Hello All,
About a week ago I hosted an Austin Audio Society Event where we took a single DSD track and broke it up into the following conversions from a single DSD file (DSF512):
*D = Dithering
The test track used was Billy Jean by Vanessa Fernandez off of her latest album "I Want You"
This test came off of the frustration I had with two things:
1) I was having to wait far too long (few minutes) to transcode the DSD file to PCM and buffer to Ram in the streamer. The Aries Cerat DAC's I use are all PCM so it has to be converted prior to playback.
2) When using DBPoweramp instead to convert the files to PCM, the end result was inferior sound after conversion by a fair margin.
So I did a lot of digging and asking around and ended up using AuI Converter to get the best results. It is much pricier than DBPoweramp, but having used that for years and using the AuI software it was far superior for this task.
Unfortunately A/B testing was not as popular as some of our other events and the turnout was low, but the results still proved to be useful. I will say this could be part 1 and will get a few more test subjects and do another round to be sure. All tracks were not played in sequential order and randomized and the occasional repeat to see how our subjects could tell if there were consistent differences that were obvious at a given sample rate and bit depth
The fascinating part with the small control group were identical results 50% of the time out of the 10+ comparisons. The ratings for the comparisons were Better, Worse, or Same. It could be more scientific as a double blind test, but I am afraid I dont have that kind of time or interest in going to that extent.
So the Results for Part 1?
-Dithering was a consistent preference
-Both 16 bit and 32 bit depths were least preferred.
-44 and 352 sample rates were the least preferred
-24bit depth was the most preferred with dithering applied irrespective of sample rate
-The differences were not glaring. It was hard by the panel to distinguish a clear winner when they did choose one. So we are talking minor differences that were very hard to tell
All tests were performed using my reference setup
Kaiser Kawero Classic speakers
Vermouth and Top Wing cabling
Aries Cerat Kassandra Limited Edition DAC, Impera Limited Edition Preamp, and Ianus Geminae Mono amplifiers
Pink Faun Ultra Streamer in Stylus mode
When I get around to it I will round up some more test subjects and do another round of testing. I have a very large DSD library, so before I roll up my sleeves and convert them all I want to make sure I choose correctly so I dont have to do it again at a later date.
About a week ago I hosted an Austin Audio Society Event where we took a single DSD track and broke it up into the following conversions from a single DSD file (DSF512):
*D = Dithering
| 44/16 D |
| 44/16 |
| 44/24 D |
| 44/24 |
| 44/32 D |
| 44/32 |
| 88/16 D |
| 88/16 |
| 88/24 D |
| 88/24 |
| 88/32 D |
| 88/32 |
| 176/16 D |
| 176/16 |
| 176/24 D |
| 176/24 |
| 176/32 D |
| 176/32 |
| 176/32 |
| 352/16 D |
| 352/16 |
| 352/24 D |
| 352/24 |
| 352/32 D |
| 352/32 |
The test track used was Billy Jean by Vanessa Fernandez off of her latest album "I Want You"
This test came off of the frustration I had with two things:
1) I was having to wait far too long (few minutes) to transcode the DSD file to PCM and buffer to Ram in the streamer. The Aries Cerat DAC's I use are all PCM so it has to be converted prior to playback.
2) When using DBPoweramp instead to convert the files to PCM, the end result was inferior sound after conversion by a fair margin.
So I did a lot of digging and asking around and ended up using AuI Converter to get the best results. It is much pricier than DBPoweramp, but having used that for years and using the AuI software it was far superior for this task.
Unfortunately A/B testing was not as popular as some of our other events and the turnout was low, but the results still proved to be useful. I will say this could be part 1 and will get a few more test subjects and do another round to be sure. All tracks were not played in sequential order and randomized and the occasional repeat to see how our subjects could tell if there were consistent differences that were obvious at a given sample rate and bit depth
The fascinating part with the small control group were identical results 50% of the time out of the 10+ comparisons. The ratings for the comparisons were Better, Worse, or Same. It could be more scientific as a double blind test, but I am afraid I dont have that kind of time or interest in going to that extent.
So the Results for Part 1?
-Dithering was a consistent preference
-Both 16 bit and 32 bit depths were least preferred.
-44 and 352 sample rates were the least preferred
-24bit depth was the most preferred with dithering applied irrespective of sample rate
-The differences were not glaring. It was hard by the panel to distinguish a clear winner when they did choose one. So we are talking minor differences that were very hard to tell
All tests were performed using my reference setup
Kaiser Kawero Classic speakers
Vermouth and Top Wing cabling
Aries Cerat Kassandra Limited Edition DAC, Impera Limited Edition Preamp, and Ianus Geminae Mono amplifiers
Pink Faun Ultra Streamer in Stylus mode
When I get around to it I will round up some more test subjects and do another round of testing. I have a very large DSD library, so before I roll up my sleeves and convert them all I want to make sure I choose correctly so I dont have to do it again at a later date.
