Spotify Chat Assistant

Conversational music discovery powered by AI

Hybrid AI Mode

👋 Hi! I'm your AI music assistant. Tell me what kind of music you're looking for, and I'll find the perfect recommendations for you!

Try these prompts:

Add existing songs (optional)

Press Enter to send, Shift + Enter for new line

Spotify Chat Assistant

Spotify Chat Assistant - Hybrid Music Recommendation System

Project Summary

One-Line Description:
A hybrid music recommendation system that provides intelligent Spotify song recommendations using AI-powered natural language understanding and traditional genre-based filtering.

What It Does:
Natural language query → AI interprets mood/culture/genre → Personalized Spotify recommendations with quality metrics

Purpose

Problem Solved

Generic playlists - Spotify's recommendations don't match specific cultural or mood preferences
Complex search - Hard to describe what you want ("beautiful nepali songs", "sad bollywood music")
Cultural discovery - Difficult to find authentic regional music
Time-consuming - Browsing through endless playlists

Our Solution

Natural Language Search - Describe what you want in plain English
Cultural Awareness - Accurate language detection (Nepali, Hindi, Korean, etc.)
Hybrid AI Approach - LLM-powered query interpretation + Spotify data
Quality Metrics - Artist diversity, cultural accuracy, freshness analysis
Dual Modes - AI-powered or traditional genre-based search

How It Works

Two Approaches Available

1. Basic Genre-Based Approach (2-Step)

Step 1: Get Artist Genres
   → Input: Artist name
   → Output: List of genres

Step 2: Get Recommendations
   → Input: Artist + Selected genre
   → Output: Similar songs based on genre

2. Hybrid AI Approach (1-Step) Recommended

1. User sends natural language query
   ↓
2. LLM (Groq/Mistral) interprets:
   - Language (Nepali, Hindi, Korean, etc.)
   - Mood (romantic, energetic, sad)
   - Genre preferences
   - Cultural context
   ↓
3. Sequential + Ranking + Embedding algorithms
   ↓
4. Strict language filtering for accuracy
   ↓
5. Returns curated recommendations with metrics

Tech Stack

Category	Technology	Purpose
API Framework	FastAPI	REST API framework
Music Data	Spotify API	Music metadata and streaming
AI/LLM	Groq API	Primary LLM for query interpretation
AI/LLM	Mistral API	Optional LLM for evaluation
Language	Python 3.10+	Core programming language
Algorithms	Sequential + Ranking + Embedding	Hybrid recommendation logic

Use Cases

Use Case	Example Query
Cultural Discovery	"beautiful nepali songs"
Mood-Based	"sad bollywood songs for late night"
Activity-Based	"energetic workout music"
Genre Exploration	"romantic hindi songs"
Language-Specific	"korean pop ballads"
Time-Based	"relaxing evening jazz"

Key Features

1. Hybrid Algorithms

Sequential search strategy
Ranking-based filtering
Embedding-based similarity
Combines multiple approaches for best results

2. Cultural Awareness

Accurate language detection (20+ languages)
Regional music discovery
Cultural context understanding
Strict language filtering

3. LLM Query Interpretation

Natural language understanding
Mood detection (romantic, energetic, sad, relaxing)
Context awareness
Intent classification

4. Quality Metrics

Artist Diversity - Variety of artists in results
Quality Distribution - Premium/Popular/Emerging ratio
Cultural Accuracy - Language match percentage
Average Popularity - Track popularity scores
Freshness - New vs. classic tracks ratio

5. Smart Filtering

Exclude already-known songs
Popularity-based ranking
Duration preferences
Preview availability

6. Performance Optimization

Query caching
Fast response times
Efficient API calls

Response Structure

Hybrid AI Response Includes:

Basic Info:

Query interpretation
Strategy used
Language detected
Genres identified
Mood analysis

Recommendations:

Track name, artists, album
Genre and language
Popularity score
Duration
Preview URL
Spotify link

Quality Metrics:

Artist diversity score
Quality distribution (Premium/Popular/Emerging)
Cultural accuracy percentage
Average popularity
Freshness percentage

Analysis Summary:

Total tracks found
Primary language
Primary genres
Detected mood

Example Queries

Cultural Discovery

"beautiful nepali songs"
"romantic hindi songs"
"korean pop ballads"

Mood-Based

"sad bollywood songs for late night"
"happy upbeat songs"
"relaxing evening music"

Activity-Based

"energetic workout music"
"study focus music"
"party dance songs"

Genre-Specific

"indie folk acoustic"
"classic rock 80s"
"modern jazz fusion"

Workflow

For Basic Approach:

Search for artist genres
Select a genre from list
Get genre-based recommendations

For Hybrid Approach (Recommended):

Send natural language query directly
Get AI-powered recommendations instantly
View quality metrics and analysis

The hybrid approach is more powerful and requires only one API call!