Music, Programming and a Cat

If you want to spread your music there is no way around Youtube at the moment. 

The command first, explanation and instructions below:

ffmpeg -loop 1 -f image2 -i yourImage.png -i audiofile.wav -shortest -c:v libx264 -c:a aac -strict experimental -b:a 192k out.mp4

Despite the fact that everyone uploads their music and if you are searching for some obscure or rare piece you can find it most likely there, YT still requires you to upload real videos because they started as a video platform or whatever reasons.

So you want to make the most simple video possible: Your music and a still image you can create in gimp or by any other means. It is not only a trick. Even the most basic still image adds to the mood. You can also include information like your name and address directly in there.

Ingredients: 

1. Your music file.
   Use pure wave or a flac. Don't use a compression with loss (mp3, ogg etc.).
   Make sure your audio is normalized. If you have sox installed (I bet you have) just type "normalize myfile.wave" into the terminal. This will change the original file.
2. A still image.
   Create it yourself, just text on black, or search the web for a freely licensed image with CC-by or CC-by-sa.  Skip the "non-commercial" ones. Youtube is commercial.

Youtube will determine its compression algorithm and output formats (=quality levels) by your video size. Make sure to make your image, by cropping and scaling, has a good aspect ratio (16:9 the most common) and make it big enough, even if that means upscaling.

In other words:  Make your image 1920x1080 pixels no matter what it was before. This will get you the "1080p" option in Youtube.

After that, upload to youtube via their webinterface and enjoy your 10 views and 3 downvotes.

One note about the "lossles" audio and video formats. If you only have a wave or a jpg to begin with, and you don't want to modify it any more, then there is no value in converting it to wav first. It will not make the quality better. ffmpeg takes what you have so just use your mp3 directly. But the audio quality will suffer, no way around that.  Get good source material in a proper format!

P.S.
You can find many blog posts and tutorials about that topic on the internet. But thanks to changes in ffmpeg they mostly don't work. Common bugs are that the video encodes forever or various glitches of some sort. ffmpeg is rather picky about the order of parameters, so don't move them around.

Here you can see the a short music vidoe I've created. If you can't see the embedded video click this link: http://youtu.be/415tvMAvoow

It is the third test using the free sample lib SSO. 

This time it features the choir samples, brass and a solo instrument. 

We can clearly hear the limitations of the library here: The choir has a very specific, fixed, articulation, only useful for short bursts. Even they sound bad. So best use the choir only far in the background with "too much" reverb and don't rely on it.

The brass is ok. It has staccato samples. Since staccato is easy to use and easy to record they sounded quite good in the 90s already. If you don't overuse them of course.

The strings are the same as the last two times. Don't rely on them. They sound quite good in a combination with a woodwinds+horn (as I did in experiment two).

The solo instruments have a few strong points, so you can use them for short snippets. But I wouldn't want to write a whole solo concert for one of them. That is why this is an experiment, so I could just cut it off after the beginning :)

Current state, after three videos: Use the strong points of the lib, forget the rest. Don't try to make big orchestral stuff. There is a progression from cheap/artificial (few instrument) to "I can't hear a single instrument anymore, just mush!" (all instruments). I doubt the useful sweet-spot (a medium sized ensemble that sounds good) can be found easily.

The Software was the same as in Experiment Two :  Laborejo, lisaoQt, SSO, zita-rev1, non-session-manager, jack_capture.  Check the link for more information.

I've written a guide to the communities and social places in the Linux music and audio scene (no cats, sorry). 

You can find it in the menu or just click this link: 

http://nilsgey.de/Linux_Audio_Social_And_Community_Guide

The focus is on distribution-agnostic places. Ask your technical and dependency questions in your distribution but ask how to create music or get less latency in the specialized audio help channels.

It is both for developers and users of all kinds (if there even is such a distinction). 

Have fun reading! Maybe there is even another Linux musician living nearby (check the map!)

Here you can see a new short piece I've created. If you can't see the embedded video click this link: http://youtu.be/1EQIlp4mWUo

This short music video is called "Experiment Two - For Sonatina Symphonic Orchestra" (here is the first) and is, in principle, a tool-chain test.

Linux Audio is modular. You get the best results if you interconnect different programs through JACK and control them with the Session Manager . Naturally things develop so you can improve your tool-chain constantly.  This video was produced with:

Laborejo - For composing and as midi sequencer. No "Piano & Paper" step; directly from brain to software. I can also produce a Lilypond Score with this.

lisaloQt - A sampler based on calfbox . Both highly experimental software.  This loads the instruments:

Sonatina Symphonic Orchestra - A set of sfz instruments. I wrote about them and other open source sample libraries on this blog before. Check the archives here and here and here .

zita-rev1 - A high quality reverb.

jack_capture - Check your distribution. I used this to record the audio. A simple command line tool that waits for jack transport to start and stop and records anything in between. 

Technically I had also non-mixer running but that was redundant routing. It did nothing than mix it all to a stereo output which then went into the zita reverb.  So no mixing, no plugins (EQ, compressor etc.) no panning. I did all this via midi directly in the sampler engine already. Do you know if that makes any difference? Panning first or panning later? Mixing before the reverb or in the reverb? etc.

There is always room for Experiment 3!