How Google Maps, Spotify, Shazam and More Work | WSJ Tech Behind
The Wall Street Journal・2 minutes read
Google Maps is a widely used platform that collects data from multiple sources to create accurate and updated maps, utilizing technology like photogrammetry and GPS. Spotify's recommendation system blends machine learning and natural language processing to refine music suggestions, with a goal of transparency and fairness while addressing potential biases and inequities.
Insights
- Google Maps has a vast user base of over 1 billion monthly users, collecting data from diverse sources like users, satellites, cars, and camels, integrating it to create accurate and updated maps while utilizing machine learning for precision and predicting travel times.
- Spotify's recommendation system employs collaborative and content-based filtering, dissecting tracks' temporal structure and cultural context, aiming for transparency and fairness, with a focus on mitigating biases, especially for new artists, relying on human editors for recommendations.
Get key ideas from YouTube videos. It’s free
Recent questions
How does Google Maps collect data?
From users, satellites, cars, and camels worldwide.
What technology does Spotify use for recommendations?
Collaborative and content-based filtering algorithms.
How does noise cancelling technology work?
Through passive, active, and adaptive noise reduction.
What technology does Shazam use to identify songs?
Unique audio fingerprints using spectrograms and scatter plots.
What is the Sphere in Las Vegas known for?
Featuring the world's largest high-def screen.
Related videos
dottotech
5 FANTASTIC Google Maps Features You Want To Know About
wpsu
Geospatial Revolution / Episode One
IGotAnOffer: Engineering
Google system design interview: Design Spotify (with ex-Google EM)
Kevin Stratvert
Top 20 Google Maps Tips & Tricks: All the best features you should know!
The Royal Institution
There is No Algorithm for Truth - with Tom Scott
Summary
00:00
"Google Maps: Evolving Mobile Platform with Immersive Data"
- Google Maps has over 1 billion users monthly and has evolved from a desktop application to a mobile platform.
- The map is continuously updated with information from over 150 million contributors worldwide, with 50 million daily updates.
- Google Maps collects data from various sources like users, satellites, cars, and even camels.
- Google Maps uses photogrammetry and GPS data to create accurate maps, starting with satellite and aerial imagery.
- Street view cars equipped with cameras, GPS, and LIDAR sensors capture imagery for Google Maps.
- Google Maps detects changes in the environment to keep maps updated, like new signs or businesses.
- Data from local municipalities, public transport schedules, and businesses is integrated into Google Maps.
- Google Maps uses machine learning and map operators to edit public data for accuracy.
- Google Maps predicts travel times and busyness using historical traffic patterns and user-contributed data.
- Google Maps aims to be more immersive by combining satellite, aerial, and street view imagery with user contributions for a photorealistic view.
15:52
Spotify's Advanced Recommendation Algorithm and Challenges
- In 2014, Spotify acquired the music analytics firm The Echo Nest, blending machine learning and natural language processing to create a database of songs and artists.
- Spotify's recommendation system begins with collaborative filtering, analyzing data patterns to understand which tracks are frequently playlisted together.
- Collaborative filtering creates a map of music and podcasts, with points representing different tracks in Spotify's catalog.
- To enhance recommendations, Spotify incorporates content-based filtering, which analyzes metadata, audio characteristics, and cultural context of tracks.
- Spotify's algorithm dissects each track's temporal structure, cultural context, and metadata to refine recommendations.
- The platform's content-based filtering technology has evolved over the years, incorporating advanced proprietary features.
- Spotify's algorithm aims to mitigate potential biases and inequities, striving for transparency and fairness in its recommendations.
- The platform faces challenges with new artists due to the lack of user data, relying on human editors for recommendations.
- Spotify's audio analysis has faced criticism for potential cultural biases, particularly in labeling musical elements inaccurately.
- Tap-to-pay technology utilizes NFC (near-field communication) technology, with antennas transmitting radio frequencies for secure and quick transactions.
31:59
Advancements in Noise Cancelling and Audio Technology
- Noise cancelling technology can be categorized into passive noise reduction, active noise reduction, and future non-stationary noise cancellation.
- Current noise cancelling algorithms can effectively identify and cancel out predictable, continuous sounds like generator or airplane noise.
- Airpods Pro and other noise cancelling earbuds offer modes like transparency mode to recreate normal environmental sounds.
- Labs are essential to understand human perception for noise cancellation algorithms to provide a continuous experience.
- Adaptive audio, combining noise cancellation with transparency mode, is coming to Apple's airpods Pro for a seamless experience.
- The Sphere in Las Vegas features the world's largest high-def screen, requiring a new camera system for content creation.
- The Big Sky camera system uses a giant fisheye lens and a sensor to capture a wide-angle, high-resolution view for the Sphere screen.
- The Sphere screen consists of 32 LEDs per foot, totaling 16k x 16k resolution, providing an immersive experience.
- Shazam simplifies songs into unique audio fingerprints using spectrograms and scatter plots to match songs quickly and accurately.
- Shazam initially launched in 2002 as a service accessed by dialing 2580 in the UK, evolving over time to improve recognition rates and profitability.
46:51
Shazam: From Struggle to Success
- Shazam initially struggled financially but showed resilience until the App Store launch in 2008, leading to Apple acquiring it in 2018, now a popular free music app with access to Apple Music's vast library of over 100 million songs, providing a unique music discovery experience in daily life.
