May be offline convertion to DSD will be more simple? Just prepare some drive for converted records and go ahaed with minimum load on PC.
Indeed however the filters HQ Player uses are unavailable for offline upsampling and these filters somehow produce superior sounding results.
Indeed however the filters HQ Player uses are unavailable for offline upsampling and these filters somehow produce superior sounding results.
Hmmm... technically, can't see why the live stream cannot be dumped to a file?
Hmmm... technically, can't see why the live stream cannot be dumped to a file?
Hmm, indeed, didnt even consider that, could be done with something like an asio driver writing to disc.
there was a long discussion for this offline convertor on computeraudiophile: Audiophile Inventory - musical and audiophile software.
Offline Upsampling
If I remember well: people tried that soft, played with HQ and compared with HQ online convertion, and were more or less happy.
Such a way:
1) mate DSD drive with BBB in one box, with linux (DSD256 is the limit, I think) + simple and clean PS for BBB.
2) Keep on shelf some external drives with converted music.
3) Use without external PC, managing by some tablett
Offline Upsampling
If I remember well: people tried that soft, played with HQ and compared with HQ online convertion, and were more or less happy.
Such a way:
1) mate DSD drive with BBB in one box, with linux (DSD256 is the limit, I think) + simple and clean PS for BBB.
2) Keep on shelf some external drives with converted music.
3) Use without external PC, managing by some tablett
Last edited:
Hmmm... technically, can't see why the live stream cannot be dumped to a file?
Jussi knows how to do it. I had him offline convert 3 tracks for me. He just doesn't have user friendly software to do it.
Jussi knows how to do it. I had him offline convert 3 tracks for me. He just doesn't have user friendly software to do it.
conversion details?
how did they sound compared to the originals? 😉
I missed the companion thread on the filter requirements for the 10s of MHz DSD components
I also don't see RF capable connectors or pads for them that let you even connect to the DSD without problems
I also don't see RF capable connectors or pads for them that let you even connect to the DSD without problems
I missed the companion thread on the filter requirements for the 10s of MHz DSD components
I also don't see RF capable connectors or pads for them that let you even connect to the DSD without problems
At DSD512, the signal frequencies are around 25MHz. The is a 6GHz U.FL coax connector option on board!
HF noise needs to be filtered out. More information on tuning and optimization will be provided in the users manual.
conversion details?
how did they sound compared to the originals? 😉
I had him convert 3 redbook tracks to DSD 128. I loaded them on an SD card to play on my Resonessence Mirus. Since it's the best way possible to listen to audio through the Mirus, it was a great test that would directly compare the SDM and SRC of the ESS 9018 chip with HQplayer's algorithms. Well I must say I was extremely impressed. Much smoother and analog like. It really opened my eyes to realize a large part of the compromise with redbook quality, is the SDM and SRC used inside resource constrained chips.
I had him convert 3 redbook tracks to DSD 128. I loaded them on an SD card to play on my Resonessence Mirus. Since it's the best way possible to listen to audio through the Mirus, it was a great test that would directly compare the SDM and SRC of the ESS 9018 chip with HQplayer's algorithms. Well I must say I was extremely impressed. Much smoother and analog like. It really opened my eyes to realize a large part of the compromise with redbook quality, is the SDM and SRC used inside resource constrained chips.
Thanks for this! Made me think further...
Yes, whilst upsampling using the best tools is the way to go but it will be not a good as a natively mastered version. Control Systems theory 101 tells us that in order to capture transients effectively the system bandwidth must be wide enough
A further illustration based on my own spin of this understanding:
An externally hosted image should be here but it was not working when we last tested it.
As you can see even at the mastering level the Redbook material is quite bandlimited to about 20KHz. Why, because of this Nyquist thingy!
For a 44k Fs the aliasing components creep into the audio band and must be aggressively filtered usually brickwall types at 20KHz or even less before digitizing. This has to happen at the mastering stage itself and the entire characteristics of the filter/sonics gets impressed onto the recording. You cannot change this afterwards, afaik. So the poor bandwidth results in the fairly subdued transient response. Upsampling/transcoding will not better anything from the system response point-of-view
Increasing the Fs gives some hope as the Filter Fc can be move away from the audio band and I reckon with 352k DxD this can be safely set at around 100KHz with simple LPF rather than brickwall type. So assuming recording engineers adjust the input Fc accordingly we can expect the transient response to approach that of an Analog system.
The point to emphasize here is that the apparent limitation of Redbook PCM is not actually due to the digitization process but the effect of the anti-aliasing filter happening right in the analog domain. In other words the impulse response characteristics is that right after the filter before going into the ADC itself!
Thanks for this! Made me think further...
Yes, whilst upsampling using the best tools is the way to go but it will be not a good as a natively mastered version. Control Systems theory 101 tells us that in order to capture transients effectively the system bandwidth must be wide enough
A further illustration based on my own spin of this understanding:
An externally hosted image should be here but it was not working when we last tested it.
As you can see even at the mastering level the Redbook material is quite bandlimited to about 20KHz. Why, because of this Nyquist thingy!
For a 44k Fs the aliasing components creep into the audio band and must be aggressively filtered usually brickwall types at 20KHz or even less before digitizing. This has to happen at the mastering stage itself and the entire characteristics of the filter/sonics gets impressed onto the recording. You cannot change this afterwards, afaik. So the poor bandwidth results in the fairly subdued transient response. Upsampling/transcoding will not better anything from the system response point-of-view
Increasing the Fs gives some hope as the Filter Fc can be move away from the audio band and I reckon with 352k DxD this can be safely set at around 100KHz with simple LPF rather than brickwall type. So assuming recording engineers adjust the input Fc accordingly we can expect the transient response to approach that of an Analog system.
Yes definitely not as good as native DSD. But much better than using the SRC and SDM inside the ESS 9018. However I do prefer 24/192 PCM without the HQplayer SRC/SDM through the Mirus.
Came across this for off-line conversion, (have not tested it myself yet):
AuI ConverteR 48x44 - Hi-End audio converter high resolution files
AuI ConverteR 48x44 - Hi-End audio converter high resolution files
Came across this for off-line conversion, (have not tested it myself yet):
AuI ConverteR 48x44 - Hi-End audio converter high resolution files
I tried that and found the algorithms to be not near as good as HQplayer.
The point to emphasize here is that the apparent limitation of Redbook PCM is not actually due to the digitization process but the effect of the anti-aliasing filter happening right in the analog domain. In other words the impulse response characteristics is that right after the filter before going into the ADC itself!
Actually, all ADC's work at around 5 to 6MHz from 1 to a few bits. So the
impulse response is dependent on decimation filter in ADC. The front end of
ADC can capture very wide bandwidth.
This is a very complex and deep subject when you look at it from a
complete signal chain POV. There is no 1 single thing that is really spoiling
the party for redbook, it's more like a combination of many gate crashers,
from the decimation filter, DS modulators in ADC, noise floor modulation
and then all the usual suspects from DAC side of things.
IMHO, I think that if we capture audio with a current SOTA 20 bit SAR
converter at a VH SR, say just below 1MHz with zero dig filtering and then
use that format to convert to whatever other DSD / PCM using powerful
software like Jussi's, we would have the best current capture quality -
subjectively speaking.
The industry got focused on ADC dynamic range and DS is the way to get it.
Terry
Came across this for off-line conversion, (have not tested it myself yet):
AuI ConverteR 48x44 - Hi-End audio converter high resolution files
That offline versus online upsampling is a very interesting subject, thanks for
the link. I haven't played with this much yet.
It's a trade off between quality and file size, keeping a library 384k PCM files
on your PC would start to chew up a lot of memory.
I wonder if a combinational approach would be a good compromise. For
example upsample files to 96k offline for storage at 96k then do the rest online to 192 or higher.
Just thinking.
T
memory is getting cheaper all the time, also its not to be even compared with big storages of CDs or, even, LPs on the shelves)).
memory is getting cheaper all the time, also its not to be even compared with big storages of CDs or, even, LPs on the shelves)).
Yes 5TB HDD is only just over $100 U.S. I wish Jussi would offer offline conversion software. I tried to convince him but he doesn't want to. I did find his offline conversions to sound better than online. Probably due to the CPU running at 1% vs 35% on my computer.
First on my list to convert would be my Beatles discography.
If Jussi is not coming to the party then we may have to look at Weiss-Saracon for quality off-line conversion ...$$$$
- Status
- Not open for further replies.
- Home
- Group Buys
- Direct Drive DSD