Decreased Sound Quality

[Atop Seid]

I have been unable to find a forum more suited to this question and have therefore elected to inquire here.

I have been attempting to use Goldwave to create a particularly loud sound. After increasing the volume all seems to be working perfectly and the sound is still high-quality. However, once I save the sound in the format needed for upload to Second Life, the sound becomes irritatingly reduced in quality, grating my ears. This does not occur with sounds where the volume has not been altered.

After uploading it to Second Life, the terrible sound quality seems to be unchanged still.

Any assistance in how I can prevent this problem from occuring would be most welcome.

Best regards.

[Freya Mokusei]

First the basics:-

Sounds are made up from 'waves' (in Second LIfe, we use Stereo waves, 2 at a time) composed of high 'peaks' of intensity, and low 'troughs'. These waves run as a scale from 0% (no sound) to 100% (maximum sound), with dips inbetween. A loud, intense shrill sound will have the 'peaks' of the wave close together, and a low heavy sound will have the peaks low and far apart - we call this frequency. These peaks become vibrations, which turn into different pitches of the sounds we hear, from an order of about 20Hz (20 wave-oscillations per second) to about 2kHz (2000 wave-oscillations per second). All of this is further complicated by things such as digitisation and conversion, but it is fairly obvious that we hear 'best' when sounds reach between the 'mid-range' of this 0 - 100% potential - because it allows for flexibility (e.g. music has highest quality when listened to on noise-cancelling headphones, vs. at a club with your head against the bass speaker).

Incidentally this quality loss may not be apparent in Goldwave (which I have no experience of, but using other examples), due to the nature of sound - applications have to turn analogue sound into digital (for recording on a PC or Mac), and then back into analogue to sound through your speaker. Modern applications accomplish this through clever methods of referencing the wave data, and using algorithms providing maipulation and control of - what PC's like to pretend - is an analogue interface. This is a little too complicated to explain - basically changes to the waveform are only guaranteed to be 'set' when they're recorded to disk (under the paramaters of a .WAV file, for example), and away from whatever clever mechanics is used within your soundcard, MIDI controller or the application controlling the sound.

Now, the problem:-

All points of measurement on any wave must work in the range of 0 - 100%, you can't have 110% volume (even if Spinal Tap did). When you increase the volume programmatically, it essentially moves all of these points outward (e.g. inflating the wave), toward 100%, where they stop. The more measurement points in a wave that move away from the mid-range (toward 100%), and what you'll eventually achieve is a flattening of the wave - a removal of peaks and troughs -  at a high intensity (~80-100%). This will have the natural effect of removing any clarity in the sound, causing the effect that you speak of.

If you're playing near the top of the wave-height potential, you have no room to transform the wave to a higher intensity without damaging the height of the waves (and therefore, the quality). It's the same as listening to Metallica or Blind Guardian at full-blast through your cellphone speaker, it can't produce the required 'depth' of sound.

The solution(s):-

Record more clearly, or at a reduced initial range (say, no higher than 50% intensity) if you definitely need to raise the volume programmatically.

Depending on the application, you might just be able to 'layer' this sound in Second Life, play it multiple times on top of itself to increase the apparent volume of the sound without changing the wave.

[Atop Seid]

Many thanks for your response. Unfortunately, the sound effect I am utilising was not recorded on my microphone and has been on my computer for some time and therefore I cannot record it at a higher volume. Although I shall try your suggested strategy of overlaying the same sound over the other in-world.

Many thanks once again, and best regards.

[Madelaine McMasters]

Hi Atop,

 The distortion you describe (as surmised by Freya) sounds like "clipping". Somewhere in the path from your good sound to the bad one, the amplitude of your passage was incorrectly limited in range. The result of this is that the smooth crests of the sound waves have their tops and bottoms clipped flat, like skipping a jump rope inside the house, where the rope must go straight where it hits the ceiling and floor. Clipping is very annoying, as I think you've just discovered. The question is, who clipped your sound?

I don't know why your sound would be fine within the Goldwave app, but distorted in SL. Have you tried importing your exported sound back into Goldwave to give it a listen? If it sounds terrible, then Goldwave is the culprit and you must dig into the owner's manual to find out why the software is misbehaving. If the re-imported sound is fine, make sure it truly is in the correct format WAV PCM, 44.1KHz, 16-bit, mono. I don't think SL will import anything else, but you never know.

Presuming you can locate the cause of the clipping, you may be done. Even so, I have a recommendation for getting the most volume from SL sound clips.

I use Audacity to process sound files. When I want to maximize the volume of a clip, I select the entire thing and "normalize" it. I expect Goldwave has a similar function, which looks for the loudest passage in the clip and makes that the loudest possible sound, minus an amount of headroom you can specify. If your sound will be the only one playing, you'd need no headroom at all. If it will be part of a melange of sounds playing, you may want a few dB of headroom. so that when added to the melange, the result isn't driven to distortion.

Now, if you have a sound clip with quiet passages you want to be louder without distorting the loud passages, or you want to reduce the loud passages without losing the quiet ones, so you can increase the overall volume, you will want to apply a "compressor" to the clip. I'm sure Goldwave will have a compressor that accepts at least a parameter or two to control the degree and nature of the compression. Experiment with those parameters until you get the sound the way you want. You may have to normalize again when done.

Good luck!

[LepreKhaun]

I second using Audacity for sound editing. It's feature rich, powerful and flexible, with a user interface I wish Blender had. Plenty of tutorials on the web for using it and a wealth of plugins as well.  And the cost is free!

 

I've used it since 2004 for various tasks and have not considered any other program since. It does have a few format limitations but those shouldn't vex anyone other than professional broadcasters that are limited to proprietory formats.