First, please email me at firstname.lastname@example.org and I will get you the latest installer for DPS 2.0.
Regarding downsampling: We have measured the effect of the sample rate converter in Windows. While it is not pro-audio quality, it is pretty good for a real-time converter. I can tell you the difference between music converted from 44.1kHz to 48kHz and playing native at 48kHz is very hard to hear. Even music converted from 96kHz to 44.1kHz is difficult to hear unless you have a very good listening environment. You should experiment with this without DPS by setting the output driver of your audio hardware to a different rate than your music file. If you have a good listening situation you will hear the automatic sample rate conversion.
An overall “boxy” sound you describe indicates there could be something else wrong or the profiles you are testing just aren’t right for your speakers. Let us know what speakers you have and the profiles you have tested. Maybe we can make a recommendation.
More details about DPS and audio quality:
The Digital Power Station virtual audio driver is acoustically transparent because it’s only job is to store audio sample in memory for processing by the DPS application. However, it does turn down the volume by 0.5dB to prevent any signal from exceeding 0dBFS. Exceeding 0dBFS sounds like nasty distortion!
In the older version of DPS processing there is nothing to prevent very loud sounds from distorting. So DPS v1.2.4 does have a volume pad of about 4.5dB for external speakers and headphones modes. The pad is in place to create headroom for DPS processing to occur. But the result is DPS “off” is about 5dB quieter than when DPS is not running. So when DPS is off there really is nothing happening to the audio other than Windows sample rate conversion (if the audio file and/or your output device is not set to 44.1kHz) and a small drop in volume.
Be careful not to mistake a drop in volume as a drop in audio quality! This is an easy mistake to make since our ears change frequency response depending on the loudness of the source. Look up Equal Loudness Contour. Understanding this psychoacoustic phenomenon, we always try to be sure DPS “on” is very similar to or just a bit louder in volume than DPS “off”. We actually calibrate this on/off relationship with VERY LOUD recordings like Rihanna’s SOS or something by Skrillex. If the voice stays about the same when turning DPS on/off we know we have good calibration.
DPS 2.0 only has a 3.5dB pad (and we are considering removing this) because we now have a peak limiter to prevent any signal from going over 0dBFS. This peak limiter gives us a lot of freedom to do more cool stuff with DPS without worrying about causing distortion by exceeding 0dBFS.
Just so you know, the internal processing of DPS occurs with floating point math so the end result is always of the highest quality. We are always looking to create the best possible sound without a lot of hassle to the user. That is why we are using a virtual audio driver even though that requires us to be limited to a single sample rate. The next best solution is for a developer to take our brand new Windows SDK and put DPS directly into their music player! Then no sample rate conversion will take place and you can be assured you have the cleanest possible audio path.
Whew! Hopefully I answered your questions. Let me know if I can clarify anything.