Recording of April 2018: Fairytales: Original Master Edition (MQA)

Radka Toneff & Steve Dobrogosz: Fairytales: Original Master Edition (MQA)
Odin LP03 (original LP, 1982); Odin CD9561 (24-bit/48kHz MQA-encoded FLAC file; Tidal Masters stream; hybrid MQA-CD; original sample rate 192kHz; 2017). Arild Andersen, prod. (1982, 2017); Andreas Risanger Meland, exec. prod. (2017); Tore Skille, Tom Sætre, original engs.; Svein Vatshaug, Rune Sund Nordmark, recorder restoration; Thomas Baårdsen, Geir Iversen, digital transfer of original tapes; Morten Lindberg, Peter Craven, Bob Stuart, digital restoration; Erik Gard Amundsen, technical advisor. DAA (original LP); DDD (MQA). Except: "My Funny Valentine," ADA (LP), ADD (MQA). TT: 40:11
Performance ******
Sonics ******

"See her how she flies . . ." When I first heard that lyric, from Jim Webb's "The Moon Is a Harsh Mistress," sung by a hauntingly fragile woman's voice and supported by a sparse yet lyrical piano accompaniment, at an audio show in 1983, I got chills. Who was this empathetic singer? Back in my cabaret-musician days, more than four decades ago, I backed so many singers with beautiful-sounding pipes but who didn't seem to comprehend the meaning of the words—yet this unknown woman directly communicated the song's emotion.

I grabbed the LP jacket and learned that the singer was Radka Toneff, a Norwegian of Bulgarian ancestry, and that the piano was played by Steve Dobrogosz, an American who had relocated to Sweden. The LP had been released in 1982 on Odin, a Norwegian label, and ended up selling more than 100,000 copies. By the time I'd bought my copy, Toneff, at the age of 30, had taken her own life.

Other than "My Funny Valentine," recorded in analog in 1979, the tracks on Fairytales were recorded in February 1982 on a two-track Telefunken/Mitsubishi MX-80, a 16-bit digital open-reel recorder running at a sample rate of 50.35kHz. The recordings were not mixed or processed in any way—the original LP was cut from the recorder's analog outputs. Subsequent CD releases were made from a copy on a Sony PCM-F1, again made from the Telefunken's analog outputs, meaning that these were corrupted both by the F1's single A/D converter, shared between channels, and the fact that the PCM-F1 operated at 44.056kHz rather than 44.1kHz. The master tape was subsequently lost, and the original Telefunken machine was converted to run at 48kHz.

In 2010, the original master tape was discovered in the archives of Norway's Ringve Music Museum—which also had the original MX-80 recorder. In 2015 there was an attempt to recover a version from the repaired machine, converting its analog output to 192kHz PCM. "To persuade the machine to lock, it was necessary to manually apply resistance to the rollers (slowing the tape speed below 15ips) while the digital electronics locked at a rate near to 48kHz," I was told by MQA Ltd.'s Bob Stuart. As reported by Norwegian journalist Jan Omdahl in December 2015, the 2015 restoration was flawed. In particular, while the recording's original pitch had been A=442Hz, the restoration was tuned to A=446Hz, and there was noticeable wow.

What to do? A team set to work: Bob Stuart of MQA, Peter Craven of Algol and MQA, Thomas Baårdsen of the Norwegian National Library, and Morten Lindberg of Norwegian record label 2L. A proprietary chirp signal was used to extract a high-precision estimate of the amplitude and phase frequency responses of the MX-80's ADC and DAC using the pass-through monitoring path (which cascades the ADC and DAC). Using a gentle method to establish lock at 48kHz, they could capture the library machine's analog output playing the 50.35kHz master at 192kHz for 8 of the 10 tracks. As the transfer was at the wrong pitch—about 4.7% too low—they did a lot of research to make sure that the original pitch was indeed A=442Hz before correction.

While the MX-80's DAC didn't show major bit errors, quantization distortion was apparent in the original master due to misadjustment of the MSB trim for the L/R A/D converters. Craven created tools that allowed the resultant bit misalignments to be corrected, reportedly reducing inharmonic converter distortion by around 40dB.

To maintain the highest transparency, the entire set of corrections was carried out using just two 24-bit operations. As described by Stuart, this comprised the following steps:

1) Stuart: Starting with slow stems, correct the ADC bit errors.

2) Lindberg: Top and Tail slow files; fix gross errors (original splices and big clicks); determine level adjust to make for each song; insert metadata.

3) Stuart: Process in one operation to: reverse out MX-80 ADC/DAC in the time domain; adjust track levels; correct the sample rate for pitch; LF phase correction; stabilize noise floor.

4) Andersen and Dobrogosz: Review results; decisions concerning minor track-level changes; repeat step 3.

5) Lindberg: Perform magic with very light "airbrush touch-ups" of very fine dynamics; cuts to length and assemble deliverables.

6) Stuart: Encode for download and for CD.

This workflow was used for tracks 2–4 and 6–10. For track 5, "My Funny Valentine," a generation-corrected transfer from the original 1979 analog tape was used to avoid the generation loss of the MX-80. For track 1, "The Moon Is a Harsh Mistress," the analog transfer from the MX-80 had unacceptable wow, either because the tape had been stretched or as a side effect of the manipulation and error interpolation during the transfer. However, according to Stuart, the "direct digital transfer was very high quality from the second musical line onward. Accordingly, track 1 was assembled from pitch- and level-corrected digital and analog stems using 'highly transparent' custom processing."

Although the restored recording has been available in Norway since late 2017 for download and for streaming via Tidal Masters, as well as an MQA CD prepared from a 24-bit/176.4kHz master, Fairytales would not be released in the US until March 2018. For this review, therefore, I played pre-release MQA in 24/48 FLAC files that I'd been sent with NAD's Masters Series M50.2 server feeding S/PDIF data to a Mytek HiFi Brooklyn DAC. (The NAD performed the first MQA "encapsulation unfold" to 96kHz data, the Mytek the further "type-L" unfold to 192kHz.) For reference, I'd made 24/192 needle drops from my well-worn 1982 LP. I cued up track 1 . . .

"See her how she flies . . ." When I heard that familiar lyric, I was transported back 35 years to the first time I heard it. The occasional trembling in Toneff's vulnerable voice, the subtle shadings of pitch she uses to point a phrase, the careful way she uses vibrato on held notes for emphasis—they were all there. What was old is new again. The key change after the brief piano solo tore me apart as it had done the very first time. The vast emptiness opened up by Dobrogosz's falling-fifths intro to "Nature Boy," which is answered by a rising fourth at the end of the song, sounded more vast than it did in the needle drop, even after I optimized the Brooklyn's reconstruction filter for straight PCM. The space of the recording venue, Bergen's Grieg Hall, was more coherently presented. The transients of the piano sound throughout the album came across as more believable, the individual notes within sustained chords more identifiable. Oh my!

"The moon can be so cold," wrote Jim Webb, "though she looks as warm as gold." This restoration of one of the 1980s' greatest albums has turned digital cold into musical gold.—John Atkinson

COMMENTS
dalethorn's picture

I have the older version, and eagerly await the MQA version here in the U.S., but sources I read say the release is in mid-April, not March.

tonykaz's picture

I can recall Karen Sumner giving me this Album, back in the mid 1980s ( I was one of her Electrocompaniet Dealers ) and I can recall everyone at the Winter CES having this music 'wafting thru the Nevada Ether'.

I'm playing "The Moon is a Harsh Mistress" as I write this note, I should probably Order this new Release and have Karen Autograph it while sending me some Transparent Interconnects ( for the price of a Car ).

Thanks for the Memory,

Tony in Michigan

ps. Doug MacLeod's "Unmarked Road" and everything of his I've so far listened to is Suberb..!

Priaptor's picture

Let's see how "MQA Masters" sounds.

pat mcginty's picture

We've been using The Moon is a Harsh Mistress to voice our designs since the early 90s. To us, it's indispensable for judging emotional connection. Can't wait to hear the MQA.

leedom's picture

Now we have no way to avoid MQA? I think I'll sit this one out.

cgh's picture

... that A=442 is about two-and-a-quarter cents away from the JND of the "right" tuning. Between the august ears of the average SP listener and the Schumann resonance everything should be alright!!!

All joking aside, and without touching the MQA kryptonite, the whole process seems pretty cool

ok's picture

..does this mean that the same care applies to any given MQA (re)issue?

dalethorn's picture

If anyone needs a shortlist for remastering, I'm ready with mine. I'll even pay the caterer.

TG04's picture

Hardly, I have listened to a good number of MQA encoded albums on Tidal with software decoding and it's a mixed bunch. One of the worst I encountered was The Beach Boys "Pet Sounds". In all fairness, this is a very poor sounding album regardless of media, but the MQA version on Tidal must be the low-water mark - IMHO, of course

dalethorn's picture

I'd clarify my comment with the same opinion I have about the Radka Toneff - if the original master is restored to a good sound *before* applying MQA, then the MQA version should sound good. One further point is that while Apple iTunes isn't the first place most audiophiles go for downloads, if Apple were to offer more "MQA" albums, they likely would have much higher resolution (~500 kbps) than other iTunes albums. But there's no guarantee obviously.

TG04's picture

Precisely, I couldn't agree more. It's a "garbage in - garbage out" type of situation. It strikes me odd (or maybe not so odd) that the MQA releases we see are the same tired old Hi-Rez issues we've been sold as downloads by HDtracks and others for some years now. It occurs to me that "they" (i.e. the MQA guy behind the curtain aided and abetted by the audiophile community) just keep cranking them out by the numbers. No individual involved, just that bulk converter pumping out MQA "certified" reissues. What a difference when compared to JVC's XRCD program, which would reissue only worthy albums and painstakingly restore and issue those using the best technology (at the time) - supervised by knowledgeable, caring individuals. No wonder they're no longer around...

dalethorn's picture

For those of us who are still buying hirez downloads, we have to hope that the preview/sample button gives us the same sound as the tracks we're about to purchase. If that ever becomes unreliable, we'd have to be sure that our downloads are the same mastering as what the trusted reviewers describe.

TG04's picture

well, the review here is of a file delivered concurrent with a NAD product presumably to be reviewed -- however, the file is of unknown progeny and hitherto unheard of tech specs (24/48 FLAC)...

anyway, have you ever tried to register a complaint about a download? and what would your argument be? at least with physical media, there is sporting chance you may be able to sell it to some other (greater) fool, or at least donate it to a nearby thrift shop.

with a download, well it's just money down the drain, isn't it?

dalethorn's picture

Each of those arguments is valid, but I'll refer to my experience with CDs going back to 1990, and digital files going back to circa 1999. I have a large bag containing about 150 CDs that I ripped lossless with bit-perfect verify, and those tracks are backed up onto about 6 drives (hard drives and large flash drives). Each drive contains all of the files. That bag of CDs is marked to sell or give away whenever I get around to it. I have another 200 or so CDs that I've also ripped likewise and backed up the same way, and those CDs are keepers that I won't let go of, for several reasons (liners for one, or because they might be premium trading material). Adding to all of that are several hundred CDs that I've bought and already sold or given away.

I have about 7000 permanent digital tracks, from CDs and downloads, where the filenames and the metadata track titles are set by me as "Artist - Title". An example would be "Mahler (Bernstein-New York Philharmonic-1968) - Symphony 3 Part 1 Adagio etc. etc." That way I have only one large playlist where I can get to any track in 5 seconds or less, and it's easy to cull items from the playlist that I'll no longer listen to.

Besides the extreme convenience in listening that I've gotten used to, that system saves me tons of time in testing.

tronds's picture

The new reissue available in Norway is actually a SACD, with both the DSD layer and a MQA-encoded CD layer. I would find it strange if the international release would differ from this disc layout.

dalethorn's picture

The SACD should be the same. I'm waiting on the MQA download.

leedom's picture

Hhmmm, now a DSD Digital Download would definitely be interesting ...

TG04's picture

correct, both layers on the physical media received from www.grappa.no are from the MQA encoded master (as I understand the written material in the CD package). Personally, I would have preferred the SACD in straight DSD as a comparison, but that's not what I got.

On second thought, another way to view it is that you get a FHD recording incl the benefit of MQA without any additional hw or sw (assuming that you consider DSD and 24/176 to be of equivalent resolution)

dalethorn's picture

Apple is selling this title ("Fairytales - Original Master Edition (MQA)") on iTunes in the U.S. now, exactly as shown in the album cover image above. They also sell the older edition. But the tracks are not MQA'd. Here is what I found so far.

1) These iTunes files (AAC, file extension '.M4A') have a much higher playback bitrate than any of the other iTunes files I have in my collection - several hundred of them. All previous iTunes files I know of are 256 kbps variable bit rate, and these "MQA" tracks go as high as 461 kbps variable bit rate.

2) I converted these .m4a tracks to FLAC, as I did with the older edition iTunes tracks, and the MQA-edition FLACs total size is 165.5 mb, compared to 123 mb for the older FLACs.

3) I played these .m4a tracks as well as the FLAC conversions I made on my Mac's music player using the Meridian Explorer-2 DAC, making sure to try all of the bitrates provided by the Mac's MIDI settings, and never could get a green light or blue light on the DAC as I get with the Steve Reich Pulse album.

4) These iTunes tracks are 16/44, as are the Reich/Pulse MQA tracks I ripped from the CD I purchased. Those Reich tracks do produce the correct light on my DAC, but these iTunes tracks do not.

dalethorn's picture

Lest I forget - the sound: The new version on iTunes is obviously better than the older version - the actual volume seems slightly higher (perhaps 1.5 to 2 db), although the overall dynamics seem very much the same, and this album has some serious dynamics. You can really hear the difference in the loud higher-register piano notes. The remastering of this involved a lot more than just MQA coding, so that probably accounts for most of the notably better sound, and raises the ongoing question: How much improvement do we hear due to MQA, and how much due to the other factors?

TG04's picture

I ripped the CD layer of the new SACD Hybrid to WAV (file size 418 mb) and also resampled to FLAC at 16/88.2 (225 mb). Don't know if the MQA encoding was lost in the process. Regardless, the playback bitrate is an unremarkable 742 kbps v. the expected 1000+ but it sounds very good indeed through JRiver and a Chord Mojo DAC (no MQA support), much like the physical media played back in DSD on my Oppo 95 (again, no MQA support).

Regardless, this one's a keeper - downloads come and go but I will never sell this disc

dalethorn's picture

I have two opinions on the 742 kbps bitrate - one, that the music is low complexity, and I've seen that especially on solo piano works. The second thing is the MQA mastering itself, which is not truly high-res, but "effectively" they say through the magic of their codec.

TG04's picture

What file format did you rip the Reich CD to?

Have you tried the same with the Toneff Hybrid SACD?

While at least some Windows compatible formats (such as FLAC and WAV) will carry MQA information, I don't believe Apple formats can manage that. So once in M4A (and maybe AIF also?), whether downloaded or ripped, the MQA aspect will be gone and irretrievably lost.

I am not trying to knock Apple product here, this is simply how I understand the current state of affairs.

dalethorn's picture

I haven't tried ripping to an Apple format for certain reasons**, nor did I have an SACD of the Toneff album. I ripped the regular CDs of each to WAV and then compared those to the WAV or FLAC downloads I got from the same site, and they were the same, and they lit the proper lights on my MQA DAC. As I remember, I got the non-MQA Reich high-res download, but with the Toneff, my only non-MQA download was the iTunes version labeled as "MQA", which despite not being MQA, had a bitrate approaching 500 kbps, a first I think for iTunes. And the Toneff iTunes tracks converted to WAV sounded about the same as the CD rips, i.e. with enough liveness and extension that they were obviously in a different league from the typical iTunes 256 kbps tracks.

**After my lengthy and unsuccessful foray into trying to convert 256k DSD/DFF/etc. tracks to 96k or 192k WAV or FLAC using JRMC, I've given up for the time being on anything other than WAV and FLAC. Although ultimately JRMC was unable to make 96 or 192k files playable on Foobar2000 on my PC, with help from 3 expert sources, all 96/192 downloads I've purchased from several high-res sites play perfectly on Foobar2000.

From everything I've seen and tested, the MQA signature in the music file that lights the proper light on an MQA DAC is no guarantee of anything, other than the file has probably been processed through the MQA codec. The Reich downloads I purchased that were MQA and non-MQA sound almost identical, with arguably a little more "air" above 12 khz or so on the MQA side. So far as I know, while a few people have speculated that what they've heard with some MQA releases is just a minute increase in treble, I don't know of anyone who's doing extensive testing of MQA releases against non-MQA releases to see what the differences are. Those would have to be from the same master of course, which would likely have to be recent releases.

TG04's picture

I have also had some difficulty getting JRMC 23 to play perfectly fine files. You may want to check out the JRMC forum about "taming Windows Defender" -- turns out that adding certain processes, folders, and files as Exception in the Defender setup did the job for me.

dalethorn's picture

My Windows computer isn't connected to any network or Internet, nor does it run JRMC. I download files on the Mac and then copy them to the PC to play on Foobar2000. Every track from every download site plays perfectly, but not conversions made by JRMC, whether those conversions were made on the Mac, or supplied by the experts at JRMC, CA, etc.

TG04's picture

Ah, but you see there is no way of disabling Defender on a Windows 10 computer, it inserts itself in every process you run - also offline. It's really weird, and about all can be done is to alter the setup profile.

dalethorn's picture

1) I run Windows XP, not 10.

2) Regardless of the Windows version, when you go under the O/S and remove certain aspects of "File Security" (NTFS etc.), as you would do if you were making OEM installs for a large corporate network, then you disable that stuff and put your own security in.

3) Every high-res download I've purchased from HDTracks, Presto Classical, Linn, 2L, Qobuz, ProStudioMasters, and several other sites I've gotten files from - all of them run perfectly in Foobar2000. I could not, with expert help from gurus from Stereophile, CA, JRMC, others..., make a 96k or 192k conversion from a DSD file that would work on Foobar2000.

4) I have many IS/IT tools as well as certain hacker tools (I wrote for 2 leading hacker mags), which can make any PC run and jump like a gazelle. There's no reason on Earth that JRMC can't do a high-res conversion that runs where ALL other files run, unless JRMC is doing a stealth DRM. Foobar2000 is open source and JRMC is not, and a reasonable assumption is that JRMC will "respect" certain things internally. DSD files may be one of those things - after all, the principle of DSD is basically the same as SACD if I'm correct.

TG04's picture

I ordered the physical media online from the Norwegian publisher (www.grappa.no). They offer w/w shipping at NOK 50 and the landed cost in NYC was less than $20. While the sound of the SACD layer on my Oppo 95 is very impressive, it seems to me somewhat "MQA-ish" in the sense of excessive detail (especially sibilants) and altogether too loud. Don't know if compression and reverb have been applied but I suspect so. Seems to me this is a closed-miked recording and the heartfelt delivery & fine diction is what sets it apart from ordinary elevator-music. In my opinion, unfortunately, the new SACD fails to convey what JA said he heard from the Odin LP (when reviewing a loudspeaker):

"The delicate fragility of the late Radka Toneff's voice in her reading of Jimmy Webb's "The Moon Is a Harsh Mistress," from her Fairytales (24/192 AIFF needle drop from LP, Odin LP03), was fully preserved. I'd made a number of needle drops of this track using...."
Read more at https://www.stereophile.com/content/magico-s5-mkii-loudspeaker-page-2#zu...

X