Search should honour special letters. Ø≠O, A≠Å

[Hestepare]

I speak a language which, along with many other languages, uses a particular set of letters that isn’t used by every language. Say I search for “øst”, that’s a very particular search that I would like to return either hits for “øst” or no hits at all. Let me say right away that if there is a way of searching verbatim I haven’t found it. I’ve tried all the usual suspects – ", ’ and +.

Hits such as “ost”, “lost”, “Shostakovitch” or “ghost” are not at all similar to “øst”. “Øst” means “east”, and “ost”, while a Norwegian word, means “cheese”. “Lost” is not “east”, neither is “Shostakovitch”, and what relevance “ghost” might have to “east” is not immediately clear.

“Bodø”, the Norwegian town, is featured in band names, song names and album names, but is not in any particular way similar to “bodo”, “bono”, “bobo”, or, curiously, “body”. If I’m searching for bodø it’s because I want to find, say, a particular song or band I can’t immediately remember the name of, it’s not because I’m feeling adventurous and might as well listen to some U2.

Likewise, “åre” means “oar”, not the present plural of “to be”, “står” is “standing”, not “celestial object”, “ære” means “honor”, not “oar” or the present plural of “to be”, and so on.

I’d be more happy if Roon returned just a handful of hits or no hits at all. I know what I’m searching for and it seems that Roon has no idea but tries anyway. It seems to be to be easy to fix this, for instance by not equating one letter with another. If I’m searching for “ø”, please return hits for “ø”, not “o”. A and Å are at either end of the alphabet and are not pronounced the same. I’m not a programmer but I would think there were standards for this – surely multi-language searches are nothing new in 2021.

[ilkay_uzun]

The problem is, I am not Norwegian but I may need to search “Bodø” as well. I don’t have that letter on my keyboard. How will I input that into search? Am I missing something?

[Neil_Russell]

The iPad has special symbols built into their keyboard, engaged by holding down the key. Just the “e” key has seven special characters (ė, ę, ē, ê, è, é, and ë). if I keep the “e” depressed, it shows these characters, then I slide my finger to select. I assume it’s the same for. I suppose iPhone, Android and desktops have their own peculiar was of doing it. But I have never tried using them in Roon. If they don’y, they should. If it can, then there should be a tutorial.

[Njal_H (Njål)]

Roon should ideally ‘understand’ special characters, and present these results first, but not only show these results.

[Neil_Russell]

I think Roon needs to improve it’s adapting to people that are of different nationalities, religions, ethnic groups, etc. that have special “needs” (thing’s like special characters!) that a white Anglo-Saxon Protestant like myself might not need. Every user’s experience should be as good as mine. 'Cause I love y’all no matter who you are, or where you’re from.

[Mike-48]

A fairly universal solution might be to search more generally by default (i.e., as now, equating o, ö, ò, ø, etc.) and have a quoted string (e.g., “øst”) search exactly.

Both options are needed, and not just to cater to those using languages with smaller alphabets. Manufacturers’ metadata are wildly inconsistent: sometimes, everything is reduced to the ASCII character set, sometimes not, and sometimes only a subset of “special” characters is retained.

Even with a simple umlaut, “ö” seems more often replaced by “o” than by the correct “oe”. As the OP noted, this sort of thing leads to annoyances and difficulty with searches.

[Jez]

I absolutely agree with the original poster, except to say that these special letters aren’t special! They are his language! All languages should be respected and treated equally.

[Freqout]

Could make it an option in the search settings whether or not to differentiate. That would allow you to toggle that as neded.

[Neil_Russell]

Or, maybe let the user select order of search between special characters, no preference, no special characters, etc.

[Mike-48]

I suggested quotes as a very easy way to toggle this on and off. I would like to see any method used that is easy and doesn’t require many clicks each time. Roon has been somewhat justly criticized for needing too many clicks in places. Many of us who have been operating mice for decades have had to deal with shoulder, wrist, and hand issues from overuse.

[Chris_Rudmin]

I personally would expect that if I search for “Bjork” that Roon would return results for “Björk”. I would also expect that exact matches would be be listed ahead of fuzzy or partial matches.

[Hestepare]

Here’s what a search for the song “låst” by the band No. 4 returns. Search phrase “låst no. 4”:

Screen one. Top result not particularly relevant, but to be fair the band name popped up:

Scrolling down, nothing relevant:

Scrolling more, nothing relevant:

It’s not that the song isn’t there – it’s track 10 on the album “Henda i været” by the band No. 4. I’m using Qobuz if it matters:

Here’s what the same search returns on Spotify:

Here it’s obvious Roon ignores language-specific letters:

[Neil_Russell]

Although I only speak American English, it annoys me that Roon doesn’t adapt to make it easier for you, and anyone who doesn’t have English as their first language, to find what they want. Roon should give you as good of an experience as it gives me.

[Hestepare]

Thank you very much Neil

[Neil_Russell]

You are most welcome, my friend.

[Thomas_Nielsen]

@Mike-48

Even with a simple umlaut, “ö” seems more often replaced by “o” than by the correct “oe”. As the OP noted, this sort of thing leads to annoyances and difficulty with searches.

You are right, i18n/l11n is complicated and a regular wild-west in metadata sources. Incidentally, “Ö” is not pronounced like “Ø” in all languages and thus doesn’t necessarily map to “oe”. Take Turkish or Icelandic, for instance (and we are not even getting into weird quasi-literations like Motörhead or Hüsker Dü ). Ohh, and then we still haven’t touched upon Cyrillic, Greek, Japanese, Arab and Urdu etc. – or transliteration between languages. This is certainly as much a problem in metadata as it is in the software that consumes it and really does suggest a new standardisation in metadata layers.

I do think that @Hestepare’s suggestion to let generics match specials in a search and let specials match their literals is a good start, e.g. have “Motorhead” match “Motörhead” and “Motorhead”, but only let “Motörhead” match “Motörhead”. Just as “pare” would match “pare” and “pære”, and “pære” only “pære”.

[Hestepare]

The translator gets it!

[Hestepare]

For the programmers among us, would this be a huge job to fix?

[Bill_Janssen (Wigwam wool socks now on asymmetrical isolation feet!)]

Programmers: there are good programmers, mediocre programmers, and bad programmers.

Most reasonably capable programmers would be using Unicode throughout their application, which would automatically support the world’s various alphabets. Capabilities are often limited by the inherent capabilities of the toolkits and libraries used, as well; injudicious selection of such can create application shortfalls like this. That could be an additional factor.

[Hestepare]

I was surprised to see this happening, to be honest. Usually search with proper letters (ÆØÅ) goes fine. In other apps, I mean.

Is this Roon or Qobuz that has this problem, do you think?