Shopping, Transport, Weather, Review (18 new + 10 review)
- Starting to combine phrases into full sentences
- User requested more restaurant vocabulary — added extra food day

Why Cantonese?

Most language learning apps and translation services have limited Cantonese support — they default to Mandarin. The MiniMax TTS API includes a dedicated Cantonese_GentleLady voice, making Fliiq one of the few tools that can generate natural Cantonese audio on demand. This pattern works for any language — swap the voice and prompt for Mandarin, English, or other supported languages.

Managing the Job

fliiq job list                        # See status and next run
fliiq job logs cantonese-tutor        # View lesson history
fliiq job output cantonese-tutor      # Last lesson text
fliiq job run cantonese-tutor         # Manual trigger

The Key Insight

The combination of text-to-speech + Telegram delivery + per-job memory creates a personalized language tutor that adapts to your progress. Each lesson builds on the last, audio pronunciation lets you practice listening, and Telegram delivery means the lesson arrives on your phone ready to study during your commute. This pattern applies to any language learning scenario: Mandarin tones practice, English pronunciation drills, or vocabulary building in any supported language.

​The Setup

​What Fliiq Created

​How It Works

​First Lesson

​Progressive Lessons

​Over Weeks

​Why Cantonese?

​Managing the Job

​The Key Insight