Analog Tape As High-Resolution? No Way.
I’ve been working with analog tape a lot over the past month. My Nagra and Ampex 440C 1/4″ machines have been put back into action after many idle years. The Ampex was parked upstairs for over 10 years until about a year ago. Following the RMAF show last fall, I decided I should dust it off, get it overhauled and sell it to an analog tape devotee. The machines are getting top dollar. It’s been parked just outside of my wife’s office since then. When she arrived the other day, she noticed that it had been moved and I’m sure she was hoping that I had finally junked it…but no, it was in the studio in active use as an editing platform. I knew there was a reason that I had kept it all these years. Yesterday was definitely a “living in the past” moment.
Figure 1 – Mark Waldrep (aka Dr. AIX) editing the master tape of Christian Jacob for a vinyl LP release.
So are other audio engineers, producers and the dedicated group of analog high-end users that prefer the sound of analog tape living in the past everyday? Well yes, in a way they are. And I can appreciate why after spending so much time listening to the sound of my recent piano recording of Christian Jacob. If pressed, would I choose the analog tape over the 96 kHz/24-bit HD PCM tracks of the same material…no way. But I do understand the appeal and the “color” or “timbral” modifications that analog tape imparts to a track. Is this why so many equipment designers and music sites use some form of the phrase, “sounds just like the analog master tape”?
Audiophiles should acknowledge that analog tape is simply one flavor among many different formats. It served the music industry very well for many decades and yes; it still has a place in this business. But it is not a high-resolution audio format knowing the limited dynamic range, moderate frequency response and other distortions that it suffers from. I use as my point of comparison the clarity, accuracy and transparency of HD PCM at 96 kHz/24-bits or better. This is a case of different strokes for different folks.
Yesterday, I lauded Cookie Marenco for the marvelous recordings that she produces. She prefers the sound of analog tape and DSD to PCM. She’s stated that when a group of audio professionals evaluated a recording made to analog tape, PCM and DSD, that the analog was chosen by 100% of them. I would have to get more information about that test because I’ve experienced just the opposite in a test we did at Snow Ghost Studios in Whitefish, Montana. Using state-of-the-art microphones, preamp, converters and recorders, Peter McGrath (a well-known recording engineer and representative for Wilson Speakers), myself and a group of about 6 other professionals rather quickly ruled out the playback from a new Studer A 820 24-track analog deck through an SSL Series 9000 console to VTL monoblock tube amplifiers to Wilson Alexandria speakers (total value of the signal chain…well over $500,000). The analog tape just didn’t have the sparkle and detail that the digital formats did. I wrote about the weekend at Snow Ghost some months ago.
As promised, here’s the analysis of the free “Jimmy and the Crows” track that Cookie recorded and which is available free of charge at the Blue Coast Records website. I downloaded the MP3, 96/24 WAV/FLAC (which are identical) and both versions of the DSD format in 2.8 and 5.6 MHz (which are also identical). The annotated spectragraph Figure 2 is below:
Figure 2 – The spectragraph of “Jimmy and the Crows” from Blue Coast Records as recorded and produced by Cookie Marenco. (Click to enlarge)
We all know that analog tape has audible amounts of hiss associated with it. During the heyday of analog tape, people like Ray Dolby (who passed away last week) created systems for both professionals and consumers to minimize the amount of hiss we had to suffer through. For professionals it was Dolby A (and later Dolby SR) and for consumers it was Dolby B. Looking at the spectragraph, you can clearly see the hiss. And it is audible when listening to the track using both speakers and headphones. The plot on the right and the “purple haze” in the left hand graphics is the contribution of both the hiss AND high frequency noise associated with DSD or 1-bit encoding and the need to apply fancy noise shifting to the signal.
The MP3 at 320 kbps actually sounds very, very good. The high frequency noise is gone but there are no ultrasonic components to the sound. I would guess that most casual listeners would not be able to tell the MP3 file from a CD resolution version or the high-resolution PCM and even the DSD files.
The analog tape converted to 96 kHz/24-bit PCM sounds the best and is the most accurate representation of the original. The sound is warm and smooth throughout the audio band and only hints at the hiss compared to the DSD files. If I had my Nagra slung over my shoulder and Cookie would make me a copy of the analog master, I’d prefer to listen that way. The reasonable alternative to my mind is the accuracy of a HD PCM transfer…and that’s what you get with the 96 kHz/24-bit FLAC or WAV. There is absolutely no difference between the FLAC version and the uncompressed WAV file. During a realtime decoding of the FLAC file, it might suffer on an underpowered machine, but to say that a FLAC file “sounds” different than a WAV file of the same source is simply untrue.
So let’s take a look at the DSD versions. Cookie told me that she records the output of her analog mixing desk to a Korg MR 2000 DSD recorder capable of 2.8 or 5.6 MHz 1-bit recording. In the plots on the spectragraph, the two versions are identical. There is no benefit from using the 5.6 MHz sampling rate over the 2.8 MHz standard. In fact, when I first did the analysis, the lines were exactly the same and I wasn’t sure whether I had actually completed the scan. Only when I added .5 dB to the DSD 128 file, was I able to see the spectra of both versions.
The DSD versions don’t have to work very hard. There’s not much in the way of ultrasonics in the analog original file (they top out at around 30 kHz) and of course, there’s the hiss. In reality having any recording scheme extending beyond 30 kHz is unnecessary in the case of an analog tape source. The DSD files are rolled off at 42 kHz (a PCM file at 96 kHz would extend to 48 kHz). And then there’s the additional HF noise caused by the DSD encoding. Who would reasonably want that?
Cookie is a staunch supporter of higher prices for DSD and in general for her “premium quality” productions. In fact, I may try converting some of my HD PCM recordings to DSD and elevate the pricing. I’m already considering issuing limited releases on 2-0 track analog tape. The philosophy of “give the consumers what they want” might just be a better way to go.
However, I’ve based my reputation on a different way of operating. I prefer to “give consumers the very best that I can offer at a reasonable price”. I didn’t get into high-end audio to get rich. I started AIX Records and iTrax.com to show what was possible with no compromise audio production techniques AND new high-definition audio standards.
So the ultimate question still stands. Is anything that goes through an analog tape production step capable of high-resolution audio specifications? The answer is clearly no. As wonderful as Cookie’s recordings are, they are not high-resolution…but they are very high-quality standard resolution recordings.
Hi Mark, thanks for this article.
” If I had my Nagra slung over my shoulder and Cookie would make me a copy of the analog master, I’d prefer to listen that way. ” Could you explain a little more about why this is so, please? Is it because you would prefer the sound quality?
Because, the tracks that Cookie produces are made on analog tape. Having access to the source (or as close to the source as possible) is a good thing. I have nothing against analog tape…it was great for its time. But most of us have moved on to something that is more accurate and much more efficient.
some people like a bit of hiss , analog is magnetic power & magnetic flux is a magical medium while digital
is made to be cheap and practical , got two nagras .
Someone who can not hear the difference in sound between 96 24 and same material from the original tape
is an average engineer , not born for the task .
Grant, I would challenge you to perceive a difference between master analog tape and a good 96/24 bit clone. There is nothing lost in this sort of transfer. Analog tape (magical magnetics?) has a sound and less fidelity than an equivalent 96/24 digital PCM version. I’ll take the greater dynamic range, extended frequency response, and convenience any day. Having been a mastering engineer for hundreds of albums and producer of award-winning audiophile recordings, I don’t consider myself an “average engineer”
Hey Mark,
FYI, that wasn’t tape hiss.. it’s the noise floor from my method of recording. Because I choose to place the mics farther away, that “hiss” sound can be heard above any noise coming from the tape. Also, because of my use of 3-5 different reverbs, there are times I can only optimize the gain a certain amount that can still be heard above any potential tape ‘hiss’. Also, there is a sacrifice using Dolby SR, but we do and for 15 years in our equipment chain, tape hiss is a non issue. Someday when you’re in the Bay Area, come by and I’ll demonstrate that the tape ‘hiss’ is a non issue, but the noise floor isn’t.
Before you transfer your 9624 to DSD, let’s chat. I’ve been a pretty staunch supporter of knowing the ‘provenance’ of a recording before increasing the price for DSD. Anyone can get Audiogate’s converter for free and do the same as you’re planning to do.
As you said, I’m not sure if the music customer cares, but I care. At least for now. A larger file in DSD, comes with additional costs of bandwidth for delivery and customer service issues, which is part of why we charge more for DSD.
About pricing… if you’re on our mailing list for Blue Coast Records, you’d know that our customers are always offered discounts and periodically 2 for 1 downloads. 🙂 No one is getting rich owning an audiophile label, my friend. You must know what I’m talking about. It’s love, not money. If you’re making money, you can sell my discs, too.
It’s an unpopular opinion, but I think that recorded music is highly undervalued. Musicians and small labels have been trained to believe that the $9.95 they pay at Walmart actually covers the cost of recording, musician fees, promotion, etc. What’s happen is that less music is being purchased, the worse the production sounds. Studios going out of business. Great engineers call me all the time for a job, which I don’t have to offer. Music sales are less than 50% of what sales were at their peak more than 10 years ago.
Lowering the price is not the answer. I believe that choice is the answer. At Blue Coast Records, we offer CD quality WAV at $1.50- $2 per single. The 9624 and DSD are priced higher. Someone can download the 44.1 and get audiogate free to convert to DSD if they don’t care about the source.
This is not a hobby for me. I have a crew that works here tirelessly for hardly any money. I’ll be paying the rest of my life for the equipment I need to make a great recording. It should be known that starting a label is not the way to “get rich”. I will say, though, selling SACDs and CDs to China is much more lucrative than DSD downloads and keeps our doors open. We’ve developed a custom 24K Gold CD we call MQD that is selling better than DSD downloads.
You posted a shot of my studio in the your last blog. Do I look rich? LOL. Common, Mark, we’re both in this crazy business. Have a little heart. 🙂 I’m sure you know the joke.. How do you make $1 million in the music business? Start with $2 Million.
How about tomorrow making the Chesky Brothers the subject of your blog. :0 Love you, Mark!
Cookie, I have called out the misrepresentations on HDTracks.com many times and posted spectragraphs as well.
Hi Mark,
Thanks for your comments, this discussion made me think of a question I’ve had lately about the proper way to compare DSD vs PCM using frequency spectragraphs. Eventually I will get a proper DSD DAC to perform fair auditory comparisons, but in the meantime, I’ve been relying on JRiver Media Center on a PC for playback. It also has a converter to PCM WAV using a low pass filter (48kHz @48dB/octave), which I then open in Audition for comparison to 96/24 PCM.
I’ve been wondering if this is a fair comparison to PCM? There might be biases against DSD during conversion with different filters or differences in software vs hardware DSD-to-PCM. Also do some DSD DACs skip PCM phase and go directly to analog?
The reason I bring this up is because my DSD to PCM spectragraphs for Cookie’s DSD sample doesn’t match in the high frequency area as the one posted above; I see no roll-off. Is there a right/wrong way to perform the conversion?
Hi Mark,
The comparison between PCM and DSD spectragram doest not seems to be fair. While I understand how to plot directly a spectragram of a 24/96 PCM file (using Audacity for instance), I do not understand how the DSD spectragrams in the above diagram are realized. For sure there must be some untold DSD to PCM conversion before creating the DSD spectragrams, either in digital (eg. using Korg Audiogate) or in analog using a DAC and then an ADC to filter and resample at 24/96…
I am pretty sure by using Korg Audiogate, I can get different spectra at frequencies above 42 kHz, by converting DSD to 24/176.4 using no roll-off, soft roll-off (-3dB at 50 kHz) or sharp roll-off (-3dB @ 42 kHz) options.
I have the feeling that the 42 kHz roll-off exhibited in your DSD spectragrams – on which you base the conclusion of the superiority of PCM over DSD) is more attributable to this underlying DSD-to-PCM conversion rather than to DSD by itself.
Many thanks if you can comment on this.
I’ll give you both the short answer and prepare a post dedicated to the whole subject of spectragraph creation and analysis. I download the DSD files on my MAC as a .dff file. I play them back using Audirvana Plus through my Benchmark DAC2, which has been reviewed as a great DSD DAC. The balanced outputs of the Benchmark DAC2 is taken at a calibrated output level (using a reference 1 kHz tone) to the input of my Nuendo HD PCM system through the Euphonix 96 kHz/24-bit ADCs. Now I have a 96 kHz/24-bit PCM file of the original DSD file that I can use in Audition to prepare the Spectragraphs.
The rise in the HF noise happens in the available frequency range of the 96 kHz samples (up to 48 kHz or Nyquist). The roll off are happening in the DSD file well below the LPF that are part of the DACs and are typical (and acknowledged by the DSD camp)..
The fundamental question is whether the HF noise matters at all. For me it does…for others maybe not.
Hi Mark,
The roll off are happening in the DSD file well below the LPF that are part of the DACs
But what are the characteristics of the low pass filter in the Benchmark DAC2 that you use to convert DSD to analog ?
It looks like at the contrary that the roll-off @ 42 kHz is the exact signature of the LPF of the Benchmark DAC2, indicating that both DSD 2.8 MHz and 5.6 MHz are processed by the same filter in the DAC2.
On my side, I converted the “Jimmy and the Crows” DSF file to PCM 176.4 kHz using Korg Audiogate, with the 3 options – no roll-off, soft roll-off and sharp roll-off – and I got very different high frequency spectrum for each option.
I can post them somewhere if you are interested.
I would love to see them. I could be creating a problem given the nature of the Benchmark but I doubt it. I’ve done DSD analyses that don’t show the same roll off characteristics. Since DSD mandates a LPF be applied below 50 kHz, I ‘m pretty confident that these plots are showing the actual files.
Hi Mark,
At last, here you are : http://leblogdejn.wordpress.com/spectrum/
You’ll see the results of 4 different options in Korg Audiogate to convert DSD to PCM : 88.2 kHz, 176.4 kHz sharp roll-off, 176.4 kHz soft roll-off and 176.4 kHz no roll-off (you can click on the pictures to view them larger).
I am not an expert to draw conclusions, but your comments are more than welcome.
Were you recording at 30 ips?
At several recent consumer electronics shows, an analog tape venue was awarded best sound at show may have been the Tape Project). If it sounds better, i.e., more real, than any other medium, why not embrace it.
I recorded at 15 ips, which is by the way the speed of The Tape Project’s releases as well. I do embrace analog tape…I’ve been recording using analog tape for many decades. Award’s for “best sound” notwithstanding, it has a sound that some people enjoy…including myself. What I don’t enjoy is when people insist that it’s the “best sound” unless their definition of best sound involves tape hiss, crosstalk, print-through, speed variation, limited dynamic range and frequency response.
In my opinion, it doesn’t sound more real than another medium. I find HD PCM more accurate to the output of the microphones and console than any other format.
From another college professor.
OK. Thank you!
Hi Mark,
I’ve packed my away my collection of cd’s a few years ago. I found the digitized music very uninvolving. It was not a question of the convenience, distortion, noise or frequency response, of which digital can be good but an issue of the timber of the musical notes and a sence of the presence of the music in the room. Digital just sounds very “flat” to me. Digital sounds ok through computer speakers and headphones. But when you expose it through better speakers in a room and try to get a sense of the actual instruments, acoustics, etc. then its very different. its not merely reproducing a sum of different frequencies, is the pahse and time relationships, the wavefront attacks and decays that are associated with the music microdynamics that make music interesting.
We have access to high res downloads and these can be an improvement in some cases on 16 bit cd or DVD sound. But todate the essential elements that I am looking for, a sense of presence and 3-D localization of the music in our listening room, is not there. I tend to listen to a few songs and walk away to do other activities.
Over the last few years I have been investing in 3-head reel-to-reel tape decks as well as some notable 3-head cassette decks (HK CD-491, Nak 700 and LX-5). All of these have extended frequency response to beyond 25kHz as well as low distortion and a presence of very fine inner detail without the surface noise on vinyl. The tape hiss is typically masked by the room background noise. It may be more evident using headphones. However, the resulting music is very enjoyable, involving and entertaining. Its something that I want to sit down and listen to. A prime example is Miles Davis “King of Blue”. I have it on CD, vunyl and cassette. On CD Miles Davis sounds horrible, his music is reduced to a series of squonks. On vinyl, the background noise can be distracting burt the musically notes are closer to real horns with the fine reverbations and details. On my cassette with Dolby HX Pro its simply wonderful with the HK CD-491 or the Naks, very lyrical and finely detailed.
On my side, I am back to tape 100%. Luckily a lot of others aren’t, providing good availability of considerable old stock music on tape on the internet at a modest cost. For newer music, its mainly 180 or 200gm vinyl. However, there is also siome new music comming out on tape. We can always transfer to other media (digital) for portable playback, etc.
Cheers!
Roman
PS. I keep my computer system and my music systems totally separate. We have accumulated music systems representative of each decade from the mid sixties to about 2006 (our most modern system). What is surprising is that the all-discrete electronics (tube or Ge/FET transistor) and the HiFi speakers (alnico magnets) available in the mid sixties for music are still very competitive and enjoyable, with some refurbishment, to higher end modern sound systems (excluding SNR).
There are lot of individuals that feel that analog tape is the ultimate format for all of the reasons that you pointed out in your response. Having spent a large part of my engineering career working with analog tape, I can understand your attachment.
However, I’ve moved past analog tape in my own listening and productions…opting for high-resolution PCM digital. I’m thrilled with the sound of my own productions, get reviews from magazine and customers that any producer/label would die for and see no reason to limit myself to any smaller a niche than I already occupy.
Hi Mark,
thanks very much for the follow-up. Tape for music is definitely a niche today.
Digital is definitely attractive for its conveniece and the fidelity of the data storage, and will provide the larger market in the foreseeable future for music sales. However, perhaps not for music hardware as its mainly used with i-phones, etc. HD digitized music can sound acceptable if the mastering is done correctly. Shannon sampling is a requirement but not sufficient as real music is intrinsically time varying (Shannon sampling is strictly applicable only to cw time invariant signals, very boring to listen to). Additional critera and associated tests are needed for music digitization to account for the phase and time aspect of real music to make it more involving. Music involves not just reproducing a frequency or set of frequencies, but the phase relationship with other frequencies that provides the spatial info to localize the instruments and the time envelope of the note that provides the microdynamics and excitement in music (i.e. piuano note versus a similar note played on a violin). Also, the digital filtering needs to be minimized in terms of the added potential distortions on the phase information of the reproduced music. For and audiophile the best situation is if the speakers disappear and one is listening to the performance, not the speakers.
However, I think that tape and vinyl will still be around 50 years from now. The form of digital music will be much different as it seems to be ever evolving in terms of bit rates and resolution. At some point digitized music will be very close to the actual analog form.
While analog data digitization is more convenient for data storage it is very inefficient. At some point the available internet bandwidth per customer will reach some limits for downloads and file exchange. Even telecom networks are now going to multilevels, i.e. more than just 1’s and 0’s.
I think that ultimately, the future will be either analog with much higher fidelity than the current techniques, or some form of multilevel storage above our current very simple binary coding. For example tape using nano particles and technologies for the heads could provide very high resolution for information storage and retrieval, perhaps using non-contact or even optical readout.. So don’t give up on the analog format. There is room for both.
Cheers!
Roman
I’ve never ruled out analog but it’s clear to me that many of the problems with regards to noise, distortion, accuracy, scape flutter are better handled with the existing PCM sampling vs. amplitude methodologies. I want and get what comes in the microphones using an HD PCM rig better than with an analog system.
Pingback: More Fuel for the Fire - Page 2