Development:Audio

From VsWiki
Revision as of 13:31, 24 December 2008 by Turbo (talk | contribs)
Jump to: navigation, search

Introduction

This guidelines standardize the sound and music in order to achieve equal quality for different contributions throughout the game. This standards are applicable for:

  • music
  • voices for intership radio transmissions, and perhaps campaign fixers
  • other sounds

Sound Requirements

The following quality settings apply:

  • 44kHz sample rate (the best that most microphones under $200 can produce)
  • Mono for voices and sounds
  • Stereo for music
  • Vorbis OGG format (which also requires it to be stereo) for the final production version

Music Requirements

FIXME

Music Engine

The soundserver (integrated in VS) handles playing songs, while the Vega Strike engine handles sequencing them using the defined playlists (which are selected according to the situation by dj_lib.py).

The soundserver uses SDL to play them, which supports .mp3, .ogg, .wav, I don't think it supports flac (perhaps on Windows), and it does support .mid (but I have the feeling it will get dropped, at least temporarily, during the soundsystem rewrite, as supporting midi with only OpenAL is simply impossible - we'll get something later, though, don't despair).

.m3u lists are just like winamp lists, which are a bunch of files, each on its own line (relative paths are computed against where soundserver.exe is, which is usually inside the bin folder - and usually you want to always say ../music/something ), and you can add comments (on their own line, I don't think it works on the same line) by starting the line with '#'.

Playlists are stored in the user-data folder (usually under ~/.vegastrike), inside their own folder hierarchy (~/.vegasrtike/playlists). Inside playlists, you may find additional grouping, by situation:

  • battle
  • peace
  • threat

Inside each, you'll find a playlist for each faction. Usually, the faction is selected according to the system owner at the time battle/threat/peace happens, though inside peace you'll have also around_sig and away, which get selected according to whether you are near a significant unit (base or planet - dockable) or not. That all gets configured in dj_lib.py, so if you add playlists, you have to add them to dj_lib.py as well - I doubt anyone will have to add playlists, though.

Inside playlists, you can (currently - until the extension commands get added) bias the frequency of some tracks by simply repeating them on the playlist. The engine will avoid playing the same track over and over, having a record of recently played tracks.

(information provided by klauss)

Music Engine Future Plans

Eventually, we'll add sequencing commands to playlists, disguised as special comments (probably '#!<command>') - that will allow us to specify a bayesian network for the tracks, with (potentially) custom transition - that means, each track can have a certain subset of tracks follow it, with a certain assigned probability. That's not that far in the future, it needs not wait for the rewrite, as it's doable as an enhancement to the current engine.

(information provided by klauss)

Voice Acting

Voice acting provides in-game voices for ship-to-ship radio transmissions, and perhaps some day for campaign fixer conversations.

Current voice acting projects are as follows:

Voice acting is easy to learn and fun to do. For instructions on voice acting, see Turbo's tutorial.

Fixer Voice Lines

Eventually, as campaign missions are written and finalized, we hope to produce voice acting for them.

Ship to Ship Radio Communication

This type of voice acting provides audio to go with the faction-specific ship to ship communications that lend fun and atmosphere to the game.

Broadcasts are keyed to events in the game, so each line must be in a separate file. Because the voice lines will all go into the VS/data/communications/sounds folder, the file name convention must be clear and consistent. The file name format will be:

<FACTION>_<OPENING-TEXT-OF-LINE>.OGG

Examples: 

 aera_leser-being.ogg for the Aera broadcast, "The lesser being is always vanquished."
 rlaan_boring-boring.ogg for the Rlaan broacast, "Boring, boring, tiny one!  You bring us no tupperware."

A ship-to-ship factional communcication voice acting package should include the following:

  • all the factions text lines from VS/data/communications/<FACTION>.xml
  • the "greeting text" line from VS/data/modules/missions/bounty_leader.py
  • if the faction is an alien language, the following additional guidelines apply:
- the translator's voice should sound flat in dynamics and pitch, while remaining clear and understandable
- the alien voice should speak for 1 second before the translator starts
- when the translator is speaking, the alien voice should be muted by -3 Db to make the translator easier to understand
- one or more alien "fixer" lines, without the translator overlay, that we can use for randomly generated fixer conversations in bars
- there should be 2 versions of the package, one with the translator overlay for the English version of the game, and one without the translator so that voice actors can eventually create versions of the game in other languages

More detail on creating the translator's voice is covered in the section below.

Creating Alien Voices

A subset of voice acting is the creation of alien voices. The short tutorial below describes how Turbo created the voices for the Aera ship-to-ship communication.

This is the process I used to create alien-sounding speech without actually developing a language. If someone wants to do this for another race, please do, and start a new thread in http://vegastrike.sourceforge.net/forums/viewforum.php to get feedback on your work.

To begin, you should look at my voice acting tutorial to get some basics about sound editing tools and how to use them: http://www.willadsenfamily.org/us/don/ttlg/voice/va_tutorial.htm tutorial

We tried a couple of things for the automatic translator that does the "voice-over" of the alien speech. Our first attempt used MicroSoft Narrator's "Michael." MS Narrator's voice meets the criteria for a cold soulless machine voice, but he was too hard to understand. We also tried applying a "vocoder" to a human voice, but it was also difficult to understand. What we settled on, to represent 1200 years of voice-generation and translation technology, was the following:

- record all the translator lines in a flat, emotionless voice - edit out unwanted junk like lip smacking, breathing, etc - noise removal and click removal - low pass filter for 15K Hz, other values default - high pass filter for 1000 Hz, others default - leveller (heaviest setting) - if desired, adjust pitch - if desired, repeat the click removal, filters, and leveller again - amplify to full volume - one last click removal

Finally, separate each translator line into a separate file for editing.

The aera speech is based on humpback whale songs I found on the Internet. I used whales because they came to mind when I read that aera have vocal resonators (Artstyle_guide:Aeran) in their heads, instead of vocal cords in their throats. I plan to use insect sounds for the Rlaan. Other alien races' speech could be created with similar techniques, using different animal sounds or your own voice as the basis for each species.

I modified the raw whale sounds by removing noise and clicks, then adjusting the tempo to 350% and the pitch down an octave. Audacity makes this easy. For example a tempo change in Audacity will not change the pitch, so you avoid the "Alvin and the Chipmunks" effecct. Then I saved the individual bits of modified whale sound into different files so I had a variety of tones, growls, and grunts from which to choose. This took about 2 hours.

Next open each translator's line one at a time. Mute the translator's speech so it will be silent while you work with the alien speech. Import the alien speech bits into a separate track of the same file. Generally I used tones to represent positive feelings, growls for negative feelings, and grunts as modifiers, but different aliens may do it differently. Change some individual parts to add variety: change pitch here or there, reverse a few parts if it sounds good, and add silence. It is best to do this with the (muted) translator speech on the screen so you can see how the patterns line up, and it is easy to select a few seconds and play it to see what it sounds like without the voice over the top. For silent portions of the translator speech, put a particularly interesting part of the alien speech there. If the translator is difficult to understand for a particular word, make the alien speech pause at that time. There is no reason an alien language would have the same tempo as its translation, after all. The translator and alien can be approximately the same length, because when you go with the final OGG format both will be mixed into stereo anyhow.

Finally, deamplify the aera speech to about 25% of the volume of the translator voice. Do this by selecting that track and the Amplify command. Then, unmute the translator, and listen to the alien voice and translator together. Make any final edits to the alien speech. If you can play the result with your eyes closed and understand the translator, while being aware of the alien voice behind it, you are done with that line. The last part of the process, assembling the alien speech parts and final editing, took about 4 hours for the 37 lines found in the source document.

Retrieved from "https://wiki.vega-strike.org/mediawiki/index.php?title=Development:Audio&oldid=16641"