audiogene

I recently purchased a Pioneer SC-LX87-K which was adverted to be capable of native DSD playback via HDMI, USB/MemoryStick, USB/ComputerAudioDriver and Network (DLNA server/UPNP). To cut a long thing short I found out the beast only supports 2.0 DSD natively over any of the above interfaces. Even the 5.1 DSD stream of a BD player connected over HDMI would be converted to multichannel PCM (@ 6x88 or 176kHz). WTF! Apart from this drawback the Pioneer is a great sounding device with digital class D power amp stages - actually the first class D amp sounding better than my old Krell KSA-250s! This leads me to the question wouldn't it be high time for an integrated digital amp capable of playing back 5.1 DSD from commonly used digital input sources? Such an integrated design would not need any format conversion nor decimation filters. There would only be needed the digital low pass filter in the class D stage. PCM sources would have to be converted to DSD instead - better than half-hearted oversampling and extra PCM processing of today's devices. Well, I had a dream...

Maybe, it is, maybe not. I think we all have the same question but we do not know for sure because nobody ever did a double blind listening test to samples with different encodings and/or filtering. You will not get a definite answer from any of us in the world, so, the only way to handle this situation is to carry out a listening test yourself, or to make use of the setting and forget about the infinitesimal differences you may achieve with different filter settings resp. conversion algorithms. Remember if you hear some differences between different settings you will not be able to judge them quality-wise because you do not know how the original recording sounded... From an other point of view your filter and conversion setting may be absolutely superior to the conversion-only method (like Audirvana offline conversion) where no previous filtering is applied. The spectrum of the latter method contains too much energy above 22 kHz which to almost 100% stems from conversion artefacts. These will not be audible as such but in the amplifier there will be created all kinds of non-negligible intermodulation distortion which will be in the audible spectrum below 22kHz. Intermodulation distortion will be created in the tweeter, too, because it may be driven to non-linear ranges due to excessive heating and/or magnetic saturation. You may easily recognize this kind of distortion with recordings of single acoustic instruments like a recorder (resp. a flute), a triangle or a piano. Side note: Long time ago when 1-bit conversion was invented some of the inventing audio engineers carried out listening tests (regrettably, in non-documented ways) and some of them said it was not worth while the whole efforts. At that time 16/44.1 was the ubiquitous digital format... Years later 1-bit conversion led to SONY's SACD format, not beacuse of audio superiority but because of business competition and a splendid idea how to force the studios into a completely new audio production chain (DSD...)

Haha, how true (regarding the conversion artefacts)! And, you have to believe in God that the DSD 64/128 contents are true HD. Amen. Oh my... do you have a clue about how todays studios are stuffed with digital compressors, expanders, and all kinds of "digital nonlinearity" devices to add some "oomph" to certain performances? This is what I had in mind when speaking of "excessive use of digital audio technology". In this calamitous chain the AD- and DA-transfer steps are as insignificant as can be. Good point, indeed. That's what I keep saying to my students in the beginning of each semester. But I also keep saying that they have to theoretically understand what they are doing when they make use of a digital device (apart from turning knobs). Todays students do have a different way approaching technology. Whe I was a student everything had to be understood theoretically and then turned into practice. Today technology is ubiquituous and everybody makes use of it without prior knowledge. So, after hours and hours of fumbling and squeezing everybody will get a result but will not know why and how it worked. I am aware of the fact that this may be called an artistic knowledge and may even found new directions of art but here we have the same problem like, say, in the art of painting. Everybody who throws some pouches filled with colors across a canvas and then randomly smears the color will achieve a result qualifying the "drawing" as a work of art. Some people may like it and some not. But, if you intend to paint a drawing in the style of a classical artist like, say, Rembrandt van Rijn, your work will easily be identified as a work of a layman! (Even then there may be some persons who like it...) Well, you can apply this 1:1 on todays music business. While in pop music business any kind of novel manipulation techniques is almost crucial for getting some attention, this is absolutely contrary in classical music business and many styles of jazz music. That's where the "audiophile" aspect steps in whose eclectic paradigm is to conserve and transfer the audio performance as realistic as possible to the listeners environment. And that's where our discussion starts!

This is exactly the kind of task I proposed on a different CA forum where audiophile discussion about different DSD conversion types is going haywire (as to my opinion). Apples are compared with oranges in that different audio media of "the same" album are compared without knowing about potentially different mastering, pre-mastering or even studio editing on the particular media. Everybody anticipates "more airiness" or even "better soundstage" when listening to HD media even if this objectively cannot be the case because, e.g., some SACD was produced using the untouched 44.1 kbits/s CD master of a recording. As I found out this is not very unlikely with HD re-issues of top-selling CD albums, - shame on the recording industry. I did find this either on SACDs and HD downloads selling at a 4 times premium to the CD. I will explore your sound samples as far as I am able to directly reproduce them with my equipment (without format conversion), but I am sure this will take some time. If there are many skilled persons doing the same in their listening environments this might create some more insights into different things, but I don't believe it will make a foundation regarding the original question, because there are too many individual parameters involved. The most prominent of them are different listening rooms and amp-speaker-configurations with totally different characteristics. I am not sure that there is a linear function between the audio quality of a configuration versus audio format/resolution. There is even no measure for audio quality. Since I have close connection to scientific audio community I will discuss the topic, anyway, because I feel we urgently need statistically valid statements about the discernability and characterization of different digital consumer formats. As long as the scientific audio community is not able to deliver here digital audio myths will spread like in the analogue domain. I feel it even may have started with the introduction of DSD...

Thank you very much for your helpful comments on my questions! Now that I feel a little more confident about my DSD->PCM findings, I did some research into some of my SACDs to get an idea of the real amount of HF noise after Multistage conversion in relation to usable "real" audio signal in the range above 22kHz. The reason for being that nit-picky is I own audio components which are very capable in the ultrasonic range - my Devialet runs straightedge to 120/150kHz and the Beryllium tweeter of my Focals is still putting out acoustic energy at around 40kHz (apart from thermal energy...). The results of analyzing some 20 SACDs (bitperfect rips in fact) really hit me. Over 80% obviously do not contain any real audio signal in the range above 22kHz originating from the recording!! Some of the "DSD masters" seem to stem from unaltered 16bit 44.1kHz PCM CD masters wich were padded to 24bits and some of them were dithered additionally. Oftentimes there were added artefacts from inferior studio devices peaking +20dB out of the increasing noise floor (above 22 kHz)! There were only 3 SACDs containing "real" audio above 22kHz - and only one contained a "reasonable" amount of real audio spectrum clearly differentiating from the increasing noise floor. Just for illustration I will append three typical frequency analysis diagrams of real SACDs - DSD64->PCM 88.2 conversion done with A+ Multistage64 decimation: Dire Straits - So Far Away (obviously from CD master) Fleetwood Mac - Second Hand News (obviously dithered PCM master) Diana Krall - S'Wonderful (early plain DSD master w/ telling peak artefacts) After all the audiophile discussion going on for HD audio (in connection with big spending) I have to admit I am somewhat upset. While everybody is manifesting additional "airyness" or even better soundstage due to extended HD bandwidth I strongly question these statements as nothing more than audiophiles' creed or simple myths. It goes without saying that two different media of the same recording event would sound different, - not because of the fundamental superiority of one medium to the other, - but because they are the simply the outcome of two different process chains with different devices and settings. Be it as it may, I have not read or heard about a double-blind listening test setup carried out under scientific preconditions (producing systematically and statistically valid results). Such a setup would have to chose a state-of-the-art recording in pure DSD (64 or 128, whole studio chain) which is presented to the probands (1) as a plain DSD stream (up to the D/A conversion), and (2)...(n) with different DSD->PCM conversions. I would be really interested in the results of such an experiment and I am thinking about involving a well-known university because at least, the audio engineering students would get a little more "environmentally" aware of todays excessive use of digital audio technology in the studios. Regarding my personal interest in current HD audio I must admit that I cooled down quite a bit...

I would like to revive this thread because there are some open questions which I have not been able to answer by now. I am talking about offline conversion of DSD files to PCM with Audirvana Plus (Menu "Add Files to iTunes"). If I got everything right Damien implemented four fixed conversion rates if and only if one chooses "64bit Multistage" (44, 88, 176 and 352 kHz PCM). Since there is no white paper or manual available the effect of the "Audio Filter" menu section is completely nebulous to me under these circumstances and I have the strong suspicion the whole iZotope stuff is not available for 64bit Multistage conversion?! I made some tests with different filter settings and after evaluating the fft graph they in fact looked identical. Can anybody enlighten me on this? What I originally wanted to do is offline conversion of DSD-Files to 88.2 PCM with the theoretically best algorithm. Further I wanted to apply some digital filtering of frequencies above 30..50kHz to get rid of the massive HF noise after conversion (which is usually done by analogue filters in SACD players). Obviously the iZotope parameters are not working here (maybe iZotope is not working at all when doing offline conversion?). Maybe anybody can shed some light on this, too. Last but not least I would like to ask some knowledgeable people here as to what they recommend as a perfect filter setting (acoustically) for this low pass filter. Analogue low pass filters in SACD players tend to be not steeper than 24dB with passband ripple minimized - but in the digital domain we have more parameters to play with.