Update : Progress…

Changes (incomplete list) :

-Sound sources can have source position parameters/envelopes independent of the main parameters source position parameter. This feature can be accessed from the right-click menu in the sound file name in the sources list. This feature is mainly useful if you have several sound sources loaded, so that you can “scan” through them in independent trajectories, to make the source position capture the positions most appropriate for each sound.

-Audio file writing support was rewritten to allow using all of the formats libsndfile supports, as well as add support for more formats in the future. Most of the new available formats won’t be that useful for real use, but maybe FLAC and Ogg Vorbis might be useful for long live output recording “sessions”. The default format for both render and live output recording is now 32-bit floating wav, so if you want something else, visit the render/live options dialog to change it. (The old settings are no longer because that would have complicated the code needlessly just to avoid a visit from the user to the dialog.)

-More discreet indication for live output recording active state. A new button for that was added to the transport/render panel and the useless “seek forward” button was removed.

-Envelope editor time selection snaps to the grid if snap to grid is enabled. This probably needs some improvements etc, will be looked into later.

-Horizontal custom scrollzoombars were changed so that they don’t cause an immediate update to happen in the widgets they are controlling. This hopefully allows more responsive operation while scrolling and zooming in the sound viewer and the envelope editor when there are complex envelope shapes etc in the envelope.

-The sound sources panel has a new column that shows a small waveform overview of the sound. If you don’t want that, you can right-click the sources panel column names to get a menu where you can hide/show the columns.

-Bug fixed (hopefully) when removing sound sources from the sources list while playback is on. This went unnoticed as it wasn’t certain to always crash HourGlass and I often don’t myself remove sound sources from the list, but rather replace them, which hasn’t shown symptoms of problems.

The “Sound” menu has a non-functioning entry “Offline render and play/stop”. I forgot to remove that before uploading this build. The feature may or may not be implemented in the future. (It is meant as a last resort option to audition a texture if there’s no way to play it with realtime audio due to extreme CPU usage.)

Download link :

http://www.vesanorilo.com/xenakios/HourGlass_0.9.4.zip

Let’s look at the cons (ie, bad things) of disk streaming :

-It would add quite a lot of complexity into the current code if it was done properly using a background thread to do the streaming etc. Added code complexity means more potential for nasty issues like crashes and memory leaks or corrupted results. I don’t say the current  RAM based sound playback is entirely free of such issues, because it probably isn’t. Faultless code simply isn’t possible nor even desirable(*) for an application like HourGlass, and I don’t want to make things harder for myself than I absolutely have to, but implementing disk streaming would make it so. (*) Because much of the time would then need to be spent on testing before releases, allowing maybe one release a year or something stupid like that, because I am just a one guy doing it all.

-The random access nature of the Classic engine mode might mean enormous amounts of seeking the hard drive would have to do. Mechanical hard drives and seeking equals the hard drive wearing out. Do I want to cause that for the user of the software? Let me think for a second…No. This is a tricky issue as it might be a real concern or not, but there’s no easy way to test this for real. But I am a pessimist on this and consider the seeking involved would at the least mean sucky performance.

-A peaks file mechanism would have to be implemented for the sound waveform visuals pretty much by necessity. This would further complicate the code.

And the only pro (good thing) :

-You could now load your 10 minute or whatever sound file into HourGlass without the risk of running out of memory.

That single little pro just doesn’t outweigh the heavy cons for me. So sorry, HourGlass won’t be streaming from disk. If you require the ability to run very long files through a granular sound processing, there are alternatives like VST plugins you can put on the track inserts of your DAW application and stream for hours and hours, if that is your pleasure.

HourGlass was never intended to be used with long input files. It is a real limitation, but a limitation by my own design and I stand by it. You are thinking about using HourGlass in the wrong way if the way it works now isn’t acceptable.

-Per-source sound source position envelopes

-Audio files writing code has been overhauled together with the render/live recording options dialog. Many more (but mostly exotic and not so needed in real life) audio file formats implemented in libsndfile now supported for rendering and live output recording. The changes that were done should allow adding support for things like wavpack and mp3 in the future, but probably not for the next release. (The wavpack option in the dialog is now “fake” and will be removed from the public release if the actual support for wavpack won’t make it in during the coming days.)

-Sound sources list can show (optionally) a small waveform overview (click image for animation)

-Support for parameter automation envelopes in some of the built-in DSP modules has been re-enabled

-Other various changes and fixes

The rendering code renders audio from the audio mixer and the results are passed to a function in the libsndfile library to write the buffer of audio rendered. Easy enough and works. However, if I needed to support audio file formats not supported by libsndfile or even wanted to drop using libsndfile (not likely, but who knows…), this type of arrangement would quickly start to suck. It would look like this if I wanted to add new formats not supported by libsndfile :

The rendering code would now need to know the details of each of these 4 ways to write audio files (which are completely different in this illustrative scenario). This clearly isn’t acceptable. The code really needs to be something like this :

The rendering code would only interface with a module (a C++ class, really) that can internally do whatever it has to, to implement the actual writing of the files. Of course this kind of design is more complicated to implement in code and the justification to do such a more complex design hasn’t existed yet. (While it clearly would be more future-safe.)

This is just a technical ranting/reminder to myself, not a commitment that HourGlass would support wavpack or mp3 or Monkey’s Audio files in the future. (I don’t even see how mp3 or Monkey’s Audio really would be useful, but perhaps wavpack could be…) These same points of course also apply to reading audio files for use in HourGlass, as well as the live output recording to an audio file.

The video also shows another upcoming feature, the “A” buttons colored yellow for parameters that have MIDI remote controls assigned and their envelopes playback turned off.

Changes :

-”Envelope operations” panel removed, functionality now available above the envelope editor in “Ops” button.

-Knobs should now look OK on Windows XP and Vista/7 with Aero turned off

-Added actions to toggle fragment stream and aux 1/2 channels muted in the mixer

-Added actions to set previous/next fragment pitch envelope

-Fragment pitch envelopes are initialized with a 4 semitones step grid

-Live output recording checks if the previously set target folder still exists on disk and if not, uses the HourGlass executable folder as a fallback

-When live output recording is active, the Transport/render panel flashes while playing back

-Fragment panning spread mode action removed from main Sound menu and now found in the Main parameters A buttons for Pan/rotate and Pan/rotate spread

-When removing sound sources while playback is on, playback is no longer stopped (this was a bug that went unnoticed for a few releases)

-Changed threads synchronization to use WDL_Mutex instead of QMutex, as it is suspected QMutex isn’t ideal for audio applications. (WDL_Mutex on Windows is just a thin C++ wrapper around Windows critical sections, whereas QMutex appears to have some quite complicated machinery, which I quite can’t figure out why it is like it is.) While this is an internal technical detail, I might have made some mistake while rewriting the involved portions of the code and crashes/hang might result. (This hasn’t happened to myself thus far, though, while I’ve been playing around with HourGlass.)

Download at :

http://www.vesanorilo.com/xenakios/HourGlass_0.9.3.zip

Thanks yet again to Vesa Norilo for providing the hosting space.

Had some fun with HourGlass’s git history and gource…Unfortunately HourGlass was began in 2009 without using source code version control, so you

can’t see the development from the very beginning… A big The git history started back in the summer of 2010 or so…