Pugliese re-transfer February 2017

Osvaldo Pugliese at his piano
Osvaldo Pugliese at his piano

Thoughts about Pugliese’s TangoTunes transfers
by Christian Xell (Don Xello)

In December 2015 we published San Pugliese 1 with all tunes from 1943–1945 and in January 2016 San Pugliese 2 with all tunes from 1946–1947.

About the process of publishing transfers

As a reference tune I chose Adiós Bardi, the tune to the honor of Agustín Bardi, composer of tangos like C.T.V., Chuzas, El baqueano, El buey solo, Gallo ciego, Independiente Club, Lorenzo, Tierrita and a lot more.

I cut all reference tunes until the very beginning and then I used the time frame from 0:23 to 2:23, so you can listen to 2 full minutes from every example.

After cleaning the shellacs the pitch determination takes place. By testing various needle sizes and mostly 2 cartridges we do 2 transfers. One archive copy in stereo without de-emphasis, i.e., we let all frequency areas flat which sounds very bright and 1 copy with a de-emphasized curve applied.

In a few words: To de-emphasize means the process of finding the reverse curve of the emphasis curve which has to be applied due to physical restrictions when cutting the master. With the lows the amplitude can become that great (inversely proportional to the frequency of the tone) that the groove cannot be cut anymore and with the heights the radius of the curvature can become that small that the needle cannot follow anymore. So the requirements at the lows are to make the spiral coarser and at the heights to run the record at a higher speed. Both requirements condition a boost and an attenuation of certain frequency bands – the emphasis. In fact lows are being attenuated and heights are being boosted. During the transfer you need to de-emphasize, i.e., you boost the lows and attenuate the heights. But at which frequency bands it is done – the so-called turnover frequencies – is not standardized with shellacs. Please see also constant velocity vs constant amplitude if you are interested to know more about this.

Let’s listen to the RAW (archive copy) version first.

RAW version of Adiós Bardi, 0:23–2:23

Let us listen next to the first version from December 2015 with a de-emphasis curve applied. Determining the de-emphasis curve is the art, the experience of the audio engineer. As there are no standard de-emphasis curves defined for shellacs, you can do a lot of things wrong and later it is almost impossible to correct this.
You can fine-EQ some parts of the tune but the overall quality, the sensation, the harmony is created by choosing a good de-emphasis curve.

Only in the early fifties a standard has been defined which de-emphasis curve shall be applied but it was defined for vinyls.
This curve was called RIAA which stands for Recording Industry Association of America.

Adiós Bardi with applied de-emphasis curve, December 2015, 0:23–2:23

A side note – 2 examples how shellac transfers with RIAA may sound

Some private collectors do the transfer with a normal amplifier with the PHONO button pressed, i.e., the shellac is being transferred with the RIAA curve which sometimes by luck leads to good results.

Listen to 2 examples of Di Sarli, the wonderful Rosas de otoño and Julian Centeya, in fact A and B side of a recording from September 1942, a day when Di Sarli obviously had a very good day. The RIAA is wrong here (heights are too heavily cut) but the piano is impressive.

Rosas de otoño – private transfer of a collector with a normal amplifier (PHONO button, i.e., RIAA curve applied)

Julian Centeya – RIAA

Before TangoTunes publishes a tune we always go “live” on various occasions, meaning we play at different Milongas with different PA (stands for Public Adress = Sound System). I played the Puglieses in December 2015 at the Maldita Milonga where I regularly play the music. The PA system at the Maldita is heavy on basses which have the Puglieses sounded very good. In fact the Puglieses from Dec 2015 are missing basses and the overall atmosphere is too bright and sharp which is due to the de-emphasis curve chosen. Normally you have three options to determine the de-emphasis curve.

Bass – Mid – Top

The settings are normally in µs which is a time constant and is represented in Hz as well.

  • Typical values for the bass are 3183µs, 2274µs and 1591µs  –
    the corresponding Hz values are 50 Hz, 70 Hz and 100 Hz.
  • Typical values for the Mids are 159µs, …, 636µs  – the corresponding Hz values are 1.000 Hz, …, 250 Hz.
  • Typical values for the Tops are 25µs, …, 100µs  – the corresponding Hz values are 6.366 Hz, …, 1.591 Hz.

What does this mean?

Adiós Bardi in the de-emphasized transfer from Dez 2015 – listen above – had the settings 3183 | 450 | 40 which represents 50 Hz | 354 Hz | 3.978 Hz.

These 3 frequencies are turnover frequencies which means at this frequency somethings happens.

Low frequency turnover at 50 Hz means that until this frequency no boost happens and the speed of the stylus in the groove is constant.

For our Adiós Bardi this setting is fine, lower µs which corresponds higher Hz values would result in less low bass.

So the “mids” are more important it seems. The first transfer had a value of 354 Hz, i.e., the boost goes until this value.
Boost from 50 Hz to 354 Hz with the settings 3183 and 450 µs.

The treble value of 3.978 Hz means that from there a treble cut of 6db per octave takes place (which consequently means that if I want it less sharp I need a higher value, e.g. 100µs where this treble cut would start already at 1.591 Hz. – in fact with many transfers of Pugliese our sound engineer used this extreme value).

Complicated? I know, if you read about this topic the first time, it’s a nightmare and at least in the early fifties the RIAA curve standardized it all, which does not solve all problems but it made the whole process a lot easier.
By the way the RIAA values in µs are 3183 | 318 | 75 and in Hz 50 | 500  | 2.122

The values for the re-transfer of Adiós Bardi were 3183 | 318 | 40, so almost RIAA, whereas the treble cut starts later with our settings.

If you compare it with the first transfer you will realize that is has more body because of the mid value of 318µs – the boost is longer, before it was until 354 Hz, now it goes until 500 Hz.

Listen here:

Adiós Bardi, de-emphasized curve applied with µs 3183/318/40 and in Hz 50/500/2122

Here is the final tune from us re-published in February 2017 and below the CTA version from Akihito Baba which most likely has been done with a pure RIAA curve as it sounded dumber.
We haven’t found the final version with these Pugliese transfers and we have to work on other components to really make it thrill. But at least we corrected a few things from the first release. TangoTunes is a project that will hopefully continue for many years and thanks to you and your (critical) feedback we will still do many re-transfers.

I had long, fruitful discussions with Christian Tobler in this matter. He still thinks that even the re-transfers do not obey the etiquette Golden Ear as they lack of overall harmony – due to wrongly chosen de-emphasis settings – and he still favors the CTA. I do not agree but I accept this opinion and as long as critical and smart people engage with TT we are on the right way. And for that I am grateful.

TangoTunes version, re-published Feb 2017

CTA version

How can we improve?

Maybe the mono approach isn’t the holy path and we shall try at least to simulate the binaural way of listening though the recordings are in mono.

Maybe fine tuning with reverberation brings improvements but not to compensate weak transfers or to meet listeners expectations as it was the case in Argentina with a lot of recordings but to simulate a more natural listening experience. If you are listening to live music you also face reverberations as the sound waves hit against walls, objects whatever.

Adiós Bardi, Feb 2017 re-published version but in Stereo with a special Stereo Ambiance Filter applied


Osvaldo Pugliese at his piano
Osvaldo Pugliese at his piano


  1. Great that you continuosly strive to improve. What does the re-transfer mean for those who already bought the former set? Do we get a new download link?
    All the best, Mikel

    1. HI Clive, yes they have been de-clicked unless you would hear all the clicks as in the objective copy example here in this post. We smoothly de-click with oxford but most of the de-clicking is done manually. Cheers Don Xello

  2. Hello, are the new re-transfers already available? I got no email, and I am eager to listen listen to them!

    By the way I like much more the sample without the stereo effect, at least on my headphones.


  3. Hi,
    Your talk aboout adding reverb makes me scared. The dry, crisp, reverb free transfers you have put out so far is one of the things that makes your work stand out from the big record companies.
    Did you apply reverb to any of the current transfers?
    Those who like reverb can easily get it by using a DSP plug-in with their player, but once applied you cannot remove it.

    Getting the transfers in 48 kHz double mono as opposed to one channel 96 KHz could be of interest if you want to do restoration on them. Strangely, it can sometimes also sound a bit more alive, I have found.

    1. Johan, I expressively wrote how reverberation must not be applied. No reason to be scared and of course none of our transfers have any reverberation added.
      But if you have a transfer where you think that the one or the other instrument does not get the attention it deserves because the sound engineer who arranges the instruments in the room does not make a good job or the whole arrangement does sound boring you might experiment with some delays to compensate and to make it more vivid and lively.
      But with great caution because in almost all cases you do not get the desired effect. In German there is a wonderful word for that: Verschlimmbesserung, i.e., you make something worse while intenting to make it better! (some sources use the English word “disimprovement”)

Leave a comment

Your email address will not be published. Required fields are marked *