From the (Undergrad) Vault: How Spotify’s Use of Deep Learning Algorithms Allows for a Personalised User Experience

NB: This post is unedited from its initial submission aside from separating the paragraphs to be smaller. I did not voluntarily go into debt for this essay just be deleted from my laptop. I will at least have those sources that were hidden behind paywalls that the university paid for be put to good use! Also I would actually recommend looking up deep learning algorithms because it’s a very interesting topic and incredibly relevant to today’s media climate.

Spotify is an on-demand music streaming platform that boasts over 75 million active listeners and tens of millions of songs on its platform (Jacobson 373). To maintain its status as the largest music streaming platform, Spotify aims to personalise users’ experiences (Jacobson 373). Personalisation is an important aspect of digital platforms, as it has set a ‘standard’ of the way that users engage with music on platforms such as Spotify (Chen 1).

Therefore, Spotify puts large investments of its resources into deep learning to tune the platform’s algorithms to continually improve its personalisation for users (Chen 1). Therefore, it is important to understand how personalisation has come to exist on Spotify and its potential to influence user experience.

Gulmatico argues that music consumption patterns can alter because of digital platforms, as the platform provides new song recommendations and pushes favourite performers (1). This is done via a feature known as ‘automatic playlist recommendations’ (Gulmatico 1). Spotify makes use of deep learning, a machine learning that utilises algorithms to explain data, learn from collected data, and then learn how to make decisions/predictions to complete a task to personalise user experience (Gulmatico 2).

Usually, a machine learning task is called a ‘classification,’ which refers to sorting data and making distinctions (Carah 2022). A classification is made by algorithms making a judgement around a range of data that has been collected, which are often called features or variables (Carah 2022). Then, a machine will be ‘trained’ using a ‘training data set,’ where algorithms are ‘tested’ with previously unseen ‘test data’ (Briot 983).

One way in which AI systems are ‘trained’ involve what’s called ‘deep learning’ or ‘deep neural networks’ (Crawford & Paglen 22). Deep learning is dominant in training AIs such as Spotify because it is driven by increases in available data and computer processing power (Crawford & Paglen 22).

Deep learning approaches can be supervised or unsupervised (Carah). These two approaches are used for classification tasks where it is too hard to describe variables (Carah). By making use of these two approaches, digital platforms can create ‘layers’ that add more distinctions for an algorithm to learn (Crawford & Paglen 8).

Spotify makes use of the unsupervised approach, especially for its recommendations and curated playlists (Gulmatico 1-2). This means that not only is Spotify’s algorithm performing tasks such as classification, but it is also deep learning by trying to personalise user experience (Gulmatico 1-2). These processes are shown in Figures 1-3, showcasing how songs are pre-processed, processed, and recommended, how features are drawn out, and how features are viewed by the algorithms that are deep learning.

Figure 1: Jakheliya, Bhumil, et al. “System Architecture.” Using Deep Autoencoders to Improve the Accuracy of Automatic Playlist Generation. 2020. https://doi.org/10.1007/978-3-030-38040-3_71.

Figure 2: Mounika, K. S., et al. “Diversity of Filters Within Layers.” Music Genre Classification Using Deep Learning. 2021. https://doi.org/10.1109/ICAECA52838.2021.9675685.

Figure 3: Mounika, K. S., et al. “CRNNs for Music Classification.” Music Genre Classification Using Deep Learning. 2021. https://doi.org/10.1109/ICAECA52838.2021.9675685.

Briot argues that this process is successful if it contains three aspects: (1) technical progress in which inefficiency in ‘training data’ to optimise deep learning is improved upon continuously, (2) an AI is provided with multiple data sets to tune, and (3) improving efficient computing power (981). With these three aspects being used by a platform such as Spotify, the manual time-consuming creation of making playlists for users is no longer needed on Spotify – enhancing user experience (Jakheliya 626).

Therefore, Spotify offers its users the feature of ‘automatic playlist generation’ systems, which are generated by algorithms that consider users’ engagements with the platform (Jakheliya 626). Jakheliya explains that digital platforms have created this system for their platform using two main approaches: the collaborative approach and the content-based approach (626). S

ystems embedded in the collaborative approach require user data to generate accurate results, while the accuracy of playlists generated using the content-based approach systems instead relies upon features categorised from the platform’s dataset (Jakheliya 627). Spotify, therefore, prefers to utilise the content-based approach (Jakheliya 626).

The next question to ask then is: how does Spotify’s algorithm determine the features of a song in a content-based approach system so that user experience can be personalised (Matera 1)?

Jakheliya in Figure 4 showcases the deep learning model that is to be discussed in relation to Spotify in the following paragraphs (630).

Figure 4: Joshua Gulmatico et al. “System Architecture.” SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022. https://doi.org/10.1109/ICPC2T53885.2022.9776765.

Matera explains that Spotify collects the necessary data for training sets by using its API, which allows deep learning algorithms to access metadata and musical content that Spotify holds as shown in Figure 5 to extract features to turn into classifications (1).

Figure 5: Jakheliya, Bhumil, et al. “Histograms of Features Representing Distribution of Data.” Using Deep Autoencoders to Improve the Accuracy of Automatic Playlist Generation. 2020. https://doi.org/10.1007/978-3-030-38040-3_71.

Metadata is used to link together users’ likes/dislikes and media habits to recommend similar user profiles and similar songs that the algorithm learns to believe users may like, while musical content is classified by labelling genres, artists, and more (Anderson 561).

Spotify’s algorithm labels clusters of artists and determines the acoustic properties of songs (Anderson 561). This is seen in the figure below, which shows that the function of a deep learning model is to “perform content-based filtering on Spotify Song Dataset,” which Jakheliya explains is, implemented with the help of two sub-models: Autoencoder Model and Clustering Model as shown in Figures 6 and 7 (630).

Figure 6: Jakheliya, Bhumil, et al. “Autoencoder Neural Network.” Using Deep Autoencoders to Improve the Accuracy of Automatic Playlist Generation. 2020. https://doi.org/10.1007/978-3-030-38040-3_71.

Figure 7: Jakheliya, Bhumil, et al. “Deep Learning Model (Content-Based).” Using Deep Autoencoders to Improve the Accuracy of Automatic Playlist Generation. 2020. https://doi.org/10.1007/978-3-030-38040-3_71.

Spotify then uses Gracenote, a third-party music metadata service, which determines the ‘mood’ of a song by processing audio signals into features (such as harmony, rhythm, and more) – known as data extraction and pre-processing (Anderson 561; De Quirós 354). Then, the deep learning algorithm will use the features of a song to determine which genre, moods, and more the algorithm should recommend – known as feature extraction (Anderson 561; De Quirós 353-358).

Mounika explains that transferring learning is used to “insert features of specified audios and labels from various data into an allocated space with linear transformations” as seen in Figure 8 (5). This figure shows how transfer learning can classify features to then recommend songs to users in Spotify’s deep learning progress (Mounika 5).

Figure 8: Joshua Gulmatico et al. “Linear Correlated Features of Popularity.” SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022. https://doi.org/10.1109/ICPC2T53885.2022.9776765.

Gulmatico also shows varied categories that data can be sorted into, showing in the figure below how over 170,000 songs from Spotify were collected from the API to extract four classifications from them: primary, numerical, dummy, and categorical as shown in Figures 9-11 (3).

Figure 9: Joshua Gulmatico et al. “Features Under the Dummy Category.” SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022. https://doi.org/10.1109/ICPC2T53885.2022.9776765.

Figure 10: Joshua Gulmatico et al. “Features Under the Numerical Category.” SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022. https://doi.org/10.1109/ICPC2T53885.2022.9776765.

Figure 11: Joshua Gulmatico et al. “Features Under the Categorical Category.” SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022. https://doi.org/10.1109/ICPC2T53885.2022.9776765.

Now that the data is collected, the algorithm can then be trained using data. Figures 12 and 13 below showcase the results of the process Gulmatico explains:

“In this stage, the researcher conducts some pre-processing to get and normalize the data that is needed for K-means when clustering a certain dataset. The development utilizes the Standard Scaler, a module by SK-learn under pre-processing, to normalize certain values on columns.

The range of values for danceability and instrumentality is only roughly 0 to 1. On the other hand, the range for duration and popularity can be in the millions. The data was scaled so that it was homogenous and clusterable. To find the dataset’s real major components, we’ll need to utilize a decomposition method (3)”

Figure 12: Jakheliya, Bhumil, et al. “Box Plot of Tempo Feature.” Using Deep Autoencoders to Improve the Accuracy of Automatic Playlist Generation. 2020. https://doi.org/10.1007/978-3-030-38040-3_71.

Figure 13: Jakheliya, Bhumil, et al. “Box Plot of Loudness Feature.” Using Deep Autoencoders to Improve the Accuracy of Automatic Playlist Generation. 2020. https://doi.org/10.1007/978-3-030-38040-3_71.

A ‘Principal Component Analyser’ module from ‘SK-learn’ is then used to complete the algorithm’s system, allowing two variables to be determined as the ‘principals’ so that the metadata can be clustered again for tuning (Gulmatico). The scatter plot data in Figure 14 showcases how clusters led to their classifications into different features.

Following the pre- processing stage of Spotify’s deep learning, the classifications will be sorted into a hierarchy of ‘important information such as artist and genre which are more weighted than harmony/rhythm or models such as Linear Regression as shown in Figure 15 (Gulmatico 4). Linear Regression predicts if a song will be popular based upon audio features (by answering questions such as “is this song danceable?”) by calculating a total average per audio characteristic (Gulmatico 4).

Figure 14: Joshua Gulmatico et al. “Scatter Plot of 2 Components.” SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022. https://doi.org/10.1109/ICPC2T53885.2022.9776765.

Figure 15: Joshua Gulmatico et al. “Clustering of Data Using Hierarchy Method.” SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022. https://doi.org/10.1109/ICPC2T53885.2022.9776765.

Spotify recommendations, therefore, are based on users’ preferences of features such as genres and artists, and the features of songs that they listen to the most (De Quirós 353- 358). This process also goes through ‘fine-tuning,’ in which optimisers are used to continuously improve the algorithm’s ability to learn and recommend for better-personalised user experience (Mounika 3).

The five steps in improving ‘flow’ are: extracting data, pre- processing data, model development, evaluation, and prediction (Gulmatico 2). Using these five steps the table below shows the process that four distinct models are used in datasets to evaluate the efficiency of the Spotify algorithm’s process (Gulmatico 2).

Gulmatico explains that The System Architecture shown above is a process in which data extraction goes through an efficient application, where data is placed into deep learning training sets/procedures to train the algorithm until it is most optimal (2).

The Spotify algorithm once it has processed metadata turns features into song vectors (De Boom 385). These song vectors are used to predict users ‘tastes,’ and are known as ‘tase vectors (De Boom 385). Taste vectors are an output of a retrieval reference number that aggregates song vectors from users’ metadata to represent their musical habits. Therefore, the taste vectors are used to generate song recommendations by, “querying a tree data structure for the nearby song vectors” (De Boom 385).

Therefore, Briot explains that Spotify’s recommended automatic playlists feature is based upon a recurrent neural network’s deep learning that has processed combinations of “arbitrary embeddings and features” that have been supplied with song vectors to produce a “user taste vector” to personalise user experience (985).

To maintain a positive personalised user experience, this feature, therefore, is in a constant state of tuning, as with each song’s metadata and user metadata that is processed, the deep learning algorithms need to be updated often so that recommendations can immediately reflect users’ current habits – which the recurrent neural networks do to calculate new taste vectors to generate recommendations (Gulmatico 4).

Additionally, about the impact, Spotify’s deep learning algorithms have on users, despite many datasets existing for musical streaming platforms, Gulmatico argues that there is no accurate way currently to determine the impact of Spotify’s deep learning algorithm on users’ musical habits (4). Striphas does however argue that there are three key aspects to current music habits users consistently experience: information, crowd, and algorithm (395).

Striphas explains that culture and music are intertwined, such as how digital platforms are intertwined with our daily lives, and therefore argues that currently, an algorithmic culture is developing – where new songs and popularity of music emerge not from publicness but from exposure/recommendations on digital platforms that are based upon deep learning algorithms (395).

Therefore, users need to understand how algorithms function and influence their music streaming experience to see how algorithms ‘intervene’ with ‘connections’ (Van Der Nagel 83). Connections are made by an algorithm based upon artists, mood, features, and more and algorithms take this metadata and are trained to then recommend music based upon users’ engagement and habits rather than seeking out the music themselves (Van Der Nagel 83).

Van Der Nagel states that because platforms are political (as they mediate between the decisions of owners, researchers, designers, and users that allow for data to be collected and processed), people engaging with platforms such as Spotify should be aware of how their experience is being personalised by an algorithm that’s been trained and continuously is optimising its processes (83; 86). This is especially important because algorithms on digital platforms are a part of everyday life for many people nowadays, as they are suggested music by Spotify’s algorithm (Seaver 1).

Algorithms can ‘hook’ people by providing a personalised experience to encourage more engagement, which provides more metadata to the platforms (Seaver 1). Van Der Nagel and Seaver encourage users to think critically about how algorithms are providing an easy solution to searching for musical playlists/songs, and therefore consider how their data is being used to keep them on the platform via deep learning algorithms (86; 1).

Spotify aims to personalise its users’ experiences by utilising deep learning to tune the platform’s algorithms to continually improve its personalisation for users (Chen 1). Therefore, it is important to understand how personalisation has come to exist on Spotify and its potential to influence user experience (Seaver 1).

Spotify utilises deep learning, a machine learning that allows algorithms to explain data, learn, and recommend songs/playlists to personalise user experience (Gulmatico 2). Recommendations are given by classifying features of songs’ metadata to make a judgement of what songs to recommend to users based on their listening habits (Gulmatico 2). This is made possible because Spotify’s deep learning algorithms are ‘trained’ using a ‘training data set’ using an unsupervised approach to process data (Briot 983).

Works Cited
Anderson, Ian, et al. “Just the Way You Are”: Linking Music Listening on Spotify and

Personality.” Social Psychological & Personality Science, vol. 12, no. 4, 2021, pp.

561–572, https://doi.org/10.1177/1948550620923228.
Briot, JP., Pachet, F. Deep learning for music generation: challenges and directions. Neural

Computers & Applications 32, 981–993 (2020). https://doiorg.ezproxy.library.uq.

edu.au/10.1007/s00521-018-3813-6
Carah, Nicholas. COMU3110 Digital Platforms Seminars 1-6. 2022, University of

Queensland, Saint Lucia. Class lecture.
Chen, Yu-Chia, et al. “Music Mood Classification System for Streaming Platform Analysis

via Deep Learning Based Feature Extraction.” 2021 IEEE International Conference on Consumer Electronics-Taiwan (ICCE-TW), IEEE, 2021, pp. 1–2, https://doi.org/ 10.1109/ICCE-TW52618.2021.9603205.

Crawford, Kate, & Paglen, Trevor. “Excavating AI: The Politics of Training Sets of Machine Learning.” Excavating AI, MLA, 19 Sep. 2019, https://excavating.ai/

De Boom, Cedric, et al. “Large-Scale User Modelling with Recurrent Neural Networks for Music Discovery on Multiple Time Scales.” Multimedia Tools and Applications, vol. 77, no. 12, 2017, pp. 385–407, https://doi.org/10.1007/s11042-017-5121-z.

De Quirós, J. García, et al. “An Automatic Emotion Recognition System for Annotating Spotify’s Songs.” Lecture Notes in Computer Science (including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11877, Springer International Publishing, 2019, pp. 345–362, https://doi.org/10.1007/978-3- 30-33246-4_23.

Gulmatico, Joshua S., et al. “SpotiPred: A Machine Learning Approach Prediction of Spotify Music Popularity by Audio Features.” 2022 Second International Conference on

Power, Control and Computing Technologies (ICPC2T), IEEE, 2022, pp. 1–5,

https://doi.org/10.1109/ICPC2T53885.2022.9776765.
Jacobson, Kurt, et al. “Music Personalization at Spotify.” PROCEEDINGS OF THE 10TH

ACM CONFERENCE ON RECOMMENDER SYSTEMS (RECSYS’16), ACM, 2016,

p. 373, https://doi.org/10.1145/2959100.2959120.
Jakheliya, Bhumil, et al. “Using Deep Autoencoders to Improve the Accuracy of Automatic

Playlist Generation.” INNOVATIVE DATA COMMUNICATION TECHNOLOGIES AND APPLICATION, vol. 46, Springer International Publishing, 2020, pp. 626–636, https://doi.org/10.1007/978-3-030-38040-3_71.

Matera, Matteo. The Music Industry in the Streaming Age: Predicting the Success of a Song on Spotify. ProQuest Dissertations Publishing, 2021.

Mounika, K. S., et al. “Music Genre Classification Using Deep Learning.” 2021 International Conference on Advancements in Electrical, Electronics, Communication, Computing and Automation (ICAECA), IEEE, 2021, pp. 1–7, https://doi.org/10.1109/ICAE CA52838.2021.9675685.

Seaver, N. (2018). Captivating algorithms: recommender systems as traps. Journal of Material Culture, 24(4), pp. 1-16. doi: 0.1177/1359183518820366.

Striphas, T. (2015). Algorithmic culture. European Journal of Cultural Studies, 18(45), 395- 412. https://doi.org/10.1177/1367549415577392

Van Der Nagel, E. (2018). Networks that work too well: intervening in algorithmic connections. Media International Australia, 168(1), pp. 81-92. doi:10.1177/1329878 X18783002.

Stranger Things Book Tag: I Suddenly Cannot Remember Anything I’ve Ever Read

I, like Moya, haven’t seen an episode of Stranger Things but will do my best to answer all these questions! Thank-you for tagging me in this too! 😊

(I also while writing this could not remember any book I have read in my entire life, and had to do some googling of odd phrases like ‘ya novel zombies alive generation’ to find the titles.)

This tag was created by Sarah Elise and you can watch a video they created about it here!

1. Epic Intro: The opening sequence of Stranger Thingsis amazing and really grabs your attention. Name a book that grabbed your attention from the first page.

The School for Good and Evil, by Soman Chainani

“Every four years, two girls are kidnapped from the village of Gavaldon. Legend has it these lost children are sent to the School for Good and Evil, the fabled institution where they become fairy-tale heroes or villains.

Sophie, the most beautiful girl in town, has always dreamed of her place at the School for Good while her friend Agatha, with her dark disposition seems destined for the School for Evil. But when the two are kidnapped they find their fortunes reversed…”

Honestly, this book really grabbed my attention just from the blurb on the back. But the beginning few chapters especially have to be some of my favourites as they hook you in deep. They set up the characters well, their viewpoints, goals and backgrounds. Everything has a layer of mystery to it though which makes it even more interesting, especially as the start is set in the village rather than The School for Good and Evil.

2. Dungeons and Dragons: Name a fantasy world you would like to experience yourself.

Generation Dead, by Daniel Waters

“The phenomenon that’s been sweeping the country seems to be here to stay. Not only are the teenagers who have come back from their graves still here, but newlydeads are being unearthed all the time. While scientists look for answers and politicians take their stands, the undead population of Oakville have banded together in a group they’re calling the Sons of Romero, hoping to find solidarity in segregation.”

This is a romance novel which is usually a hard, left swipe for me, but the story world seemed to interesting to pass up! The story’s set in a world where zombies exist, and are exclusively young people who come back to life hours, days, or weeks after they die. They even become “more human” the more they remember their lives and are shown love by friends, family and others. I’d like to experience myself this world just because of how interesting how everything is handled. There are so many things to consider… Do you mourn until after a certain period just in case they come back? What do you do with teenagers whose parents don’t want them back as zombies? How will people cope if their child is someone who doesn’t return as a zombie? Where should you store bodies then? How can you test humanely on the zombies to learn information, and who do you chose to be in studies?

3. Squad Goals: When Eleven met Mike, Dustin, and Lucas it was a mostly perfect team. Name your favourite bookish group of friends.

Deltora Quest, by Emily Rodda

“Deltora Quest series 1 tells the story of three companions – Leif, Barda and Jasmine – who are on a perilous quest to find thes even lost gems of the Belt of Deltora. Only when the belt is complete will the evil Shadow Lord and his rule of tyranny be overcome.”

Technically this is a series but I will NEVER shut up about how much I loved this series growing up. Leif, Barda and Jasmine are one of my favourite trios in books. They are each such great and interesting character on their own, and their dynamic together is so much fun. They had such a great set-up as a group from the first book too.

4. ABC’s and Christmas Lights: Joyce Byers goes mad with grief after Will goes missing. Name your favourite mentally unhinged character.

Wide Sargasso Sea, by Jean Rhys

“Jean Rhys’s spell-binding novel Wide Sargasso Sea, inspired by Jane Eyre and winner the Royal Society of Literature Award is beautifully repackaged as part of the Penguin Essentials range.

‘There is no looking glass here and I don’t know what I am like now… Now they have taken everything away. What am I doing in this place and who am I?’

If Antoinette Cosway, a spirited Creole heiress, could have foreseen the terrible future that awaited her, she would not have married the young Englishman.”

I like this book’s ‘mad woman’ because her name is Antoinette Cosway, and she is Rhys’ version of Charlotte Bronte’s “madwoman in the attic.” The story tells of Antoinette’s life – from her youth in Jamaica, her unhappy marriage to a certain unnamed English gentleman who we all know who renames her Bertha, says she’s made, takes her to English and isolates her from the rest of the world in his mansion.

5. The Upside Down: Name a book that was the opposite of what you expected.

Flick, by Abigail Tarttelin

“Marooned by a lack of education (and lack of anything better to do), Will Flicker, a.k.a. “Flick,” spends most days pondering the artistry behind being a stoner, whether Pepsi is better than Coke, and how best to get clear of his tiny, one-horse suburb. But Flick senses there’s something else out there waiting for him, and the sign comes in the form of the new girl in town—a confident, unconventionally beautiful girl named Rainbow. As their relationship develops, Flick finds himself torn between the twisted loyalty he feels to his old life and the pull of freedom that Rainbow represents.”

This book was the opposite of what I expected because this is the first book I had read that I absolutely hated. I only read the whole thing out of shire spite and stubbornness. Nothing is actually wrong with it though! I just hated reading it. It was too slow for me, the characters too unlikeable, the main character too asshole-ish and entitled (from him it was due to him mostly spending a lot of the book saying how he wants to get out of his small town, but practically refusing to give up drugs or change pretty any of this attitudes or habits to get there and I didn’t feel satisfied with his character and his not-really-existent-arc at all by the end of the book). I also really did not need to ever read a line in a book that portrayed a teenage boy whose character I did not like at all’s armpit fetish. This is not something I say to kink-shame, it is just that as you can tell I am definitely not turned on by armpits.

6. Mad Sciensts: Dr Brenner likes to get freaky with humanity. Name the freakiest dystopian government you can think of.

Starters, by Lissa Price

“Sixteen-year-old Callie lost her parents when the genocide spore wiped out everyone except those who were vaccinated first–the very young and very old. With no grandparents to claim Callie and her little brother, they go on the run, living as squatters, and fighting off unclaimed renegades who would kill for a cookie. Hope comes via Prime Destinations, run by a mysterious figure known only as The Old Man. He hires teens to rent their bodies to seniors, known as enders, who get to be young again. Callie’s neurochip malfunctions and she wakes up in the life of her rich renter, living in her mansion, driving her cars, even dating Blake, the grandson of a senator. It’s a fairy-tale new life . . . until she uncovers the Body Bank’s horrible plan…”

What freaks me out the most about this society is that children are left to fend for themselves as squatters by the government. People under the age of 20 are called  ‘Starters’ who are under-aged and not allowed to work. People of this, they are left with little to no option but to go to a corporation to let their literal bodies be rented to people over the age of 60 as an illegal way of making money. They are chipped so that seniors inhabit the body during sessions and control the body with only their sub-conscious being there during the session so that they can use the teenagers bodies however they want. Because it’s illegal, people also do not say if they are working for the Body Bank or won’t tell anyone if they are an ‘Ender’ paying to use a teenager’s body.

7. Demogorgon: Name a scary bookish creature that you would notwant to come through your walls.

Pet Semetary, by Stephen King

“When the Creeds move into a beautiful old house in rural Maine, it all seems too good to be true: physician father, beautiful wife, charming little daughter, adorable infant son—and now an idyllic home. As a family, they’ve got it all…right down to the friendly cat.

But the nearby woods hide a blood-chilling truth—more terrifying than death itself…and hideously more powerful.

The Creeds are going to learn that sometimes dead is better.”

I am not the best handler of horror. Pillows, blankets and hands are frequently used to cover my face while watching. The Wendigo I think is something I wouldn’t want to come across because of how unsettling everything that happened in this book was.

I did some googling too, and the Wendigo is a mythological creature or evil spirit who is described as a monster with some human characteristics, or a spirit who has possessed a human and made them look monstrous. It’s said to be able to invoke acts of murder, greed, cannibalism and the cultural taboos against such behaviours. Definitely would not want to come across!

8. Cliff-hanger Ending: Name a book that left you wanting more.

Jatta, by Jenny Hale

“Jatta is an Alteedan princess, but her life is far from carefree. Ever since her mother was killed by werewolves when Jatta was a toddler, the palace has been shadowed by fear. Now, Jatta discovers that the werewolves did not just take her mother – they also stole Jatta’s future. As she is haunted by terrible dreams, it becomes increasingly clear that nothing in her life will ever be the same.

Jatta and her bother must flee the palace and embark on a journey to try to save Alteeda. But each month, at the time of the wolf moon, Jatta becomes dangerous – to herself, to her brother, and to everyone else whose path she crosses. Will she be Alteeda’s saviour – or its ruin?”

This book ending was actually very fulfilling. However, it was so good that I just wanted to read more about all the characters, world and developments of the story.

Spreading the Tag

As this tag is pretty old, I’m not too sure if who has already done it. If you want to do it, go for it even if you haven’t been tagged yet! Or even just if you want to do it again! You can also do it with anime, movies, TV series, whatever you want… Also feel free to not do it if you’re not interested 🖤

Tcrow

Merlin’s Musings

AnimeGoodReads

Twitter 🕷
TikTok  🖤 
YouTube 🎬 
Podcast 🎧

An Updated Pop Vinyl Collection

I have somehow managed to amass even more pop vinyls since last time. I often wonder how, but then remember that the vast majority were gifts or arcade prizes (along with some really cute plushies…) and I am grateful for all of them! Especially since I love decorating my room with things that I love, and merchandise is a big part of that for me.

It’s so hard to pick favourites, but this corner has to have some of them…

Fun fact about the Eeyore: it was a present ordered online for the diamond edition (which is usually glittery black). So when it arrived blue, the person who bought it for me was very confused and about to call the supplier to ask why it was the wrong colour. They then looked it up online, and realised that this was the ‘chase’ version and that the supplier did a thing where you could win it if you ordered online

I’ve had to start budgeting when I go to the second-hand book festival run each year in my city

It’s a bit of a Kingdom Hearts shrine area…

I just really love the Halloween Town designs!

I’m not going to think about Voltron Season 8

I have wayyyyyy too many favourite MHA characters

The collection has grown…

Twitter 🕷
TikTok  🖤 
YouTube 🎬 
Podcast 🎧

The 2020 Releases I’m Most Excited For

Animal Crossing: New Horizons

I love, and will forever love, Animal Crossing. I can’t wait to start playing on it the Switch and be given that golden ticket to a deserted island by the racoon I will forever be indebted to.

Ori and the Will of the Wisps

This is such a gorgeous series, so I can’t wait to stand around for hours just looking at everything.

Attack on Titan Final Season

I love AOT. And watching the anime’s final season is going to ruin me.

Yuri!!! On Ice Movie

Although there is no confirmed date yet for the movie, one can dream that it will be out 2020. BECAUSE I LOVE THIS SERIES WITH ALL MY HEART AND MY LEVEL OF EXCITEMENT FOR IT HAS NOT BEEN DAMPENED SINCE ITS ANNOUCEMENT.

My Hero Academia: Heroes Rising

I love MHA. I know the movie is already out but I haven’t watched it yet so I’m including it in this list. Can’t wait to cry over it, and then read non-stop fanfiction and go on a following fan artists spree these next few weeks.

She-Ra Season 5

Even though nothing is confirmed yet, I am putting it on this list because She-Ra: The Princesses of Power is one of my favourite Netflix series. Everything about it is so good, and I need more in my life.

Dragon Prince Season 4

This is another one of my favourite things I’ve watched on Netflix (and featured on the list despite that the next season has not yet been confirmed). Each season just keeps getting better, and that just makes me hyped for the next!

Umbrella Academy Season 2

Umbrella Academy was sooooooo good! I’m so keen for the next season! It can’t arrive fast enough!

All the Bright Places

This Netflix movie coming out is based upon Jennifer Niven’s book of the same name. I love her YA novels, so I’m interested to see what the movie will be like.

Brooklyn 99 Season 7

Any Brooklyn 99 is good.

Birds of Prey (and the Fantabulous Emancipation of One Harley Quinn)

The Harley Quinn animated series has made it up into my top 3 DC Universe TV series, which makes me excited to then go see Birds of Prey. Plus, Margot Robbie! Harley Quinn!

Wonder Woman 1984

IT’S GOING TO BE AMAZINGGGGGGG.

SCOOB!

It’s Scooby-Doo! Need anything else be told to me to make me more excited?

Twitter 🕷
TikTok  🖤
YouTube 🎬 
Podcast 🎧

 

THE ONE LINER TAG: ASK ME MORE ABOUT THE WAVE, I DARE YOU…

A big thank-you for Keiko for tagging me in this! They write awesome anime reviews, and their weekly round ups focused on seasonal anime are always great! Definitely head over there for some great anime and also music content!

The rules for this tag are as follows:
– Accept and thank your challenger(s) by linking back to their post
– Make a post of one-sentence summaries and/or roasts of at least five books
– No spoilers!
– Link back to the person who tagged you so they can see your post!
– Challenge as many or as little people as you want!
– Have fun!

So then here are the books I’m one-lining!

1. The Wave
   
   Absolute not even worthy enough to be called trash that no-one knows even exists, and for good reason.
2. Paper Towns
   
   Well written, annoying female protagonist, interesting idea about abolishing capitalism in grammar.
3. 13 Reasons Why
   
   Bad.
4. First Life
   
   An interesting idea that shouldn’t have been read like it was a better version of Twilight, but was read as such anyway.
5. Pride and Prejudice
   
   Good though long to read, but very interesting to learn about.

And I’ll tag:
– Rose @ Wretched and Devine
– Ayano @ KawaiiPaperPandas
– Phi @ Anime Athena

Thanks for reading everyone! Hope you have a good day ❤

DELTORA QUEST: WHY DOES NO-ONE EVER TALK ABOUT THIS

This is an amazing series and no-one can convince my otherwise.

Deltora Quest is an anime that was adapted from Emily Rodda’s children’s fiction series of the same title. For those who don’t know the great Australian author that is Emily Rodda, this is her pen name for her children’s books and other series such as the Rondo, Teen Power Inc and Rowan of Rin series. She’s also written The Three Doors trilogy.

Emily Rodda’s real name though is Jennifer Rowe and she writes under this name for her adult crime fiction novels, while she then uses the pseudonym of Mary-Anne Dickinson for her books for much younger children. She’s also had picture books published too.

But enough about my favourite childhood author and more about Deltora Quest!

The books are actually split into three series: the first one which is the title name, the second is Deltora Shadowlands and the final one is Dragons of Deltora. There’s also Tales of Deltora (think a fairy tale book, but for the land of Deltora) and Secrets of Deltora (an engaging book about the history of Deltora with a cute little secret message that you have to decode with hints throughout the book via riddles, pictures and puzzles) but we won’t get into those or the second and third series today.

Naturally, someone realised how brilliant this series was and decided to adapt the first series of Deltora Quest into an anime! (Which I have only watched 4 times by now.) Produced by Genco and SKY Perfect Well Think, there are 65 episodes in the Japanese sub but only 52 in the English sub (don’t worry though, I can vouch that both are good). There’s actually 8 books in the series so the storyline is quite involved, but essentially this is the synopsis (taken from the blurb Emily Rodda has written herself, because neither me nor Wikipedia could write anything even remotely close to this good):

And so as you may have gathered by now… I may or may not have had a tiny hyper fixation with this series. But quite frankly, 8-12 year old Spooky is highly insulted that no-one except for herself has ever mentioned this anime series *huffs with the thousands of arrogance*. So I’m going to write about it to rectify this mistake of society’s. (Though then again I only have been blogging for 10.5 months, and since its original run was released back in 2008 it probably has been but didn’t capture anyone’s attention long enough for it to stick around expect for reruns on ABC3 for 3 years… But I’m just going to violently ignore this as a possibility.)

The story is in-depth. It’s interesting, and you have to stay engaged to fully understand the story. There’s tons of puzzles, riddles and hints about everything to everything throughout the series so you do really need to remember stuff because chances are it will be brought up again later, especially with all the big plot twists that happen. This is by far the stand out feature of the series.

The art is then fairly standard, though that’s still cool and pretty. And the storyline is quite a standard ‘adventure quest’ format too. This tends to translate into the series then being considered a bit of a generic anime, but more so in the way in which you expect certain things to always happen in the shounen genre.

The thing that really helps make the anime nicer though to watch is the characters. They all have great chemistry with each other, and are all diverse and interesting on their own. They are characters with their own skills, opinions and are individuals – which helps make them much more easy to empathise with and connect with too. I also like the character designs in the anime (which have been altered from the book series quite a bit).

Overall:
I will forever hype up this series. I love it because of childhood nostalgia and because of how engaged with the little things and the story events you have to be to be able to link together all the pieces of the puzzle that are the storyline. This series is so well thought out! The characters are well designed too and have great engagements with each other, and although everything else about the anime is pretty standard – it’s still a good, enjoyable watch.

Twitter 🕷
TikTok  🖤 
YouTube 🎬 
Podcast 🎧

5 INTERESTING HARRY POTTER CONSPIRACIES TO TAKE NOTE OF

What do you guys think about these?

1. It’s been theorised that Harry hallucinated the entire series. He was deprived of food by his aunt and uncle and was starving in the cupboard under the stairs, hence the hallucinations. Steve Kloves, a screenwriter for the films, has even said about Harry being in the cupboard, “The point was that he seemed slightly mad… so when Hagrid appeared, you thought he was out of his imagination for a minute.” Rowling then responded to this with, “I think that’s a fabulous point and that speaks so perfectly to the books. Because I’ve heard it suggested to me more than once, that Harry actually did go mad in the cupboard and that everything that happened subsequently was some sort of fantasy life he developed to save himself”

   

2. There is a website that’s been made purely to prove that Draco is a werewolf. Ideas/evidence to support this claim is that:
   – Snape and Malfoy are close because Malfoy is getting Wolfsbane potion from Snape
   – In book 6 when Harry is hiding Borgin and Burkes Draco threatens Borgin and then shows him something on his arm. Harry thinks that what Draco has on his arm must be a Dark Mark, but Draco is not a Death Eater
   – An ongoing arc in book 6 is also that Draco is kind of sick and definitely stressed out
   – Fenrir Greyback is a character who specifically punishes people who’ve messed up by biting their children. And Lucius did stuff up what Voldemort wanted him to do…
   – Voldemort also makes a comment to Draco that he could ‘babysit the cubs’ after they find out that Lupin and Tonks are having a baby

   

3. People often wonder if perhaps Rita Skeeter actually fled the wizarding world after the Battle of Hogwarts. And then here in the muggle world what is she wanted to keep writing, but under a pseudonym? What is she wanted to write about the wizarding world that she was rejected from? Perhaps even a 7 book series?

   

4. This theory from Tumblr user harryjxmespotter suggests that Snape, Voldemort and Harry are the three brothers. Meaning that Dumbledore is Death. Dumbledore greeted Harry at King’s Cross Station and was also the one behind Snape and Voldemort’s death… Voldemort is the oldest brother, murdered by someone who sought the Elder Wand. Snape is the middle brother who committed suicide after resurrecting the girl he had once hoped to marry. And Harry is the youngest brother who escapes Death with an invisibility cloak until he gives it to his son (like how James leaves Harry the cloak, which is given to him by Dumbledore) and then greets Death ‘as an old friend’

   

5. Rowling has revealed that there about 1000 students that attend Hogwarts. This means that there have to be about 35 students in each house year. But there only seems to be 10 Gryffindors in Harry’s year… People have questioned whether this is an oversight made by Rowling, is there are just other students wondering about their own business that don’t need to ever be named, or maybe there’s another reason? Tumblr user marauders4evr writes, “What if there were less students in the Hogwarts Class of 1998 because the period when the other kids would have been conceived (1979-1981) was when Voldemort’s reign of power was at its peak? Between the dozens of adults who joined the Order, the dozens of civilians who were killed in Death Eater raids, and the dozens of adults that didn’t want to bring a child into the world, just then. It’s actually entirely possible that there was a baby drought for a few years in the wizarding world, leading to a smaller class size a decade later. While intriguing and logically plausible, Rowling most likely didn’t think this one all the way through”

   

3 THINGS I LIKE ABOUT WIDE SARGASSO SEA

It’s definitely very interesting to read!

1. The fact it’s based upon Jane Eyre
   Essentially Jean Rhys has rewritten Charlotte Bronte’s Jane Eyre. She stated that, “Bertha seemed such a poor ghost, I thought I’d like to write her a life.” I think that this is a very interesting concept that Rhys has taken and it works so well!
2. Fragmentation is always a good theme to explore
   This novel is told from different points of views (meaning that there really is no narrative authority throughout the book) and is focused upon the inability of Antoinette to form her own identity. There’s so many reasons why Antoinette can’t have subjectivity – such as how her family are kind of like ‘in between two worlds’ in a society where slave owners no longer exist so they don’t fit in the colonies and they also aren’t British so they don’t fit in in England. Antoinette is also a female with no education apart from ‘how to be a good housewife’ and her arranged marriage leads her to a husband who renames her refuses to allow her to have her own identity. Even the fact that despite that this book is about Antoinette, it doesn’t follow Bronte’s title style of naming the novel after the protagonist because Antoinette doesn’t have the some subjectivity that Jane has because she isn’t allowed to develop her own identity
3. It really does explore the question, “In order for Jane from Jane Eyre to have her autonomy, who needs to suffer?”
   In Jane Eyre, Jane gets to end up being a happy, successful, educated, wealthy wife, woman, potentially a mother etc. But Bertha has to die to allow for Jane to have this. So Wide Sargasso Sea tells the story then of Antoinette, which can be considered a critique of the 19th century feminism that was portrayed in literature. This genre allowed the female protagonist to ‘have it all’ at the end of their story, but that meant that another person had to suffer. And Wide Sargasso Sea tells us who had to suffer and what they had to go through for Jane’s story to reach its conclusion. I just think that’s very interesting

10 THOUGHTS AFTER RE-READING ‘THE REMARKABLE SECRET OF AURELIE BONHOFFEN’

I get nostalgic and re-read books I liked when I was younger too often…

“Ever since Aurelie Bonhoffen was a child, she has juggled and played the dead girl on the ghost train at Bonhoffen’s Seaside Pier. On her twelfth birthday, she stumbles on her family’s remarkable secret. It’s hard to accept at first, but when her new friend at school reveals a dangerous plot agains the pier, the secret helps Aurelie confront the greatest threat her family has ever faced.” – Blurb (how do I reference this??? It’s a blurb so that’s what I’m going to call it…) of Deborah Abela’s ‘The Remarkable Secret of Aurelie Bonhoffen’

1. Wow, I forgot how much I loved the characters! They’re so engaging and awesome! The grandma is a treasure. Actually, all the Bonhoffen characters are treasures…
2. The whole ghost part of the story is actually a pretty cool concept (though I just like ghost stories so I’m biased)
3. Aurelie is very relatable. Most of the characters are very relatable… I suppose that’s what makes them so easy to empathise with and why you get so emotionally invested in them!
4. I’m loving the pier setting. Abela has definitely created an immersive world. It’s easy to envision all the settings in the book and although the descriptions do have a bit of an almost ‘fantasical’ approach feel to them, they still are fairly realistic (though I think that magic realism is too strong a word for it). And I think that this approach to the descriptions of settings suits Aurelie very well (she is the narrator after all!)
5. The conflict is a pretty typical one, but it serves its purpose
6. I think that Abela has written a story that is very well suited for its target readers
7. Aurelie’s uncles are the best! Rolo and Rindolf are incredible! They deserve so much praise!
8. And okay, so we were definitely told us the secret too early on in the book. But younger Chloe and present Chloe still just like reading about the characters and their interactions with everyone too much too care all that much
9. This book has some great lessons in it! I’m getting a strong message of ‘acceptance’ from this story so it’s a ‘yes’ from me!
10. I love the Bonhoffen family business and their pier. All the Bonhoffen characters are so talented and although Aurelie grew up in a pretty quirky environment and the less accepting characters are concerned about how this may affect her, the Bonhoffen pier is definitely a great environment to be raised in. Aurelie’s home is filled with many incredible and supportive people! Plus I melt when it comes to families that society likes to deem as ‘strange’ because they don’t necessarily follow every societal expectation (or even pretend to), but are the most loving and caring family one could ever wish for

8 REASONS WHY YOU SHOULD GET INVESTED IN ‘ATTACK ON TITAN: LOST GIRLS’

“Going against the flow takes courage. I respect that” – Annie

1. The storylines follow Annie and Mikasa

   

   Need there be anymore reason?

2. It gives a deeper understanding into the characters (especially Annie since we really didn’t get to see her all that much before she went into the crystal) and lets us learn more about their own stories. This works to kind of do a bit of ‘filling in the gaps’ about the Marleyan Warriors, Annie’s psyche and even a bit of how the Military Police works in Annie’s volume too

   

   Annie’s such a great character

3. But apart from learning a lot more about Annie, it also lets us meet some interesting characters and see a bit more of minor characters (like Hitch and Marlowe). It also introduces some things that happen later in AOT’s timeline – such as Reiner’s identity complex. And although we already knew that Reiner developed this as a way to cope, it makes it more believable and easier to understand his actions when we get to see how it was developing and actually see how it affects him more in different situations

   

   My poor babies have been through so much

4. All three volumes have their own interesting storylines

   

   I’m so glad Isayama decided to write Lost Girls

5. The music in the OVA is great! There’s female covers from some of the OSTs in the anime! 

   

   Call Your Name is a beautiful song and I can’t really decide if I prefer the male or female version, so I’m just going to love them both

6. It lets us all cry together! We can cry over Marco’s death along with Annie, Reiner and Bertholdt as they take Marco’s gear and watch him get eaten. And also cry over how Mikasa’s volume focuses upon her imagining what her world would have been like without Titans (meaning that her parents would still be alive too)

   

   ARE YOU CRYING YET? BECAUSE I AM!! SOMEBODY GET ME A TISSUE BOX PLEASE!!!

7. We get to see Annie’s hair down! And more of small Mikasa and Kenny (for some reason that I myself cannot tell, I do like Kenny’s character)!

   

   I really wanted to see what Annie’s hair looked like down, and my wish has been granted

8. The novels and manga are already completed, so you won’t have to wait forever in between the OVA releases if you don’t want to!

   

   So for those who are impatient like me, you can binge read it!