A Solution to Wow/Flutter?

Bruce B

WBF Founding Member, Pro Audio Production Member
Apr 26, 2010
6,562
3
38
Seattle, WA
www.pugetsoundstudios.com
#1
The Plangent Processes....

In simple terms, it takes/measures the bias freq from the analog tape and corrects the wow/flutter IM. By talking to Jamie, this can be used for mag film and even LP's by tracking servo noise. Is this just a fancy DSP that makes it sound different, but not necessarily better, or does the system actually work? I will soon know and give a full review in the next few months. For now... snake oil... or miracle?

From the website:

It is a little-known fact that much of the coloration, much of the "sound" of a particular analog tape recorder results from its mechanical design, often more so than its electronic performance. What’s more, no two tape machines, even of the same make and model, are exactly alike either. Each machine will typically have one or two particularly strong mechanical components that will impart a signature sound, but the problem is that they are intermingling in a semi-random way; the rotation rate of a spindle might be 60 cycles, while the rotation rate of another component in the transport might be 47.1 cycles. None of these are harmonically related to each other, so there is a constant cascading of a variety of different culprits that vary every time you press play.

Similarly, the familiar honky and constricted sound of 50's movies is sourced not within the electronics, but in the fast 96 Hz sprocket cogging, a flutter component which is indistinguishable from classic intermodulation distortion (IM).

To eliminate these mechanical defects would be to regain the neutrality and transparency that existed at the time of the original performance and to minimize the unwanted colorations that distract us from our enjoyment of music. For years, the designers of such fine machines as the 3M series of analog recorders were aware of the theoretical possibility for improved sound based on removing scrape flutter and the "sidebands" caused by high-speed modulation of the recorded material, but the technology of the time made it unachievable.

Now we live in a different era. The best efforts by the original machines' hardworking engineers to minimize these issues via servos and sheer mechanical precision pale by comparison to the resolving power of present-day digital signal processing (DSP). For all the controversy surrounding digital recording, it is a given that digital yields a rock-solid timebase compared to even the best mechanical analog systems, with error measured in parts per million. Even the most stable analog recorders are several orders of magnitude worse than digital, at best.

Plangent Processes is a revolutionary system which uses the power of DSP to forensically identify and remove the ever-present wows and flutters which diminish the fidelity of tape and mag film. The complete system includes a sophisticated set of tape heads and electronics mounted on a highly modified super-stable ATR, followed by analog-to-digital conversion and post-processing through our proprietary pitch-correction algorithm, developed with researchers at Cambridge University in England.

The original intent was to develop a system that could unravel gross defects within stretched or poorly stored tapes, and to repair recordings where the original performance of the machine was obviously flawed. But the actual finished product surprised us; not only did it perfectly restore the pitch stability of obviously wowed tapes, it consistently improved the overall sonic quality of even beautifully recorded material.

As we continued our experimentation, to our amazement, we found that even impeccable recordings made on fine vintage equipment exhibit surprisingly high levels of FM modulation distortion due to the presence of flutter at frequencies ranging from several Hz out to beyond 4 kHz. These are the artifacts – as opposed to the pleasant harmonic distortions which give analog tape its “warmth” – that undermine the audio in subtle ways, masking detail, while adding harshness, cloudiness and “grain.” The Plangent Process cleans up the subtle time-domain smearing within these otherwise excellent recordings (while leaving the harmonic distortions intact), revealing detail, smoothness, depth, and stability previously unheard. In many cases it is even possible to hear an improvement in rhythmic feel and “groove” since the micro-timing accuracy of the processed audio far exceeds the original playback. A radical claim, perhaps, but the results speak for themselves.

Another unfortunate fact is that the classic wow and flutter specification standardized long ago by NAB and AES employed a "weighted" characteristic, which yielded results that numerically looked good but masked the true extent of the flaws present and audible. Many noted designers made efforts to revise the spec, and to work to a higher degree of precision. Indeed, the design brief of one of the best-sounding machines ever made (3M) focused particularly on the problems of HF flutter and scrape flutter, and that was in large part why these classic machines sounded as good as they did. But many of the other machines of the era also seemed to measure well; meeting the specifications in use during the heyday of analog recording was assumed to guarantee results below the threshold of audibility, which we now know was clearly not the case. Our process reveals in dramatic fashion how this overlooked phenonemon diminished the recordings of the past, and it allows us to now hear the recorded material without those flaws.

This has led toward a new understanding of the true nature of "generation loss." Our research indicates that it is not so much a result of the magnetic losses and electronic flaws as has long been commonly believed, but rather is caused by speed instabilities, previously unnoticed, garbling the sonics of recordings that measured "flat" but never sounded right. More likely, the “generational loss” we were hearing was actually FM distortion caused by accumulated flutter.

Happily, all these issues can be addressed today. With the extreme wideband sampling techniques and the subsequent digital forensics we employ, the Plangent system can clock the original recording deductively, even 50 years on, and can then reclock the audio with a fraction of the error found in the original recording, often beyond the resolution of 1 part per sample. In so doing, we undo the errors of the past, restoring much of the fresh and clear sound the mixer and producer originally heard "live" in the control room before the interference of the tape machine was introduced.

Plangent Processes' intent is strict; we take great pains not to modify the original recording in any way, other than to correct mechanically-induced timing and pitch flaws. Our goal is to restore the clarity and neutrality of the signal so that it comes closer to the original console output. To that end, our software algorithm has been carefully designed to be extremely transparent. Based upon a novel proprietary wavelet-based irregularly-shaped sampling theorem developed by Dr. Patrick Wolfe at Cambridge University, England, it utilizes a proprietary resampling algorithm never before used in an audio context. As a result, there is none of the typical harshness of conventional sample-rate-converters and pitch shifters, even though the resampling rate is changing broadly and constantly. What goes in is what comes out, with only its timing subtly altered, nothing more.

Unlike other familiar restoration tools which are additive (such as equalization or companding), or which attempt to subtract noise (but also inevitably subtract ambience and potentially add other artifacts), Plangent Processes does nothing but re-conform the existing waveform to a much more highly accurate timebase, shifting it back to the shape it took on before the tape mechanism distorted its timing. There is absolutely no level-shifting, no frequency response shaping, and no compression. The sonic results include less interstitial haze, increased ambience, less abrasive timbre to the background hiss (with an apparent reduction in the noise floor), sharper transients, and smoother siblilants. The stability in the time domain also results in pinpoint imaging, reduction in "wandering" center, and increased definition of the rhythmic underpinnings (the “groove”) of the music.
As a bonus, our processed result also presents much less of a moving target for single-ended NoNoise™ type systems to work with, since the analytical ability of these systems is affected by the phase accuracy and stability of the incoming audio. And given the improved subjective quality of the tape hiss (which is often made less obtrusive and smoother by the Plangent Process), it is probable that a lighter touch can be employed with all subsequent processing tools.



AES White Paper - Correction of Wow and Flutter Effects in Analog Tape Transfers
 
Last edited:

JackD201

[WBF Founding Member]
Apr 21, 2010
10,992
8
38
Manila, Philippines
#2
If this works it will be a very useful tool for someone in your exact line of work Bruce.
 

RogerD

Active Member
May 23, 2010
3,142
0
36
BiggestLittleCity
#3
My,my,interesting stuff! I hope there on to something big. Thanks for posting Bruce.
 
May 30, 2010
13,964
42
48
Portugal
#4
Bruce,
Are you also getting a Custom Plangent/Flux Magnetics stereo head assembly for your Studer A80?
 

rbbert

Active Member
Dec 12, 2010
3,423
0
36
Reno, NV
#5
This has been used on several Grateful Dead releases, including the Winterland 1973 box and the recent Europe '72 giant box.
 

Bruce B

WBF Founding Member, Pro Audio Production Member
Apr 26, 2010
6,562
3
38
Seattle, WA
www.pugetsoundstudios.com
#6
Bruce,
Are you also getting a Custom Plangent/Flux Magnetics stereo head assembly for your Studer A80?
Yes, I already have Flux Magnetics ER heads on all of my machines, but it will be interesting to see what this Plangent headbloc does in addition to our own Flux heads.
 
Apr 3, 2010
16,022
0
0
Seattle, WA
#7
Great read indeed! A few months ago I went researching Wow and Flutter effect to understand their perceptual effects relative to jitter in digital audio. One great paper I found had a fascinating point related to this device in that it also said the standardized weighting for Wow and Flutter was wrong since it had not kept up with time in the way the new (at that time) tape mechanisms generated a lot high frequency variations. He went through the details of what caused that.

This is important since low frequency variations in tape speed creates sidebands that are masked by the signal and hence, even high levels of it are very much tolerated. It only becomes an issue when the frequency becomes so low that we detect the variation in speed (a different problem than jitter).

Does this device put out a digital signal then?
 

Bruce B

WBF Founding Member, Pro Audio Production Member
Apr 26, 2010
6,562
3
38
Seattle, WA
www.pugetsoundstudios.com
#8
Does this device put out a digital signal then?
Yes, you use your own A-D converter at an optimized sample rate in relation to your bias freq. If your bias freq is 120k, then you will need to record to at least 352.8kHz because 192k is too low (Nyquist).
DSD64fs wouldn't work because of the UHF noise, but DSD128fs might. We'll have to see.
 

Bruce B

WBF Founding Member, Pro Audio Production Member
Apr 26, 2010
6,562
3
38
Seattle, WA
www.pugetsoundstudios.com
#9
To bring this thread back to life, I want to expand on what I said over on this post.
I was doing some Stevie Wonder tapes and noticed some Wow/Flutter and was wondering if the Plangent system could really correct this? I have posted 2 samples on my server. Your job, if you wish to choose it, is to download the 2 files, both at 24/192 and determine which is the original and which is the corrected version.

ftp://pugetsoundstudios.com/

User: sample
PW: whatsbestforum
 

bblue

New Member
Apr 26, 2011
360
0
0
San Diego, CA
#11
To bring this thread back to life, I want to expand on what I said over on this post.
I was doing some Stevie Wonder tapes and noticed some Wow/Flutter and was wondering if the Plangent system could really correct this? I have posted 2 samples on my server. Your job, if you wish to choose it, is to download the 2 files, both at 24/192 and determine which is the original and which is the corrected version.

ftp://pugetsoundstudios.com/

User: sample
PW: whatsbestforum
Sample A sounds distinctly different than B, and B seems like what I'm more used to hearing. In fact, it is almost identical in character to an 88k SACD rip I have of the same song. If you play B after A, B starts out sounding a little flat.

Sample A overall, is clearer, more open with more inner detail. It's easier to hear individual voices in the background vocals, individual cymbals and instrument placement. Sample B by comparison (and my 88k version) are slightly veiled and 'scrappy' sounding. In this context 'scrappy' means a little more distorted especially around sibilance-over-other-sounds.

Sample A also demonstrates how different the audio on Stevie's lead vocal track is compared to other tracks.

It's hard to imagine that Wow&Flutter were the only thing removed. It sounds like a generation closer to the master.

I'm using JPlay mini in extreme mode, 2 samples buffer and the HiFace KS driver on a Mtech HiFace USB->S/PDIF converter.

Boy, I sure hope I'm not wrong on this. :eek:

--Bill
 

rbbert

Active Member
Dec 12, 2010
3,423
0
36
Reno, NV
#12
I'm travelling with only my iPad, so I can't compare, but the improvements you mention are similar to what Plangent Processes claims will result from using their "process".
 

Andre Marc

Member Sponsor
Mar 14, 2012
3,973
0
0
San Diego
www.avrev.com
#13
To bring this thread back to life, I want to expand on what I said over on this post.
I was doing some Stevie Wonder tapes and noticed some Wow/Flutter and was wondering if the Plangent system could really correct this? I have posted 2 samples on my server. Your job, if you wish to choose it, is to download the 2 files, both at 24/192 and determine which is the original and which is the corrected version.

ftp://pugetsoundstudios.com/

User: sample
PW: whatsbestforum
Hi Bruce:

I have just downloaded the samples and will give my choices later. Thanks!
 

Andre Marc

Member Sponsor
Mar 14, 2012
3,973
0
0
San Diego
www.avrev.com
#14
To bring this thread back to life, I want to expand on what I said over on this post.
I was doing some Stevie Wonder tapes and noticed some Wow/Flutter and was wondering if the Plangent system could really correct this? I have posted 2 samples on my server. Your job, if you wish to choose it, is to download the 2 files, both at 24/192 and determine which is the original and which is the corrected version.

ftp://pugetsoundstudios.com/

User: sample
PW: whatsbestforum
Ok, after ONE listen straight through each sample, I am guessing A is uncorrected, B is corrected.

I will submit either the same answer or reverse my decision after several more listens.
 

Andre Marc

Member Sponsor
Mar 14, 2012
3,973
0
0
San Diego
www.avrev.com
#16
To bring this thread back to life, I want to expand on what I said over on this post.
I was doing some Stevie Wonder tapes and noticed some Wow/Flutter and was wondering if the Plangent system could really correct this? I have posted 2 samples on my server. Your job, if you wish to choose it, is to download the 2 files, both at 24/192 and determine which is the original and which is the corrected version.

ftp://pugetsoundstudios.com/

User: sample
PW: whatsbestforum
Ok. Sticking with my original pick. Sample B corrected. Crossing fingers!
 

bblue

New Member
Apr 26, 2011
360
0
0
San Diego, CA
#19
Here are the files.

Sample A = corrected
Sample B = original
Bruce,

The differences in these files was quite clear. But I'm having a hard time relating the change of audio entirely to the removal of wow and flutter.

I take it you have equipment for Plangent in house? Does this process produce a similar result when applied to different source? Or does it vary wildly depending on which tape machine originated the master?

Have you tried processing an all-digital recording through the process? Theoretically at least it shouldn't do much since it wasn't recorded on a tape machine originally? Does it?

--Bill
 
Aug 1, 2011
3
0
0
#20
Interesting comparo!

Just for some background: Listening through a pair of K702s, hooked up to an Apogee Duet 2 and coming from a Retina Macbook and Decibel. Due to Meniere's disease, I'm about 4db down in HF in one ear and very slightly down across the board in that ear as well. That said, I used to beta test hardware for a manufacturer and have always been blessed with a receptive and, I think, discerning aural palate, as it were. Given all that, here's what I hear:

On the processed version when Lani Groves (?) Gloria Barley (?) sings "You are the apple of my eye," I can hear here pull her lips apart at "app-le," almost as if she's taking in a small breath. (good technique to avoid popping the P?)

That's blunted on the unprocessed file. The processed file also seems to un-smear the closed high-hat, making it easier to hear the (very) damped metallic ring of it. The effect is even more apparent on the ride. At about 1:00, the drummer works his way from the middle of it toward the bell. This is much more apparent on the processed file.

The final effect I'm less sure of, possibly due to the Meniere's: The unprocessed file has more of a swishy, phase-y throbbing between L and R in the bass range--a pulsing, alternating kind of pressure. The processing diminishes it. Wonder if that's the sound of the wow going away.

Bruce, on a separate note, as a sanity check, could you give me a quick gear diagnosis?

My Duet 2/K702 combo seems a little sibilant, a little brash on top and just a bit recessed in the vocal range. The phones are well-broken-in (about 1000 hours), so that's not it. I suspect the Duet 2. It's USB2. They don't say whether it's async (Cagey bastards, they are, at Apogee). Seeing as I'm 48 and down a bit in the HF in one ear, it would have to be fairly edgy for me to notice. That is, unless a top HF loss manifests itself somehow as folded-down distortion in the sibilant range. Seems doubtful.

For reference, listening, say, to Nat Cole's Stardust, from a Steve Hoffman SACD, converted from DSD to 176.4 in AudioGate, using Gentle rolloff and Aqua dither. Unless the ribbon mike or U47 he used was wiry (it ain't), what, do you suppose is going on?

BTW, no clipping. Duet is either on Mac power or the outboard brick, so that's not an issue. Again, phones are broken in. Whaddya think? Any experience with the Apogee?

Sorry to ramble.

Marshall
 

Members online

About us

  • Founded in 2010 What's Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals we enjoy learning about new things and meeting new people and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing