XMA VS. VORBIS – Comparing Game Audio Compression Methods

Comparing compression methods of the XBOX 360 and PS3

There is no doubt that next-gen gaming has demanded more complex sounds than ever before in games.  The amount of sound necessary for a high fidelity game is many times larger than that of a film due to longer game lengths, variety of sounds per event, and needing sound to cover all possible gameplay situations.

It is for these reasons that file compression is a necessary evil of game audio.  This isn’t the signal reduction type of compression, but rather data compression to save space.  Without it there is no way to fit all the necessary physical media onto the shipped game.  On PS3 and X360 you can use PCM, XMA or Vorbis formats.

There are pros and cons to each of the formats.  PCM is uncompressed, but takes up a lot more space.  Vorbis sounds great and loops easily, but can take more processing power than XMA.  XMA is free on the X360 because of a hardware decoder, but can be hard to loop properly and I think it sounds the worst.  I decided to take some time to compare and contrast the compression qualities of Vorbis and XMA.

Below you will find charts analyzing the two methods and a summary comparing them.  This post gets pretty techy.  If you are in a hurry you can just look at the charts and read the summary below 🙂

I hope this will help you figure out what compression methods you will use on your games in the future.  I’d love to hear any advice you have on conversion methods in the comments.

VORBIS CONVERSION QUALITY

I did two vorbis quality tests. I took an explosion with glass and a vehicle engine loop.  I then converted them to Vorbis qualities 2,4,7, and 10 using normal and high quality conversion.  The difference in file size between quality 2 and 10 ranged up to 300 kb.  However, the sonic difference is pretty subtle on most assets.  This is of course unless it has LOTS of high end.

CHARTS

Note the spectrum analysis below.  The real difference is found with a 20db reduction around 11 and overall reduced quality.

Original Asset

Vorbis Quality 2 – Low shelf drop below 20 and wide dip at 11k at about 20 db down compared to original asset.

Vorbis Quality 10 – only a 10 db reduction at 11k.

Comparing Original in blue with Vorbis Quality 2 in Orange.  Note the reduction in highs and 20 db dip around 11k.

CONCLUSION:

The difference between Vorbis 4 and 10 is negligible.  Vorbis 4 seems to be the best sound to size compromise.   Loud heavily compressed sounds are less affected by these qualities.  If an asset REALLY needs frequencies above 10 k and is iconic to the game then it could either be changed to a higher quality or PCM and streaming.  Any setting under Vorbis 2 introduce noticeable artifacts.  Rarely should sounds need to be higher than 4 high.


XMA CONVERSION QUALITY

Next, I used WMA to see how XMA sounds when using it’s VBR compression settings.  WMA seems to use the same compression algorithms as XMA so this should be pretty accurate.

My previous Vorbis comparisons used values = 0,2,4,7,10

I intended to use XMA values of= 20, 35, 50, 75, 100.  I ended up using 25-50, 75, and 100 WMA because this is as close as I could get using VBR WMA methods.

Results:

XMA Vehicle Comparison

#1 = original

#2 = 75 WMA VBR

XMA 75 drops at 18,000 Hz. about 28dB.

#3 = 50 WMA VBR

#4 = 25 WMA VBR
Note the steep falloff of 40dB at about 11,000 Hz. for XMA 50 and 25.

XMA Explosion with glass Comparison

#1 = original

#2 = 75 WMA VBR

#3 = 50 WMA VBR

It drops a whopping 50dB at 15,000 Hz.

#4 = 25 WMA VBR

Note the VERY STEEP falloff at about 11,300 Hz.  It’s 60 dB down at this frequency.

CONCLUSION:

The difference between XMA qualities is much more obvious than that of Vorbis compression.  XMA 50 and below introduce steep rolloffs starting around 15 kHz.  Aliasing was also more noticeable in these tests than the Vorbis tests.

Overall Conclusion

Vorbis sounds FAR superior to XMA in my opinion.  Just by looking at the charts above you can see that XMA has much more frequency altering and falloff.  The listening results sound better than the visual results, but anything needing clarity such as glass or with anything higher than 12,000 Hz.  should use Vorbis if possible or XMA 60+ to maintain sonic integrity.  I believe it is best to use Vorbis as often as possible until CPU usage becomes an issue.

Thanks for stopping by.  I hope this was enlightening to those of you wanting to know more about game audio and compression methods.  If you have any comments I would love to hear them.

Note: I apologize for the lack of audio samples. Unfortunately I can’t post the files I used in this test.

 

Thanks for visiting the Aaron Brown Sound blog! Come back soon for more posts about video game audio, audio engineering, sound design, composing and all other things relating to being an audio professional 🙂

Linked In Profile: http://www.linkedin.com/in/aaronbrownsound

Demo Reels: http://www.aaronbrownsound.com/resume-awards-demo-reel/

About: http://www.aaronbrownsound.com/about/

How to break into the professional audio industry – Find and keep your first video game jobs and recording studio jobs.

If you are reading this then you likely aware of how difficult it can be to get your start as a professional audio engineer. Whether you want to work in recording studios or the games industry, breaking in takes more than just talent and dreams. In this post I’ll give you a lot of tips on how to get your big break and stay employed once you do.