• AI Academy
  • Posts
  • 📚Turn Your Spotify Playlist Into Language Lessons

📚Turn Your Spotify Playlist Into Language Lessons

How I use AI to create French quizzes from song lyrics

Reading Time: 4 minutes

Hello AI Enthusiast,

What if every song you added to your playlist automatically became a personalized language lesson?

I am Helin, Partnerships Lead of AI Academy and I joined the last edition of our bootcamp to build a passion project, "Mot à Mot" which transforms my French Spotify playlist into interactive, bite-sized quizzes delivered straight to my WhatsApp in seconds.

The Problem

Grammar books, language apps, and traditional learning materials are absolutely great and helpful but there are different learning types, and we all absorb information differently. As a contextual learner, I find that learning song lyrics is one of the most effective ways for me to acquire a new language. Songs are packed with cultural meaning, slang, and emotional nuance which make them the perfect condensed data set for a contextual learner, but there is no structured way for me to do that other than tons of karaoke.

How I Built It: My Product Flow

Before diving directly into building, it's important to spend time understanding what a good and efficient flow might look like. The AI coaches during our bootcamp help everyone identify the starting point, as they have extensive experience using different tools.

This guidance helps tremendously because playing around with automations is a bit like playing The Sims, there's no perfect/complete house, you can always build better, bigger, more cost-efficient solutions and re-iterate. But the goal here is to get you comfortable and get you started using tools.

The flow I'm sharing below is the first completed version of my automation:

✅ Song added to playlist → triggers Make (SPOTIFY)

🎼 Scrapes song title, artist, and real lyrics (APIFY)

🧹 Cleans & segments the text (OPENAI)

📚 Logs lyrics (Google Sheets)

🌍 Detects language, only continues if it's French (OPENAI)

✍️ Generates a fill-in-the-blank quiz (OPENAI)

📚 Logs questions (Google Sheets)

📤 Sends quiz via WhatsApp (TWILIO)

Step 1: Trigger

I used Make.com to watch my Spotify playlist for new additions. Every time I add a song to my French playlist, the automation kicks in automatically. This means I can discover French music naturally and know that each song will become a lesson without any extra effort on my part.

Step 2: Extract Song Information

When a new song is detected, the system connects with Apify API, which helps me connect Genius and automatically scrapes the song title, artist, and most importantly, the actual lyrics from online sources and checks if they are in the target language.

Step 3: Clean and Process Lyrics with AI

This is where the magic happens. I use AI to clean up the raw lyrics and prepare them for learning. What does this mean exactly?

When we are learning contextually from quizzes, I need to be careful. For example, if in an Italian song it says "In bocca al lupo," I need to understand that it doesn't literally mean "mouth of the wolf" but actually means "good luck." AI is already quite good at this but I needed my prompt to capture all these cultural meanings to split the quizz-bits accordingly.

Detailed prompt I use:

"You will receive raw French song lyrics as a single text input. Your task is to clean and segment them into a list of standalone, coherent lines for language learning.

Follow these steps precisely:
1. Remove all section labels (e.g. [Refrain], [Couplet 1], [Pont], [Instrumental], etc.), including malformed or non-standard ones (like (Refrain), REPEAT, or empty brackets).
2. Remove empty lines, extra spaces, and any text that is not part of the actual lyrics (e.g., credits, timestamps, note markers).
3. Preserve the order of original lines.
4. Split the lyrics into individual lines or complete sentences. A line should:
   * Contain a coherent unit of meaning.
   * Be less than 25 words (if longer, break into shorter logical segments).
5. Group the output into **meaningful two-line pairs**. Each group should contain two lines that form a coherent thought or flow together naturally. Insert a blank line between each pair.
6. Output as a plain text list, with one sentence or lyric line per line.
7. Do not translate, add, interpret, or reorder any lines. Only clean and segment.

Example Input:

[Refrain]  
Je veux voir le monde danser  
Je veux sentir la liberté  
[Couplet 1]  
Chaque jour est une chance  
D’écrire notre vérité  
(Instrumental)

Expected Output:

Je veux voir le monde danser  
Je veux sentir la liberté  

Chaque jour est une chance  
D’écrire notre vérité

Important: If the input is blank or contains no usable lyrics, return only: NO LYRICS FOUND
"

Step 4: Generate Interactive Quizzes

Next, I have the AI create fill-in-the-blank exercises from the cleaned lyrics:

Detailed prompt:

Create a fill-in-the-blank quiz from these French lyrics. Select 8-10 key words or phrases that would help a French learner understand the song's meaning. Include a mix of vocabulary levels - some basic words for confidence, some intermediate for challenge. Provide the complete quiz with blanks marked as _____ and include the answer key at the bottom.

Focus on words that are:
- Emotionally significant to the song
- Commonly used in everyday French
- Culturally relevant expressions

Lyrics: [Insert cleaned lyrics]

Step 5: Deliver via WhatsApp

The finished quiz gets sent directly to my WhatsApp, making it easy to practice during commutes or breaks:

WhatsApp message format:

🎵 Song: Je ne sais pas danser  
🧠 Quiz:

1.⁠ ⁠Quand les lumières brûlent __

Step 6: Log Everything for Review

The system saves both the original lyrics and quiz questions to a Google Sheet, creating a personal database of songs I've learned from.

Make workflow that fetches Spotify lyrics and sends them to Google Sheets

The quiz question sent via WhatsApp using Twilio

What I Learned Building This

Creating Mot à Mot taught me several valuable lessons:

Don't be afraid to try new tools: This project required web scraping, and connecting tools like Apify & Twilio & Spotify which I'd never done before. It turned out to be a lot simpler than I thought.

Consider the broader implications first: I initially focused on the technical challenge without thinking about copyright considerations! This made me look for alternative solutions and update my flow.

Be flexible in your thinking: I have changed where to store the data, how to display the quizzes etc. multiple times after having calls with our coaches. It is good to have a solid flow but being open to feedback takes the project further, quicker.

Ask for help: When help is available, sometimes all you need is to ask. As much as I enjoyed being stuck in a part of the automation for hours, I set a call with a coach for each issue I had and asked feedback about "this is how I think the issue might be and this is how I'm solving it." A lot of the solution can come up during that conversation which makes life easier.

Save your work: Hoping auto save will be implemented in no-code tools soon, but until then, Just do it!

Your Turn

Ready to Build Your Own AI Solution? Helin's Mot à Mot project is exactly the kind of real-world automation you'll create in our AI Agent Bootcamp. Our next cohort starts September 2 with hands-on coaching to help you build working AI systems that solve your actual problems.

Want to get even more practical? Explore hands-on AI learning with AI Academy:

  • AI Academy Membership: Get 12 months of access to all our cohort-based programs, live webinars, on-demand courses, and tutorials.

  • AI Agent Bootcamp: Accelerate processes and solve business problems by mastering prompts and building AI Agents, without coding.

  • Corporate Training: Equip your team with the skills they need to unlock the potential of AI in your business.

  • Practical Introduction to ChatGPT: A free course on using ChatGPT confidently, understanding its workings, and exploring its potential.

We'll be back with more AI tips soon!