Dynamic Comparison of LPs vs CDs - Part 4 - page 2

Main Titles from Chariots of Fire

Let's look at some statistics first, comparing:

• the digitally ripped copy of the track (using Exact Audio Copy)
• the analog recording of the SCD-XA777ES playing the track
• the recording of the LP

The following statistics were obtained using the " Analyze Statistics" function of Cool Edit (which produced separate results for Left and Right channels, which were then averaged):

Note that the values for the CD recording are not as "good" as the EAC digital rip. This gives an indication of the difference between CD "real world" performance vs "theoretical" performance.

Now clearly LP loses out in dynamic range, even against the CD recording (by comparing the figures in the Maximum - Minimum RMS Power row). This is to be expected, given the surface noise on LP (which is quite audible on my system).

But before we chalk this as a "win" for CD, let's look at the noise flloor more closely, on both the CD and LP recordings.

This is the noise floor from the CD recording (in the "silent" bit just before the music starts):

The noise floor seems to be hovering around the -88dB level up to 20kHz, and then drops down to below -108dB (the underlying noise floor of the sound card).

By contrast, here is the recording of the LP around the same point:

We can see why statistics often "mislead". LP's noise floor is actually quite low over most of the spectrum, ranging from -84dB around 1kHz to -96dB for frequencies above 10kHz . In other words, the LP recording has a lower noise floor than the CD recording for the majority of the spectrum (frequencies above 2kHz ).

LP's surface noise, which is responsible for the poor dynamic range, is mainly concentrated below 500Hz where the noise level is around -50dB.

And this is for a mass-produced commercial LP, purchased second hand from a thrift store for around $1!

The noise floor is even lower for an "audiophile" pressing on good quality vinyl. This is the noise floor from a Mobile Fidelity Sound Lab "Original Master Recording" of Three Works For Jazz Soloists & Symphony Orchestra ( Don Sebesky ) MFSL 2-503 (200J-3):

As you can see, the noise floor is convincingly below -90dB all the way down to 400Hz. So it would appear LPs do have a reasonable dynamic range for the majority of the audible frequency range.

Many vinylphiles have long claimed that they can hear "below" the noise floor of their LPs. My observations would seem to partially support this claim: surface noise is fairly "structured" (it has a distinct "sound" as opposed to random noise) allowing our brain/ears to "filter" it away and listen to the "music" all the way down to the "real" underlying noise floor which is comparable to CD.

A far more interesting statistic to look at is Maximum - Average RMS Power . As I have mentioned in previous articles, this is a good indication of "relative dynamics." Note that relative dynamics is a different concept from dynamic range. Relative dynamics is the difference in dB between any two points in a waveform. Dynamic range is the difference between the loudest signal and the noise floor.

Here, LP actually "wins" over CD. LP's difference between maximum to average is around 11.56dB, compared against the CD recording at 11.11dB and even the digital rip at 11.35dB. In other words, LP "real world" relative dynamics is better than CD's "theoretical" performance.

At this point you might say: But isn't the higher noise levels on LP perhaps contributing to the values, and therefore the results are not statistically significant?

To investigate further, let's look at the actual waveforms.

This is the waveform for the digital rip (I'll explain the significance of the highlighted selection around 30 seconds into the track later):

The CD recording looks very similar:

And so does the LP recording:

Now, let's zoom in into the highlighted section. This is the CD recording:

Note the relatively quiet section at the beginning is peaking around -24dB. The louder section that follows is around -18-12dB.

By comparing, this is the LP recording at the same section:

Notice that although the quiet section at the beginning is also around -24dB, the louder section that follows has much higher peaks, ranging from -18-9dB! This is despite both recordings having the same average RMS power (in other words, same perceived average loudness). In particular, the loud bit around 30 seconds into the track is peaking at up to 3-4dB higher on LP vs CD!

This finding supports my own subjective impressions comparing the CD against the LP. I much prefer listening to the LP over the CD on my system. The CD sounds dull, congested, muddy, and lacking in dynamics. If I push up the volume, the sound becomes noticeably harsh and artificial. The LP on the other hand sounds more "dynamic" and "exciting."

Now let's compare the spectral views of the recordings. First of all the CD recording:

As you can see, there is no spectral content above 20kHz , due to the Nyquist cutoff at 22.05kHz. The few spikes here and there are probably distortion frrom my setup.

In contrast, this is the LP recording:

The LP would appear to have better frequency response, with spectral components all the way to 48kHz . However, I would caution against this interpretation of the graph (just yet). I suspect a lot of this spectral information is harmonic distortion and there is not much useful frequency content significantly above 20kHz in the original master tape. We'll investigate LP's usable frequency response in more detail later.