Gripes With Music Streaming Apps
Update (2025.7.21): Additional gripes:
Romanization, English name & hard-to-input characters: If you listen to music from other places in the world where people uses letters with diacritics, you'd want to be able to search through your music collection by typing names in Latin letters w/ no diacritics, because that's what's available in most (if not all) keyboards and that's what you can almost always be sure to be available. Should your prompt of "Motorhead" be able to find "Motörhead", and should your prompt of "Maneskin" be able to find "Måneskin"? If you really care about these kind of things then the answer is obviously "absolutely yes", but then you'd have to deal with the problem of base character vs. standalone character w/ diacritics vs. base character + Unicode combining diacritics, the latter two of which you can't directly do a codepoint-to-codepoint compare with.
If you deal with music where they don't use the Latin script, things would be even more confusing. You might be tempted to say "well just store their English name then!"… if only things are this simple. For example, each Japanese band which uses Japanese characters for their name and has somewhat of a following overseas would obviously have at least 2 ways of referring them: original Japanese name and the romanization of that Japanese name (e.g. 人間椅子/Ningen Isu), and sometimes they'd have official or quasi-official English name (e.g. Tokyo Incidents/東京事変/Tōkyō Jihen), and sometimes that English name is like the romanization but only with slight differences (e.g. Ling Tosite Sigure/凛として時雨/Rin Toshite Shigure), and not to mention that Japanese have at least two different ways of romanization (Kunrei vs. Hepburn). As someone who do know Japanese, Japanese names feel much more natural and thus the more comfortable option for me is to store the Japanese name and be able to search w/ romanization instead of having to additionally memorize how the English name goes, but since Japanese has this thing that is Kunyomi vs. Onyomi and sometimes they just straight up don't respect any of the two and come up with their own pronunciations, you don't necessarily know how a kanji is pronounced within a name, so sometimes people would resort to searching by typing out the kanji within the name directly, and if the user is e.g. Chinese, they'll likely type those characters with a Chinese IME, and despite the Unicode people's best effort at Unihan, sometimes you'd have characters which are conceptually "the same" but being assigned different codepoints since they have (slightly) different shapes which means that despite you using the same character there's no guarantee if you simply compare the codepoint, and even if they choose to type traditional Chinese characters, they have no guarantee that what they would've typed would directly match the kanjis in Japanese, because for some of the characters the Japanese have their own simplification that is different from the one the PRC uses and probably have a different codepoint because of it.
Every single one of these problems takes a heap amount of effort to solve despite being very tiny detail-only problems, but when they occur they stick out like the biggest sore thumb you'd ever come across. It's actually insane.
Seamless transition between tracks
Some apps don't do this. I have yet to find a Chinese music app that implements this (apparently someone took this complaint to NetEase, and they promptly did not give a fuck; he was understandably furious.). Spotify do have seamless transition, but when it decides it should unleash its ads on you it's at least a 30 second seam.
Seamless transition is vital not just in the context of progressive rock (and the long one in Abbey Road by The Beatles). Many J-Pop & J-Rock albums adopted the style of separating the intro of the first track from the main part; I can tell you two of such examples right off my mind:
- 極彩色 by Reol; you can see there's an intro track titled "-nil-" and the title track "極彩色", but they are obviously intended to be one single entity as shown by the official music video of the title track;.
- 虚言NEUROSE by MY FIRST STORY, which were split into two tracks titled "Monolougue" (sic) and "虚言NEUROSE" in the original album but are combined as one single track in their latest compilation.
Speaking of long progressive rock tracks with multiple sections…
One track vs. Multiple track
It's currently fixed on YouTube, but there has been a long-standing (10+ years) problem about Amon Düül II's track "Burning Sister" on many Chinese music apps. The track is from their second album "Yeti", and to understand this problem we should see the album's listing:
- "Soap Shop Rock" 13:45
- "Burning Sister" 3:45
- "Halluzination Guillotine" 3:10
- "Gulp a Sonata" 0:46
- "Flesh-Coloured Anti-Aircraft Alarm" 6:04
- "She Came Through the Chimney" 2:57
- "Archangels Thunderbird" 3:33
- "Cerberus" 4:21
- "The Return of Rübezahl" 1:41
- "Eye-Shaking King" 5:40
- "Pale Gallery" 2:12
- "Yeti (Improvisation)" 18:09
- "Yeti Talks to Yogi (Improvisation)" 6:18
- "Sandoz in the Rain (Improvisation)" 8:59
The first track, "Soap Shop Rock", is composed of multiple parts, as is many long epics produced by their progressive rock peers of the time. There can be two major forms of errors:
- Each part of Soap Shop Rock is split (possible for CD reissues), but the track listing in the music database is not; this makes the title "Soap Shop Rock" refers to the first 3m45s (i.e. "Burning Sisters"), the title "She Came Through the Chimney" refers to the following 3m10s (i.e. "Halluzination Guillotine"), etc.. Depending on the actual implementation, a few tracks is going to be missing, and most of the tracks would have the wrong name. Xiami famously labelled "Cerberus" as "Pale Gallery" due to this error, and most Chinese music apps do the same thing because they seemingly copy the data of each other.
- Each part of Soap Shop Rock isn't split, but the listing is; this would end up in the situation where "Burning Sister" has the full length of 13m45s, "She Came Through the Chimney" being mislabelled as "Halluzination Guillotine" and etc..
The fact that all music player I've seen till this day assumes "each track can only have one name and it's its canonical name" also didn't help: should you be able to see that it's a part of "Soap Shop Rock" when playing "Burning Sister"? I'd say you should. This is not just a progressive rock thing; classical music can benefit from having a "series name" as well: one of the common convention about classical music CDs is that multiple works are often thrown together because one single work is often not enough to fill the length allowed by the format (e.g. this Kapustin record is titled "Eight Concert Etudes" when it also contains his Sonata Fantasy Op. 39 split by movements.)
Name clash & name split
Have you ever had this happened to you?:
A: Hey B! How are you doing?
B: I'm doing wonderful! You know I've just discovered this band called Nirvana, they have some really great music!
A: Oh wow, you love Nirvana too? Have you listened to "Lounge Act"? The bass line is really killing it!
B: No I didn't! Oh I'll have to check that out when I get home! But have you listened to "Wings of Love"? The instruments is really great.
A: No I didn't! Oh I'll have to check that out when I get home!
(…After they both went home)
A: Wait, is this also what they did?? This ain't grunge rock??
B: Wait, is this also what they did?? This ain't psychedelic pop??
Me neither, but I knew at a very early stage that there's two bands both of which called "Nirvana", because the music streaming app I used at the time gives me the result from both. The problem of "two (or more) different artists with the same name being lumped together as a single entry" happens in every major music software I've used: QQMusic, Xiami (now defunct - you'll be sorely missed because you arguably had the best recommendation algorithm among them all), NetEase Music, Spotify, and those YouTube "Topic" channels. Most cases it's artists that are obscure outside of their own respective niches but managed to put out enough work to be picked up by web scrapers; in the more recent times it's often DJs and those who make easily-made raps and/or electronic music that happen to use the same name as some other previously existing band which they definitely have no chance of knowing (or in the case that they do know, I'd imagine most of them simply don't care), since their convention is to put absolutely every single track as an independent release. For example, the YouTube Topic page for Ambitious Lovers listed music from both the music project by Joel Saunders and the band by the much more famous Arto Lindsay, despite they have very different styles and were active in very different time frames.
Speaking of Ambitious Lovers, YouTube also sometimes treat different names of the same band as different topics: there's a separate Topic page titled "Arto Lindsay & The Ambitious Lovers" listing only one out of their three released albums, despite being obviously the same band. You can argue that YouTube is correct by separating Vektroid/MACINTOSH PLUS and James Chance & The Contortions/James White & The Blacks (which is arguably very ambiguous because they're pretty much the same band lead by the same James Chance and the only difference is that one's music is less jarring), but I really don't like how they're treating the Ambitious Lovers case.
Genres should be tags instead of categories
You can argue that classifying music into neat little genre boxes is fool's errand and I would probably agree with you. People have claimed that Imagine Dragons is "alternative rock". As an old-school rock lover I genuinely can't tell which part of their music is supposed to be the "rock" part; it's as "rock" as AWOLNATION (not to bash on AWOLNATION; they have good tracks and they at least have Passion, which counts as rock in my book).
Unfortunately no one ever implements genre labelling with tag semantics (except the possibly-existing sites designed for genuine music lovers), despite there has to be some form of it already existing as a part of their recommendation algorithm. Does Arto Lindsay's solo albums count as Indie Pop or MPB? You'd say it's Indie Pop, but it do have a similar feel to A Mulher do Fim do Mundo (by Elza Soares) and that album got counted as MPB; and if we look at Ivan Lins, whose work also got counted as MPB, it's yet another very different feel. Album-wide genre labelling is even more absurd if you consider "Greed" (by Ambitious Lovers), which contains both absurdly stereotypical 80s pop music (e.g. Love Overlap) and absurdly stereotypical MPB (e.g. Para Nao Contrariar Voce).
The way the artists are handled
In the context of classical music there are two major axes when you ask the question "who made this music": the composer and the performer. The composer part is obvious, the performer less so; but if you care enough, you can tell there is a difference between Itzhak Perlmann's and everyone else's Chaconne (BWV 1004) and between Hilary Hahn's and everyone else's Gavotte en rondeau (BWV 1006). The current convention is that an album, depending on what it is, can be both attributed to a composer (often times with multiple performers) or a specific performer (in which cases the album often only consists of works from a single composer). This is hardly ideal; my personal opinion is that both names should be credited and shown. This is also a thing that happens with popular music, with featuring artists, covers and re-arrangements: should the original track also be featured in the page for a cover? If it should, should the tracks made from sampling other people's songs also contain links to the sampled songs? If all sampling doesn't count because the sampling artists do make artistic transformations of the original, then how should we deal with genres like vaporwave, within which one of the iconic albums comprises of nothing but other people's MIDI files? You see, the user experience design on this front is far from complete; things are how things are now because it's too much effort for too little people that cared.
Last update: 2025.7.18