Not that Cleaner:
This is it:
Digital audio reproduction has many believers and I am definitely one of them. To this day, even the best DtoA converters have their weaknesses. We hear differences never thought possible with ever lower levels of clock phase noise. We can hear small differences between different DtoA chips or discrete FPGA converters. At some point, one wonders whether there's a different way to deal with nonlinearities that ALL DACs produce. Dr. Uli Brueggeman has been thinking about these problems for some time. We've previously discussed his method of introducing frequency dependent crosstalk. You can read more about that here:
http://www.whatsbestforum.com/showthread.php?15471-Crosstalk-digital-more-like-Vinyl
The idea is to introduce frequency dependent crosstalk so as to reduce the load on the brain when decoding phantom sources. The result can be a more relaxed listening experience and even better stereo imaging. The "Cleaner" was developed to address the difficulties of localization. In particular, the "Cleaner" is designed to deal with left/right nonlinearities produced by the DAC. The theory according to Uli is that any small nonlinearity between the R/L channels stresses the brain which is attempting to localize the phantom sources while listening to a stereo image. In the worst case scenario, if left and right channel do not correlate our brain has to do a lot of decoding work.
The "Cleaner" is a mid/side analog decoder designed to high quality standards. The signal in the digital domain is converted into a mid/side pair of channels and sent out the DAC into the "Cleaner." The "Cleaner" then decodes it in the analog domain. (analog in and analog out, 1:1 gain, low output impedance output).
Mid/side encoding is not new technology. It's become very popular these days with many mastering engineers. Of course, they are using it for a completely different purpose. Mastering engineers can use mid/side to enhance spatiousness and adjust vocal in middle of the stereo image. The "Cleaner" does not change the original waveform.
In Uli's words, here's how mid/side encoding and decoding work to reduce any nonlinear distortion caused by the DAC:
"m/s processing means:
M = (L+R)/2
S = (L-R)/2
Then you can reconstruct L/R by
L = M+S
R = M-S
If the processing is perfect then you get the original signals back. This is the intention. There is no reason to change the original content (whereas with mixing and mastering there is a tradition to apply m/s processing with changes of M and S)
Now assume that during DA conversion you get some signal dependent dirt. So you may get
Lanalog = Ldigital + Ldirt
Ranalog = Rdigital + Rdirt
Ldirt and Rdirt do not necessarily correlate.
Now instead of feeding the DAC with Ldigital and Rdigital you feed it with
Mdigtal = (Ldigital+Rdigital)/2
Sdigital = (Mdigital - Rdigital)/2
The analog signal thus is
Manalog = Mdigital + Mdirt
Sanalog = Sdigital+Sdirt
Mdirt and Sdirt do not necessarily correlate.
With m/s decoding now you get
Lanalog = Manalog+Sanalog = (Migital+Sdigitla) + (Mdirt+Sdirt) = Ldigital + (Mdirt+Sdirt)
Ranalog = Manalog-Sanaog = (Mdigital-Sdigital) + (Mdirt-Sdirt) = Rdigital + (Mdirt-Sdirt)
So the output signal is the original signal plus dirt again. We cannot avoid the dirt. But now the resulting dirt on left and right side has some correlation as both sides contain Mdirt and Sdirt.
In reality the decoding circuit cannot be expected to be 100% perfect, it will also add some crosstalk. As we have learnt from flow the crosstalk is not necessarily bad. Anyway the decoding circuit has to be designed carefully.
So you understand, the basic idea is very, very simple. But is has never been published before. So I have built a prototype and I have been excited by the result. Then I have got the chance to visit ABACUS for a three day workshop for demostrating Acourate. I have decided to test the prototype by demonstrating its function to the people. It has been a full success. The audience has finally baptized the circuit as "cleaner". And ABACUS and I decided to start a product AcourateCleaner."
I have tried many preamps over the years. They were all some of the best out there. Everytime, I felt that the preamp reduced resolution compared with my DAC direct to amp. So, I've never used a preamp. The "Cleaner" is not a preamp. It simply decodes the m/s signal back into original 2CH signal and drives the amps from it's low impedance output. What does it sound like?
The image is wider and more spacious. The usually harsh recordings are rendered more relaxed with increased detail retrieval. The really incredible improvement is with vocal located at the middle of the soundstage. Voices are deeper, more natural, more articulate and much more relaxed sounding. I highly recommend it for digital folks searching for a more analog feel to their favorite music.
Michael.
This is it:
Digital audio reproduction has many believers and I am definitely one of them. To this day, even the best DtoA converters have their weaknesses. We hear differences never thought possible with ever lower levels of clock phase noise. We can hear small differences between different DtoA chips or discrete FPGA converters. At some point, one wonders whether there's a different way to deal with nonlinearities that ALL DACs produce. Dr. Uli Brueggeman has been thinking about these problems for some time. We've previously discussed his method of introducing frequency dependent crosstalk. You can read more about that here:
http://www.whatsbestforum.com/showthread.php?15471-Crosstalk-digital-more-like-Vinyl
The idea is to introduce frequency dependent crosstalk so as to reduce the load on the brain when decoding phantom sources. The result can be a more relaxed listening experience and even better stereo imaging. The "Cleaner" was developed to address the difficulties of localization. In particular, the "Cleaner" is designed to deal with left/right nonlinearities produced by the DAC. The theory according to Uli is that any small nonlinearity between the R/L channels stresses the brain which is attempting to localize the phantom sources while listening to a stereo image. In the worst case scenario, if left and right channel do not correlate our brain has to do a lot of decoding work.
The "Cleaner" is a mid/side analog decoder designed to high quality standards. The signal in the digital domain is converted into a mid/side pair of channels and sent out the DAC into the "Cleaner." The "Cleaner" then decodes it in the analog domain. (analog in and analog out, 1:1 gain, low output impedance output).
Mid/side encoding is not new technology. It's become very popular these days with many mastering engineers. Of course, they are using it for a completely different purpose. Mastering engineers can use mid/side to enhance spatiousness and adjust vocal in middle of the stereo image. The "Cleaner" does not change the original waveform.
In Uli's words, here's how mid/side encoding and decoding work to reduce any nonlinear distortion caused by the DAC:
"m/s processing means:
M = (L+R)/2
S = (L-R)/2
Then you can reconstruct L/R by
L = M+S
R = M-S
If the processing is perfect then you get the original signals back. This is the intention. There is no reason to change the original content (whereas with mixing and mastering there is a tradition to apply m/s processing with changes of M and S)
Now assume that during DA conversion you get some signal dependent dirt. So you may get
Lanalog = Ldigital + Ldirt
Ranalog = Rdigital + Rdirt
Ldirt and Rdirt do not necessarily correlate.
Now instead of feeding the DAC with Ldigital and Rdigital you feed it with
Mdigtal = (Ldigital+Rdigital)/2
Sdigital = (Mdigital - Rdigital)/2
The analog signal thus is
Manalog = Mdigital + Mdirt
Sanalog = Sdigital+Sdirt
Mdirt and Sdirt do not necessarily correlate.
With m/s decoding now you get
Lanalog = Manalog+Sanalog = (Migital+Sdigitla) + (Mdirt+Sdirt) = Ldigital + (Mdirt+Sdirt)
Ranalog = Manalog-Sanaog = (Mdigital-Sdigital) + (Mdirt-Sdirt) = Rdigital + (Mdirt-Sdirt)
So the output signal is the original signal plus dirt again. We cannot avoid the dirt. But now the resulting dirt on left and right side has some correlation as both sides contain Mdirt and Sdirt.
In reality the decoding circuit cannot be expected to be 100% perfect, it will also add some crosstalk. As we have learnt from flow the crosstalk is not necessarily bad. Anyway the decoding circuit has to be designed carefully.
So you understand, the basic idea is very, very simple. But is has never been published before. So I have built a prototype and I have been excited by the result. Then I have got the chance to visit ABACUS for a three day workshop for demostrating Acourate. I have decided to test the prototype by demonstrating its function to the people. It has been a full success. The audience has finally baptized the circuit as "cleaner". And ABACUS and I decided to start a product AcourateCleaner."
I have tried many preamps over the years. They were all some of the best out there. Everytime, I felt that the preamp reduced resolution compared with my DAC direct to amp. So, I've never used a preamp. The "Cleaner" is not a preamp. It simply decodes the m/s signal back into original 2CH signal and drives the amps from it's low impedance output. What does it sound like?
The image is wider and more spacious. The usually harsh recordings are rendered more relaxed with increased detail retrieval. The really incredible improvement is with vocal located at the middle of the soundstage. Voices are deeper, more natural, more articulate and much more relaxed sounding. I highly recommend it for digital folks searching for a more analog feel to their favorite music.
Michael.
Last edited: