|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.220.121.75
In Reply to: Native 24/96 sounds different posted by edward on October 29, 2006 at 10:13:41:
I have found that SRC actually sounds better on some tracks than the original 24-bit master. It is a very good upsampler indeed. On others, it is not the case. I think if it is a very good master, the original 24-bit will generally sound better. SRC may give more dynamics than the original.
Follow Ups:
I am a bit troubled by a generalization here. I think upsampling may sound better with some DACs, and worse with others. Let's remember that the DAC is the one that is turning this original data (and manufactured data, in the case of upsampling) into sound.I've found that with a Benchmark DAC almost any type of upsampling to something near 110K improves the sound because the resampler in the Benchmark DAC1 (the one that resamples everything to 110K) is awful.
I've also found that upsampling using a Lavry DA10 DAC (which does not resample the input unless you want it to) makes it sound worse.
I did find that the Secret Rabbit resampler sounds slightly better than the standard one in Foobar2000 when using the DAC1, however. I highly doubt that I could tell the difference between them in a double-blind test though.
By the way, I made a recent change to my DAC that has a far greater impact than any other change I've made recently (e.g. Secret Rabbit vs. default resampler, ASIO vs. DirectSound, cable A vs. ultra-expensive cable B, etc.): I replaced the fuse with one of those Critical Link fuses now available from PS Audio (which are actually AHP fuses from Germany). Now THAT made a signficant difference in the sound!!! :-)
I dont think I generalized at all. With some tracks SRC is better on my system. With others, the original 24-bit track is better. I believe my system is very transparent and neutral based on a lot of feedback, demonstrations and trials on other systems.The system will affect the outcome of course. If there are other problems in the system, upsampling can help or be detrimental. It is obviously very difficult to judge different upsampling techniques without having a very neutral and transparent system to start with.
BTW - other changes to your DAC would likely make even larger improvements. I usually replace the .5A fuses in equipment with 3.5-5A fuses. In my mind, the fuse is there to protect the transformer primarily and this is sufficient. I dont care if a few regulators or rectifiers blow...
But I used SRC to create the 24/96 files. The content should be identical. Why do they sound different? Somehow the process of having the DSP function in real time produces less jitter? or something?
There are many aspects of computer audio that are not transparenet, as we don't actually know what bit of code is doing what, and what mathematical model is being invoked.SRC actually does not sound good compared to a top class hardware processor, although it sounds better on a pc than other srcs.
If you use Audition 1.5 for upsampling or playback, then the thing sounds different again! There are so may combinations of software settings that al least I am at a loss as to what the combined effects are.
It demonstrates that producing the upsampled files rather than doing the upsampling on the fly creates different data.The 24/96 files that I have are remastered from reel-to-reel tape. Then they are converted to 16/44.1 on a music CD. I rip the CD and compare this upsampled to the original 24/96 master.
If as some say here, bits are bits and somputers can do no wrong, how can this be?Unless there is undocumented processes going on.
Going from 24/96 master tape to 16/44.1 and then upsample will of course change the sound. There is the question of dither when downsampling and filters when upsampling.
If as some say here, bits are bits and somputers can do no wrong, how can this be?Unless there is undocumented processes going on.
Going from 24/96 master tape to 16/44.1 and then upsample will of course cahnge the sound. There is the question of dither when downsampling and filters when upsampling.
This post is made possible by the generous support of people like you and our sponsors: