My first shader

Not sure if I need a dedicated devlog for my game projects (like I already have for my music stuff), but for now I don’t have one and my update pace is super slow anyway… Nevertheless, the 8-bit game of Modus is not dead. I have switched to Godot a while ago and I’m learning it pretty well, albeit slowly. Recently I did my first shader ever. It has 8 different dedicated colour channels and can swap the palette between them. You can have one tileset and the shader does the rest. It’s pretty convenient and boosts my ideas a bit further.

Nextcloud saved my file

This post is not about a big thing but rather a very nice surprise. I have successfully transitioned from Rosegarden to MusE. I still use both but MusE has better support for integrated plugins. Anyway, recently MusE sequencer crashed during my composing session. As it critically crashed during saving my over 10 minute long song, this 1.5MB sized file shrank into 0.3MB, containing only one track. (For a perspective: 1MB programmed music can be compared to approximately 1GB recorded music.) Unfortunately MusE does not seem to do any regular backups. Also both of my computers were open during the incident, so the broken file synced everywhere instantly. Needless to say, I was in a panic for a moment and prepared myself to reprogram the whole song from my memory.

Then I remembered that my cloud service might have some backup options and version history. I have never needed to use it before and I don’t have any sort of commercial plan with my cloud storage provider. I have just chosen a provider that uses open source alternative called Nextcloud as its software. So, I wasn’t sure what to expect. Luckily I found some version control from the browser version of Nextcloud. Then I could retrieve my hours of work that was lost everywhere else. Thank you Nextcloud for saving my ass! This three-part-song will safely appear into my upcoming album now. Open source Linux music making can be a bit hazardous at times, but there are open source solutions to secure your back too.

Transition from GameMaker to Godot

This year I have decided to ditch GameMaker and eventually make a transition into Godot. There are many reasons for this. Support of GM:S1 was abandoned some time ago and GM:S2 does not have any sort of free version anymore. I have been using only Studio 1 so far and waiting if the company would come up with at least Linux native IDE and reasonable pricing policy. GM:S2 has been around for some time and neither of these has been presented. Even though I lack proper programming knowledge other than GML I have no choice but to move into a better engine in order to move my workflow entirely and exclusively in Linux. Therefore, Godot’s free open-source Linux native engine without any DRM restrictions sounds too good for me to not check it out. It is not limited into 2d games either (even though I don’t do 3d).

So far my only problem has been to get fully into Godot. (Enigma project is also quite promising but far from stable.) Godot is a bit more advanced engine compared to GameMaker but I think learning to use it is going to be worth it. Besides, HeartBeast’s YouTube tutorials and this library of equivalent functions are like tailored for a GM-background-person like me to get started with Godot. With an accessible Python-like scripting language I should have some hope at least. This summer I also finally realized how the nodes in Godot are supposed to work. (Things are done quite differently compared to GM, which is only engine I have used to.)

During the university course of Game Project we learned to use Unity which is quite similar to Godot in terms of how UI works etc. However, I did not get a good impression with Unity’s menu-inferno and C#’s a bit complicated syntax. Luckily Godot seems to be way simpler engine to do practically the same thing than Unity does. Unity also recently came up with an official Linux native IDE but with Godot you get immediately in touch with the development itself. Just download a single file and launch it, no installation procedures required. Being so light-weight is one of the reasons that makes using Godot more convenient. (Of course if one is planning to make games in large teams in the gaming industry, Unity is, and probably will remain, an industry standard.)

So far Godot seems to be everything I need. I just need to learn to use it. I have already experimented with new noise functions and procedural maze generations quite successfully. But it will take some time until I will come up with some real games. For now there are at least two games I continue to develop with GM, Hazard #32 (it will have at least one update) and the random downhill ski game. My big project Modus will be likely done with Godot instead of GM since GM:S1 will not be likely supported anywhere the day Modus game will be finished (assuming it ever will).

How to get Linuxsampler working

Some time ago I hyped about sfz pianos and other (sampler based) instruments on Linux. This time I focus again on sfz-instruments (and .gig instruments). Linuxsampler is one of the most common samplers on Linux to play this stuff. However, it has not been updated much after 2007. So, don’t expect things going smoothly to get it working. I just got it working on my Ubuntu Studio 18.04, though. So far I haven’t been focusing on Linuxsampler much, since I have used mostly Carla and SFZero to play sfz files. They have quite limited features, though. One of the most advanced free sfz player is Sforzando by Plogue. However, since it doesn’t have Linux native support (and I had no luck to get it working with Wine) I decided to look for some alternatives and I had Linuxsampler already installed.

About installing
First of all it is important to make sure that every dependency to Linuxsampler is installed, there are quite many of them. Instead of Qsampler GUI, I recommend using java-based Jsampler (Fantasia). It looks nicer and is easier to use (which might not tell much in case of Linuxsampler’s user experience…) Anyway, I am not going to instruct more on these install procedures, since it probably differs much across Linux distributions. However, because of exceptional licensing things, Linuxsampler might be found only on repositories of audio oriented distributions. So, one might need to set up correct repositories first.

Little fixing
After the whole Linuxsampler mess is successfully installed, one can try how the interface works. Manual might help. However, when loading instruments something like this might pop up: “Updating database directory: Cannot open  instruments database: /var/lib/linuxsampler/instruments.db”. The directory “/var/lib/linuxsampler” does not exist. This seems to be an old bug that has not been fixed. So, we have to fix that. Since the missing directory locates outside user directories, we have to create the directory on command line with root privileges like this:

cd /var/lib
sudo mkdir linuxsampler

Then we have to create the instrument database file with Linuxsampler, based on instructions that Jsampler manual gives for creating databases (remember to write sudo as we’re still making changes outside user directories, although I don’t remember was it necessary):

sudo linuxsampler --instruments-db-location instruments.db

Now it shouldn’t give the error anymore and Linuxsampler should work… well, reasonably. I mean, it’s still quite an old software and probably that’s why it doesn’t support the newest and the most advanced sfz-files for example, which is important to keep in mind. But at least now you can import some .gig instruments, VSCO2 sfz orchestras or whatever floats your boat. Have fun, hopefully.

My fresh Debian installation

Nerdification intensifies, I just had to go deeper into the rabbit-hole and install Debian 9 on my desktop computer. This is practically my first step outside Ubuntu family towards a purer realm of Linux. I had f*cked up my update systems on my previous Ubuntu Mate and this time I wanted to make sure there isn’t at least needless bloatware messing my shit. Debian is a nice compromise between blank technical stability and compatibility with many packages I’ve used to play with on my audio workstation. Although, it has already turned out that I have to use command line more to install and keep my stuff organized from now on. But hopefully I can take it (at least it’s not CLI-hell like Arch). Even though I’m not very experienced with bash, installing and updating are pretty straightforward operations with that.

---

Suomeksi siis lyhyesti: sain pöytäkoneeni Ubuntu Maten päivitykset sekaisin ja päätin kertaheitolla mennä kaninkolossa syvemmälle ja harpata Ubuntusta puhtaampaan Linux-maailmaan ensimmäistä kertaa. Depis oli siis luonnollinen valinta, jotta audiotyöskentely hoituisi yhteensopivasti jatkossakin.

Yoshimi/ZynAddSubFX/Zyn-Fusion… wait, what?

http://zynaddsubfx.sourceforge.net/

ZynAddsubFX is as complicated as it sounds: it is one of the most ground-breaking software synthesizers ever created and until recently it was for long a Linux exclusive open source software. (Now it is available on Windows and OSX as well.) The development started 2002 by Nasca Octavian Paul. Since then the synth has gone through many changes and phases. As a result one can say that a dear child has many names, but what do they all mean? ZynAddSubFX is the main engine of the synthesizer for which the slightly different Yoshimi is based on. Zyn-Fusion on the other hand is the recently updated GUI for the original ZynAddSubFX engine (which was updated as well). For simplicity, some people (read: a little bunch of audio nerds) call the whole thing just Zyn.

The long history and a wide range of features doesn’t make this software very easily accessible and it probably isn’t a good idea for me to give any tips about this either. There are many things to learn for me as well. However, I have been using Zyn before the massive rewriting update 3.0 with Zyn-Fusion GUI. Even though it is a bit difficult for me to find some of the features now, the change has been definitely taken towards a right direction. The old GUI was really old and messy even when you got used to it. Although, at this point it is sometimes difficult to say whether I’m doing something wrong or is there a bug. Heck, it can be even both! Zyn-Fusion and the rewriting project is still on the beginning and it is understandable that it will take a lot of time to release the full potential of the engine to action. (At the moment it seems that microtonal functions work only in Yoshimi for now.)

Despite all these difficulties, Zyn has been the most used synth on my Tiānxià songs. It is not only capable generating diverse sounds, it can also handle quite big projects without too much hassling and fear of crashing. It takes the best features of many professional synthesizers and make them multitimbral and multipartial (16 simultaneous assignable MIDI channels). In layman’s terms, Zyn only isn’t a good synth warrior, it is a whole army of them! There simply isn’t many VA-synthesizers capable of handling functionality in this extent. Of course, these kind of wide tasks are usually left to different kind of audio hosts and DAWs, but at least on Linux environment Zyn optimizes all of this quite well on its own. It can be a great relief if you’re working with light hardware or “Linux-typical modular workflow” which might include many separate engines running at the same time with some Wine powered VSTs etc.

I don’t have much experience working with monolithic DAWs that can do everything by themselves. (Ardour theoretically can but I don’t see a point to use it like that for many reasons.) I guess Zyn could be really confusing for someone used to a singular DAW. Zyn is a textbook example of an alternative approach of an user interface in a modular workflow – and it doesn’t follow many familiar conventions in this sense either. While this can be confusing, it can also spark creativity and interest into alternative approaches of music making, which is always a good thing in my opinion. Zyn is also used as one of the default synths in popular LMMS sequencer (which I have never tried), but that might still use the old GUI, I am not sure.

There is a demo version of the updated Zyn with Zyn-Fusion GUI which goes silent after 10 minutes. Even though Zyn-Fusion seems to be commercial, the full version of it is still possible to install freely via terminal commands, at least on Linux… AFAIK devs are even aware of this, but I guess they just try to somehow cover up bandwidth costs meanwhile bugs get reported from users of the zynaddsubfx-git version or something, but I dunno.

Linux Sfz pianos & SRX-02

ENG: below… FIN: Tiivistettynä käsittelen joitain netistä löydettyjä ilmaisia sample-pohjaisia akustisia soittimia ja vertaan niitä uusimpaan hankintaani, jonka päivitin Fantomiini. Lopuksi esteettistä hienostelupohdintaa.

On previous post I talked something about Sfz-format which can deliver some quality sample-based sounds for computers. On Windows platform Sforzando is a popular software to handle these files. On Linux there is mostly the LinuxSampler engine and Carla which utilizes it best. Notation software MuseScore 2 also supports Sfz playback (but in my case I had to update it to the newest version 2.3.2 in order to get it working). However, at the moment Carla pretty much lacks any interface to edit settings for sfz-instruments. There is also SFZero plugin which might fill some gaps but it is quite limited as well. Personally this is not a big problem for me as long as I get the sounds itself working. Besides one can always use external effects to edit the sound of these instruments. Virtual analogue synthesizers fulfill my editing urges quite well. Anyway, I think this lack of editing interface will be supported in the future in a way or another by KXStudio developers.

So far I have already tried some free pianos and other acoustic like orchestral sfz instruments. Especially Salamander piano and Iowa piano sound very promising to me. (Iowa piano has some small velocity key mapping issues but personally I really like how the samples itself sound.) There are probably many other good free alternatives as well, if one digs interwebs enough. It is good to see free and open source music production potentially thriving on this area even though I have been focusing more on actual synthesizers when it comes to software instruments. I have couple reasons for this. Reason 1: I think that in order to have an authentic sounding acoustic instrument library one should use either the real acoustic instruments or relatively advanced software. Otherwise the production is at stake sounding plastic which practically always happens if one is limited using formats like sf2 (not the sfz!). Reason 2: I have a professional keyboard Roland Fantom X8 which has pretty much earned its place as a king of my sample based instruments. However, it has getting old…

SRX-02
Fantom X8 is almost something like two decades old instrument already and some of the sound libraries sound relatively dated in modern music. Fantom’s engine supports 4-stage velocity samples (which most of the default patches are not fully utilizing of course). To extend my instrument’s life I made a little cosmetic surgery for it and made the first eBay purchase of my life. I got an used SRX-02 Concert Piano which has a dedicated library of some rare European grand piano Premier. The card contained almost 50 variations of that (approximately 2/3 of those are mostly bloatware, though). Despite being already quite an old sound card, the main piano sounds really good and can rival at least the free piano sounds I have found from the Internet.

Aesthetic notion
All in all, not having the newest cutting-edge soundbanks or instruments is not a deal-breaker for me. At the moment I am not hunting some extremely modern sampler keyboard to replace my Fantom any time soon even if I would have more money to do that. (Although it is becoming old in terms of practical use…) Actually, I think that too hi-fi sounds might often sound too sterile and approach the plastic factor from another angle. That’s why I think it often adds positive character to have some slight imperfections/anomalies/unprofessional traits in the sound. Compared to the SRX-11 Complete Piano, my choice sounded more personal and unique based on demos even though the SRX-11 definitely is “a higher quality” product. Generally my SRX-02 could be more perfect but also as a professional product of its time – less perfect.

Good thing about free Internet samples is that if they are at least somewhat done by the rules of quality, they can not only sound good enough for production but they still might have also some unintentional interference which makes the sound stand out from others in the best case. In many cases the interference becomes just a nuisance, though. Professional products, however, tend to calculate so many variables that there remains no room for random things. In order to appeal as professional the end product often ends up being overly hygienic and sterile. This results lacking the unique characteristic of the sound.

Can aesthetic imperfection be taken into account in professional production intentionally or has it always to be unintentional in order to flourish? If the latter is true, no amount of professional brute force can ever guarantee the most beautiful aesthetics. Of course I am not claiming that anyone could make better arts and products than professionals. That could be a tempting view for someone unwilling to see much productive effort in learning and working with things. (Someone like me, lol?) Anyway, what I am saying is that even the best of us should not totally rely on our skills. This kind of total power could eliminate the seed for something undiscovered new and fresh. There should be always room to let go.

How to find good Linux software instruments

Finding a good virtual instrument for Linux comes down pretty much with the knowledge of instrument standards. (Most of the things here apply with effects and such as well, but I focus on instruments.) It is good to know, that some instruments and plugin formats work better on Linux than others. Even though, there are many cross-platform instruments, there are also operating system exclusive instruments on pretty much every OS. With certain settings Linux can – as a highly customizable OS – expand its boundaries, though. Still, the most reliable way is to seek the best instruments for Linux supported standards at first, and only after that rely on other alternatives (if one cannot find suitable native Linux instruments). That said, some Linux supported instrument formats include LADSPA, LV2, DSSI, LXVST (Linux VST), Sf2, Sfz and probably some others I don’t even know.

VST
VST – the common instrument standard for Windows – is possible to get working under Linux when using some setups with Wine (and JACK). Some important software running VSTs on Linux are Festige, Carla and Ardour. However, it is relatively heavy and gimmicky for Linux to run mostly Windows based audio plugins. Therefore, using many VSTs (especially with their GUIs) simultaneously could easily result into crashing. Also, some VSTs might work only partially* and some VSTs might not even work at all. In my experience sample based VSTs tend to be often like that, but the most of the VST synths usually work just fine.

*For example: they cannot read default bank libraries, which is not usually a big problem since the earlier mentioned Linux software can replace the saving capabilities with their own settings.

Sf2
Sf2 works well on Linux without problems. Fluidsynth (Qsynth) is probably the most common instrument to play these soundfonts. However, because of the limitations of sf2 format, I consider these banks mostly as toys. It is like an virtual equivalence of plastic GM keyboards, good maybe for beginners (and sketching quick arrangements on MIDI sequencers) but that’s about it.

Sfz
Sfz is not to be confused with Sf2. It has many more supported features, such as multi-velocity sampling. I discovered this format only recently but so far in my experience, Sfz is probably the best format to find and play some acoustic sample emulations on Linux easily. (For electronic synth sounds I of course recommend using some actual synthesizers!) One can play Sfz instruments with LinuxSampler, Carla or MuseScore 2. I recommend Carla (+ SFZero plugin) since MuseScore is focused on notation and vanilla LinuxSampler is difficult to get working. (It is difficult to install and I guess the development is still really WIP anyway.) Also, Carla uses LinuxSampler’s engine anyway AFAIK… Final option is of course to use Windows software Sforzando under Wine but I don’t guarantee anything. More about my sfz piano experiments here.

Some good instrument sources for Linux (list will be updated):
https://bigcatlists.blogspot.com/2014/05/totally-free-instrument-by-type.html (lotsa’ free stuff!)
http://bigcatinstruments.blogspot.com/2015/09/all-keyboard-instruments.html (some sfz pianos but mostly stuff for commercial Windows synth Kontakt 5)
https://bedroomproducersblog.com/2010/07/01/free-sample-shootout-3-acoustic-electric-and-toy-pianos/ (more pianos)
https://freepats.zenvoid.org/ (miscellaneous instruments)
http://linux-sound.org/linux-vst-plugins.html (some LXVSTs etc.)
https://www.kvraudio.com/plugins/linux/instruments/newest (optionally one can filter all the commercial stuff away)
http://www.vst4free.com/ (VSTs mainly for Windows, see my VST description above)

Some good new instruments that deserve a mention:
https://tytel.org/helm/ (I’ve hyped about this before and I’ll probably keep hyping about this in future)
https://vcvrack.com/ (modular synthesizer, and now with a JACK support after the latest update!)
Yoshimi/ZynAddSubFX/Zyn-Fusion (I’ll post something about this later)

Käyttäjän vieraantuminen

Linux-yhteisön on vaikea avautua suurelle yleisölle, koska esimerkiksi komentorivin asema korostuu edelleen niin vahvana. Monille tämä näyttää nörttien elitismiltä, mutta mistä tällainen asennoituminen johtuu ja kuka tässä oikeastaan onkaan “asenteellinen” (sillä asennoituminen sananakin oikeastaan pitää ironisesti sisällään aika asenteellisen latauksen)?

Suurin osa ihmisistä tuntee tietokoneen käytön vain graafisen käyttöliittymän (GUI) tasolla. Teknologisessa kehityksessä GUI on kuitenkin melko tuore osa tietokoneen rakennetta. Tämä tulee esiin esimerkiksi siinä, kuinka jokainen graafinen komento periaatteessa aina käännetään koneelle tekstipohjaiseksi komennoksi, mutta jokainen tekstipohjainen komento ei käänny graafiseksi komennoksi. Tämä riippuu siitä, luodaanko tekstikomennoille graafisia vastineita ja se vaatii sovelluskehittäjiltä ylimääräistä työtä. Tässä kokonaiskuvassa lienee myös selvää, että jos käyttäjä tietää mitä tekee, tekstipohjaisessa työskentelyssä saa asioita enemmän aikaan. Vaikka hallitsisi GUI:n “täydellisesti”* eikä tietäisi komentopuolesta mitään, tällaisessa reifikaatiossa tietokoneen perusasioihinkin voi jäädä mustia aukkoja, jolloin pienienkin ongelmien tapauksessa on käännyttävä sellaisen ammattilaisen apuun, joka hallitsee tekstikomentoja. Samalla logiikalla voisi tosin sanoa, että komentorivinörtitkin ovat vieraantuneita tietokoneen järjestelmästä, kun eivät näpyttele kaikkia nollia ja ykkösiä sellaisenaan.

Kuten graafinen käyttöliittymä kääntyy tekstikomennoiksi, tekstikomennot kääntyvät lopulta binääriksi. Koska GUI:iin on yleensä rakennettu vain käyttökelpoisia komentoja, on niissä asioissa mitä tekee hankala mennä pieleen – jos tietää edes mitä suunnilleen tekee. Tekstipohjaisessa käyttöliittymässä täytyy puolestaan tietää tasan tarkkaan mitä tekee, koska toisin kuin muu ihmiskielen teksti, koodi ei anna tulkinnoille armoa. Tietokonetta ei voi komentaa sumealla logiikalla – käyttöliittymien kehitys on tämän tosiasian häivyttämisen historiaa.

Ihmiset tulevat tietokoneen käytössä hyvin erilaisista taustoista ja suurimmalla osalla ei ole kovin tarkkaa kuvaa tai eriteltyä suunnitelmaa siitä, mitä he oikeastaan aikovat koneellansa tehdä. Sellaista he eivät usein ehkä edes tarvitsekaan. Edistyneillä käyttäjillä on puolestaan selkeämpi kuva tavoitteistaan jo sen puolesta, kun tietävät mitä koneella on ylipäätään mahdollista tehdä (vaikka siitä seuraava hifistely esimerkiksi Arch Linux -hipstereillä menisikin sitten vain käyttöjärjestelmän käyttökokemuksen itselle kohdistettuun räätälöintiin). Näistä käyttäjien eroista luultavasti johtuvat ne lukuisat yhteentörmäykset, jossa eri tavat käyttää konetta linkittyvät oman arvomaailman asenteisiin ja sitä kautta toisista eroa tekeviin makumieltymyksiin. Tarkan huomion puute on kuitenkin otollisinta kasvualustaa reifikaatiolle, mikä voi osaltaan synnyttää tietokoneen toiminnasta virheellisenkin mystifioituneita käsityksiä, jolloin jää asiantuntijoiden tehtäväksi estää niistä haitallisimpia leviämästä.

Vaikka en itse henkilökohtaisesti pidäkään komentorivin käytöstä typojen ym. epävarmuuden pelossa, yleisesti pidän siitä, kuinka voin Linux-järjestelmäni virittää juuri haluamanlaiseeni multimediakäyttöön. Tämä kuitenkin vaatii jollain tasolla sitä, että opin tuntemaan tietokoneen/käyttöjärjestelmän toimintaa ja suhteuttaa tekemistäni sen asettamissa rajoissa.

*GUI:ta ei juuri koskaan rakenneta kattamaan kaikkia tietokoneeseen vaikuttavia toimintoja, mistä on sitäkin etua, ettei keskivertokäyttäjä vahingossakaan pysty rikkoa järjestelmää… tai ainakaan niin usein. (Siksi suositut kaupalliset käyttöjärjestelmät rakentuvat usein kompromissiksi kahden asian välille: ominaisuuksien määrä – saapaskäyttäjien määrä. Usein juuri käyttäjäprofiilista muodostuu käyttöjärjestelmien välisiä sosiaalisia arvostushierarkioita. Tämä ei silti välttämättä suoraan korreloi käyttöjärjestelmän ominaisuuksien kanssa: varauksella laskisin Windowsin pahnan pohjimmaiseksi, koska sen maailmasta usein kantautuvat ongelmat voivat lempeästi tulkittuna vihjata siitäkin, että asioita on mahdollista tehdä laajemmin kuin Macillä – vaikka jälkimmäinen UNIX-pohjaisena monia Linuxin kanssa edistyneitä ominaisuuksia jakaakin. Tämä on kuitenkin vain mutuani, koska en Maccejä juurikaan tunne.)

My first Linux bash script (audio setup)

Couple years back QJackCtl’s internal scripts stopped working properly for me, possibly broken by some update. This caused GUI crashing and I had to always manually set jack-sink as default audio output in order to make alsa/pulseaudio and jack work at the same time and run other things manually as well. Since music making on Linux can get quite modular with different software, scripts become eventually handy. So, I decided to fix these issues myself once and for all.

At first it is good to know that “&” makes following commands running on background. Otherwise they become executed only after previous processes are killed. The script requires at least following packages installed: qjackctl, jackd, a2jmidid (Ubuntu Studio OS or upgrade is recommended)

#ASS (Arello's Sound Script) 0.1
qjackctl -s & sleep 3 #Runs Jack immediately and waits for moment before next commands
pactl set-default-sink jack_out && a2jmidid -e #sets jack sink as default and runs a2j MIDI bridge
#Optionally after this it is possible to execute additional software, like: & software_name_or_location

Writing this as a text file script_name.sh and permitting it as executable makes this script usable. ATTENTION: At this point the script is not perfect though. It does not actually check whether jack transport is already running. In the code I only presume that under three seconds it is already running. For my purposes this is enough for now.