DSEE: Restoring the High Frequencies
I’ve been playing most of today with new Sony HAP-S1 “HDD Audio Player System”. I’m planning on writing a full review soon but thus far I’ve been focusing on the Digital Sound Enhancement Engine or DSEE feature. If you missed yesterday’s post, I’d recommend reading that before continuing with today’s post. You can click here to access it.
I’m struggling with the whole concept of developing a feature that re-generates high frequency information that is lost when original CD quality tracks are compressed into MP3 files (typically at around 128 kbps…recall that CDs have a bandwidth of 1411 kbps). I guess there are so many casual music lovers out there that have vast catalogs of poor quality, heavily compressed audio files that will believe that they can restore them to “their original quality” that including DSEE in their new line of players is worth the investment.
Why not just go back to the “original quality” CD and import them into the new hardware? Then you really would get the real deal.
The online information about the DSEE process makes it sound pretty promising. I talked about the fact that MP3 compression does a lot more than just roll off some of the highest frequencies yesterday. I promise a post that will address the damage that can be done during the 10:1 data compression scheme that Fraunhofer came up with many years ago and which launched the downloadable music era.
So I thought it would be interesting to see how good the DSEE process actually is. By looking at the spectra of the files along the way, we should be able to see whether the DSEE process actually does restore an MP3 files back to its original fidelity.
So here’s what I did. I took one of my favorite high-resolution audio tracks, the “Mosaic” track of Laurence Juber, and converted it to standard CD resolution (44.1 kHz/16-bits) from the original 96 kHz/24-bit stereo master (I used the Sonic Process application set at steep and noise shaped…which maintains the highest amount of high frequency information). This is the track that is available on the FTP site. The SRC files are available on the site as well (I know I still have to talk about the various types of sample rate conversion etc). Then I took the CD res file and created an MP3 128 kbps stereo file, which I transferred to the HPA-S1’s HDD.
I captured the playback of the DSEE processed MP3 file into my Pro Tools rig at 44.1/16-bits and created spectragrams at every file involved. Take a look at Figure 1 below:
Figure 1 – The spectra of “Mosaic” in various formats [Click to enlarge].
So does the DSEE process restore the high frequency content that was absent from the MP3 file? No, clearly it does not but it does add some information at the very high end. Let’s take a close look at the plot.
The green line is the high resolution original file and looks wonderful all the way out to 48 kHz. The red line in the plot represents the CD version. The amplitude is several dB lower than the original (more on that in an upcoming post). As we expect, it stops at 22.050, the max for a compact disc. The light blue line is the MP3 file. It tails off at about 17.8 kHz but doesn’t really resemble the uncompressed version after around 16 kHz. Those high frequencies are data devils with regards to space…lots of information up there that isn’t going to get through your ear buds anyway.
The purple line is the DSEE processed output from the HPA-S1, which does departs from the MP3 file’s spectragram at exactly the same place…16 kHz or so. And you’ll notice it also has some spikes that don’t belong (please forgive the different amplitudes…I lowered the DSEE processed file so that it would be easier to see). The segment of the purple line above 16 kHz can be attributed to the DSEE process. We’re talking about 4 kHz of bandwidth that was removed from the original CD and “restored” by the DSEE process at much lower amplitude and with a completely new shape.
It’s your call on this one…is it worth having a processor try to “fix” the problems of the MP3 format (or any other heavily compressed audio file format)? You have three choices: go back to better sources and get definitive copies at real fidelity, live with the degradation of the MP3 format or turn on DSEE and get some made up high frequency information added to the highest range of the audio spectrum.
It is utterly baffling to me that Sony should go down this path…………….or is it?
Sony have a “technology” clause in their contracts with their artists. If they wish to re-release an album from their catalogue in a “new” format, they ( Sony ) receive a larger percentage of the royalties. This way, they have a large bite of the software AND the entire chunk of the hardware. It is in their own interest to keep introducing “new” formats. DSD was, and should remain, their trump card. Save for people with the early PS3, no-one has been able to rip SACD. I would have thought that this would been their saviour?
Mark, If it was not for people like you, I would have thrown my system in the garbage years ago. My much loved “Nitty Gritty Surround” disc with the wonderful Jennifer Warnes track “Somewhere Somebody” restores my faith in the recording industry.
The “dumbing down” of audio is the equivalent of a movie director spending months on end waiting for the ideal “light” in which to film a scene and then agonise over the colour palate in the final print…only to have the production company release it in Black and White.
It must be soul destroying for a record producer and engineer to have their “children” treated in this way?
Many thanks again for your wonderful contributions and I look forward to the next posting.
Looks like Harman Kardon has something similar called Signal Doctor. I bet it sounds really, really good
“Signal Doctor a proprietary software solution that automatically analyzes and improves the audio quality of all types of compressed, digitized music sources. The technology leverages HARMAN’s decades of unmatched experience in music recording, signal processing and psychoacoustics to naturally restore the full, satisfying sound that is forfeited in the compression process.
Signal Doctor is available now in HARMAN’s new JBL Authentics Series wireless home entertainment sound systems (a CES 2014 Innovation Award winner), and HARMAN plans to launch Signal Doctor technology across a range of Home, Multimedia, and Automotive products in the coming year.”
— my comments include a huge portion of sarcasm
How does HAP-S1 compare to your Benchmark and OPPO? Does the fact that it does not need to connect to a (noisy) PC makes it better, SQ-wise? I am really curious.
I just bought the NWZ-ZX1 which also has the DSEE function. Although I completely agree that DSEE does NOT restore the original high end frequencies, there is still some merit to this technology(or algorithm!). I’ve actually been using the DSEE function for some 16/44.1 m4a files (CD standard) and the DSEE function does make some harsh sounding high notes sound a lot smoother (sorry for the lack of more scientific description). The NWZ-ZX1 is 24/192 capable so I am assuming the DSEE fills in the gap between 16/44.1 and 24/192. I find this is similar to the upsampling function of some very expensive CD players.
I’m not a fan of these restoration algorithms. If you think of them as some processing that can be used to tweak a sound to your liking…then great.
Good morning,
do you have notice differences between DSEE software processor and Dolby Atmos software processor in term of equalisation or in term of saving some high frequencies?
Thanks in advance
Alberto, I can’t say I have enough experience to properly answer your question. Although, I cannot endorse the kind of processing that SONY does with DSEE.
I use the DSEE option on my Sony Experia Phone. Other settings include a concert hall 3D affect and an option that includes equalization settings. After playing options, settled on DSEE. While it has no EQ options and reduces the overall volume, it did seem to add warmth and balance to the music. I am less conscious of hearing a “digital file”. Good enough that I traded the desired extra volume and EQ control to just let DSEE do its thing. As it never gets load enough I do use a now discontinued Boom 360 Cloud stick. This provides mild amplification and psychoacoustic properties (EQ and sound stage enhancement). This probably sends audiophiles into a paroxysm, but Boom combined with DSEE provides great excitement on my runs, and also suffices at night time with a bourbon. I am a budget audiophile type-NAD and entry level Grado’s. So care about sound but never getting around to fixing the Grado headphones. As DSEE is for wired headphones, I suspect a future DAC/AMP will circumvent this, so I am good with this system for now. DSEE does have a purpose.