Have you ever wondered
how seriously record companies regard
comments by reviewers in magazines and
web-sites? As can be seen below Dunelm
records felt a criticism by Don Satz
was an impeachment of their high standards
of production and took the complaint
very seriously indeed.
Ludwig
van BEETHOVEN (1770-1827) The
Piano Sonatas, Volume 1
Murray McLachlan, piano Recorded at
Whiteley Hall, Chetham’s School of Music,
Manchester, Summer 2003
DUNELM RECORDS DRD0201 [78:33]
In his
review Don Satz commented:
Unfortunately, crackling
noises embedded in the soundstage dampen
enthusiasm for the production. This
is not acceptable for a 21st
century recording, and Dunelm needs
to address and correct the problem.
Otherwise, McLachlan’s Beethoven series
has little chance of competing with
all the alternative versions on the
market.
The results of the investigation by
the record company are shown below.
July 2004
Ref: 0201ds
problem
Dunelm Records: Investigation of review
copy of DRD0201:
McLachlan plays the 32 Beethoven Sonatas:
Volume1 sent to Classical Musicweb
Contents
Introduction
Problem
Addressing the problem
(1): Request for more information
(2): Tests on the returned CD-R
(3): Examination of the tracks on the
music computer
Table:
Results of a "glitch search on
the Review CD-R and the Master CD-R
(4): Examination of the original
manufacture of the Master
(5): Independent examination of the
Review CD-R
(6): A new clone
(7): Previous experience
Appendix 1: Optical discs
Further reading
Introduction
Recordings of the Beethoven Piano
Sonatas were made in The Whiteley Hall,
Chetham’s School of Music, Manchester
during 2003 with the pianist Murray
McLachlan, Head of the Keyboard Department.
The piano used was the Steinway D, prepared
by Peter Lyons, and the microphones
used were four AKG 414 B-ULS one at
90 degrees to the keyboard, a stereo
pair on the diagonal and one at the
tail. The distances from the piano were
250cm and the height from the floor
190cm.
The sound
was recorded at 24bit on DAT via a Mackie
Designs 1604 VLZ Pro 16-channel MIC/Line
mixer [2003] through a TC electronic
Triple C signal processor [2001] to
a Tascam Professional Digital/Analogue
Audio Tape Recorder Model DA-45HR [2001].
Editing
was done using Sound Forge 6 and the
Master CD was made on a Mitsui gold
Ultra low speed CD-R [Serial number
3167 2221 4170*] at a speed of 4 × using
a Plexwriter 48/24/48 CD-R writer.
Clones were made from this master using
a Marantz CD Recorder / CD player CDR500
at 1× speed.
A Mitsui gold Ultra CD-R clone [Serial
number 3167 2221 1844*] was supplied
to Classical Music Web on January 14th,
2004 for consideration for review.
* A batch
of 300 CD-Rs with serial numbers 3167
2221 XXXX was purchased from The CD
Team at Henley-on-Thames in 2003. All
have been used on a variety of recording
projects. No problems have been reported
with the 298 others!
Problem
On June 6th, 2004, a copy
of the review written by Don Satz of
Albuquerque, New Mexico, USA was received.
[http://www.musicweb.uk/net/classrev/2004/Jun04/Beethoven_Sonatas_Dunelm.htm]
The paragraph
causing great concern at Dunelm Records
is that placed third from the end of
the review together with the first sentence
of the final paragraph. These are:
"Unfortunately,
crackling noises embedded in the soundstage
dampen enthusiasm for the production.
This is not acceptable for a 21st
century recording, and Dunelm needs
to address and correct the problem.
Otherwise, McLachlan’s Beethoven series
has little chance of competing with
all the alternative versions on the
market."
Given
the sound problem I noted above, my
recommendation is to wait and see if
Dunelm corrects the crackling soundstage.
[Note:
The
irony of the situation is that Mr. Satz
also reviewed DRD0212: Bernard Roberts:
Celebrity recital and, although the
method of manufacture of the review
CD-R was exactly the same as that with
a problem, no "crackling problems
were reported. The two copies on file
bear serial numbers 3274 2221 0021 and
3274 2221 0013.
http:// www.musicweb-international.com/classrev/2004/Mar04/Roberts_Satz.htm
Addressing
the problem (1): Request for more information
1. Following
a request for further elucidation
of the "crackling noises",
Mr. Satz replied:
"I'll help as best I can.
I used the word "embedded" because the
crackling
didn't sound like an extraneous noise
but like the crackling that tended to
be prevalent in LPs.
First, it is most noticeable on my best
equipment; when using just an
ordinary portable CD player with low-end
headphones, there is just a little
projection.
Second, the crackling is fairly persistent,
essentially throughout the disc,
although its loudness increases as the
program progresses.
…I'll get the disc back to you.
And later,
"I haven’t had any tracking problems
with the disc; the noise is most noticeable
on my Sennheiser headphones, if that
means anything."
2.
Jim Pattison replied:
Obviously, I am extremely worried about
the (crackling) noise you report.
May I ask one further question? Did
you realise that the disc supplied is
a custom-made CD-R and not a commercial
pressing? Some CD players may
refuse to play CD-Rs because of their
hybrid nature. Your third paragraph
gives the clue for, if tracking problems
were being experienced this is exactly
the sound effect it has.
(See New Scientist, 20th May
1999 page 10 "Hi-fi anxiety"; I can
send you a copy if you would like to
see the article.)
Since that article was written, most
manufacturers have improved their laser-tracking
specifications and include a little
plate on the front of their machines
saying CD-R/RW playback. I assume your
machines have this.
Addressing
the problem (2): Tests on the returned
CD-R
1. CD-R Serial number 3167 2221
1844 was received at Glossop on June
16th, 2004.
A visual examination showed nothing
untoward.
2. The CD-R was placed in a Sony Super
Audio CD Player, Model SCD-XB780 [New
on November 16th, 2002] and
played back via a Sony Integrated Stereo
Amplifier Model TA-FB740R {New on March
27th, 2001]. The sound was
heard firstly through headphones – Beyerdynamic
Model DT 250 80 ohm / System Serial
no. 09156 [New on July 30th,
1999].
These headphones
are very "bright" and it was
possible to hear – with great clarity
– the occasional page turn and pedal
application. There was no "crackling"
noise.
3. As for paragraph 2 but listening
via two wall-mounted B & W Speakers
Type DM602 S2.
Listened to at higher volume, nothing
untoward was heard.
4. The CD-R was next listened to via
a consumer-use (costing about £200)
Panasonic CD Stereo System Model SA-PM25
with CD-R / RW playback symbol on the
front [New on September 8th,
2001]. Again, the results were clean
and of a standard expected from this
efficient domestic unit.
Addressing
the problem (3): Examination of the
tracks on the music computer
Introduction:
To try
to locate (I quote) "crackling
noises embedded in the soundstage…"
– which noises should be visible and
audible on each track – each track on
the "problem" CD-R was extracted
from the CD-R using CD Architect. This
procedure was repeated for the Master
CD from which the "problem"
CD-R was cloned.
Sound recording
engineers are aware that unwanted noises
– called glitches – can spoil a sound
recording. Glitches are commonly the
result of analogue audio editing, analogue
to digital transfer, or electronic noise.
One of the hazards of location recording
is that the audio cable(s) which not
only carry the "sound" wires
but also the phantom power for the microphones,
must be really secure. If the XLR plug
"droops" (owing to the softening
of the plastic sheath in a warm hall)
and the phantom power pin separates
from its socket – even by an imperceptible
amount – an electrical "spike"
is caused which gives a massive "glitch"
on the waveform. The shape of a glitch
is an almost vertical front with an
exponential "tailing off".
Because of their shape, they can easily
be recognised by a computer algorithm
which examines the audio file for instances
where the waveform matches the specified
threshold slope and sensitivity criteria.
The cursor then moves precisely to the
location of the glitch to allow it to
be repaired. Since the "Find"
tool only locates one glitch at a time,
all the glitches can be identified sequentially.
If the "Find" tool finds no
glitches it reports, "No events
of the specified type were found"
which has been abbreviated in the Table
to NE.
There are
two possible pitfalls, however! These
are:
(i) Where the music is loud and percussive,
for example, from a piano, "attack"
waveforms can look like glitches to
the algorithm when they are not! Thus
it is necessary to "listen through"
all reported glitches to ascertain if
they are real or spurious.
(ii) There are three methods of repairing
glitches but each of these is only partially
effective, in other words, if the glitch
is a small one, it will often be masked
by the music. No engineer can "miss"
a big glitch!
It follows
that, if the source of the "crackling
noises" is a series of glitches,
they must be large and there must be
very many of them!
In the
following operation, the code for the
Review copy of the CD-R is CD-966703098
and that for the Master CD-R is CD-1796841454.
Both codes were "read" by
the computer from the discs.
The tracks
of the Review copy were examined first.
Then those containing "alleged
glitches" were checked against
the Master to ascertain if they were
in the original or whether they had
been generated later. Allowing for millisecond
differences, if they were in the original
the timings would be similar. Every
"glitch" was:
(i) listened through on Beyerdynamic
headphones for the typical sound of
a "glitch", and
(ii) the waveform on either side of
the "glitch" was selected,
expanded and viewed for its shape.
(iii) This shape was compared with the
waveform shape on each side of the "glitch"
to decide whether it was "musical"
in shape, even though with a sharp (vertical
almost) slope at its front.
In addition,
if there was "just noise"
it could be random as against reproducible
if embedded in the soundstage.
Table:
Results of a "glitch search on
the Review CD-R and the Master CD-R
The Threshold Slope (range –60 60
–6dB) was set at –20dB
The Sensitivity (range 10 to 100) was
set at 50
Track
|
Review
CD-R time
|
Master
CD-R time
|
Comment
|
1
|
08:08.186
|
08:08.140
|
A
very small glitch.
|
2
|
NE
|
|
|
3
|
02:27.185
|
02:27.205
|
Steep
music slope
|
4
|
NE
|
|
|
5
|
NE
|
|
|
6
|
01:59.447
02:58.595
|
01:59.467
02:58.615
|
Steep
music slope
Steep music slope
|
7
|
NE
|
|
|
8
|
NE
|
|
|
9
|
NE
|
|
|
10
|
00:33.520
|
00:33.540
|
Steep
music slope
|
11
|
02:42.873
|
02:42.907
|
Steep
music slope
|
12
|
NE
|
|
|
13
|
NE
|
|
|
14
|
04:48.690
05:33.076
|
04:48.724
05:33.110
|
Steep
music slope
Steep music slope
|
15
|
NE
|
|
|
Results
There
were eight "events" reported
in the Review CD-R of which only one
was a very small "true" glitch.
These were "matched" – within
acceptable mathematical tolerances –
by the information from the Master CD-R.
Comment:
The conclusion drawn is that the
"crackling noises" could not
have come from the embedded audio material
or artefacts therein.
In addition, these "noises"
have not been reported by anyone else
who has purchased this product.
Addressing
the problem (4): Examination of the
original manufacture of the Master
Unfortunately,
the waveform from which the Master was
made on November 11th, 2003 is no longer
on the active part of neither the Music
Computer nor its Parking Area. However,
a 1:1 Safety copy has been retained
on DAT.
Because another source of unwanted noise
can arise at the Bit-depth Conversion
stage, the original records have been
checked. These show that the conversion
from 24bit to 16bit was done with a
Dither setting "Highpass Triangular"
and the Noise Shape was "High-pass
contour". These are settings that
have been used for all of Dunelm’s recordings
made at 24bit. No noise will emanate
from these settings.
Addressing
the problem (5): Independent examination
of the Review CD-R
Jim Pattison will give this report –
and offer the returned Review CD-R –
to Mr. Nigel Winchcombe, an independent
sound engineer who formerly worked for
the British Broadcasting Corporation,
on his return next week from Poland,
and ask for an independent assessment
of the problem*.
*Another
copy of this CD, viz. DRD0201 has been
reviewed by David Rothery who commented:
"The recording is clean and well-rounded.
The booklet notes by McLachlan and Ian
Milnes are interesting. A recording
to treasure."
Addressing the problem (6): A new
clone
A new clone, on different stock,
using the original master on the Marantz
copying machine, will be made and sent
to Mr. Satz for assessment of the sound
quality.
Addressing
the problem (7): Previous experience
The use of CD-Rs stock for making
CDs in small numbers has been extant
since about 1995.
Although
it was believed that the manufacturing
process was sufficiently developed to
enable CD-Rs to be played on conventional
CD players without difficulty, it was
found that the pseudo-hybrid nature
of those CD-Rs which used a dye layer
as the recording medium could "fool"
the laser-reading/tracking mechanisms
of the early player models because the
angle of incidence of the laser beam
was unprecise and did not focus properly.
In addition, lenses were hardly ever
cleaned and a film of condensation,
dust/cigarette smoke also diffused the
beam.
Further,
it was not made clear – for obvious
reasons – that, almost imperceptibly,
the power of lasers decreased with time
and use until a point was reached that
"noise" was being created
in the playing process. CD-Rs that would
not play on an "older" machine,
always played on the newer models (whose
specifications had tacitly been updated
in the meantime by the manufacturers!
Dunelm
Records has made more than 5000 music
CD-Rs over the twelve years of its existence.
In that time there have been approximately
10 reports of playing difficulties.
Cleaning the lens or updating to a newer
machine cured nine of these. In one
case, the laser just failed!
Dunelm’s "no-quibble" guarantee
has been invoked twice in all of that
time – and one of those was believed
to be a "con".
This is
the record that this report hopes to
restore after the severe censure of
the review that started it all.
Jim
Pattison
Dunelm Records
June 20th, 2004.
Appendix
1: Optical discs
Optical recorders have the
advantage that light can be focused
at a distance – something that cannot
be done with magnetism – so there need
be no physical contact between the pickup
and the carrier with no wear mechanism.
Basically, the information on the disc
is in a form that simply conveys a series
of numbers, which are exactly those
recorded on the master tape. The readout
of a CD is through the disc thickness
and this tolerates scratches very well.
To play an optical disc a drive is needed
that has a spindle drive mechanism to
revolve the disc, a positioner to give
radial access across the disc surface.
This positioner carries a collection
of items all needed in the reading process.
Very accurate track following is required
and it takes some time to lock on to
a track. For this reason, tracks on
laser discs are usually made as a concentric
spiral.
To replay
any optical disc, a source of monochromatic
light is needed. Care must be taken
to keep the beam focused on the information
layer.
There are
two types of "simple" disc:
(i) has a thin metal layer in which
holes are punched on recording by heat
from a laser
(ii) has a layer of photochemical dye
which darkens when struck by a high-powered
recording beam.
Whatever
the recording principle, light from
the pickup is reflected more or less,
or absorbed more or less, so that the
pickup senses a change in reflectivity.
Conventional
CD players should be able to read certain
discs, called CD-Rs. However, it was
soon realised that the specifications
of some "early" CD players
were too "loose" and they
could not play CD-Rs. There was a theory
that the dye used was obscuring/diverting
the beam from the information.
This theory
was "strengthened" when the
industry produced hybrid CDs, called
SACDs, at the end of the 20th
century. These hybrid CDs contained
two recordings at different depths,
the lower layer being the conventional
CD recording at "Red Book"
standard. The laser optics in a standard
CD player are designed to focus on the
lower layer, ignoring the second semi-reflective
layer.
Laser optics
in some existing CD players may, in
fact, detect the semi reflective layer,
fail to make sense of it and either
reject the disc as unplayable or give
an artefact (noise, in this case).
Some companies
cured this by issuing Firmware Updates;
some just changed the models and put
a plate on the front saying that the
player was compatible with CD-R / RW.
Further
reading
1. The Quiet Revolution:
Optical disk and CD-ROM (November, 1995)
http://www.soundonsound.com/sos/1995_articles/nov95/opticaldisks.html?session=eeaf…
2. Creating Audio CDs & the future
of CD-R
http://www.soundonsound.com/sos/1996_articles/sep96/pcnotessep96.html?session=eea…
3. Burning Issues: Compiling an album
on CD-R (July, 1999)
http://www.soundonsound.com/sos/jul1999/articles/cdrmastering.htm?session=eeaf16b0e2
4. Burning Questions: Getting better
performance from your CD drives (January,
2003)
http://www.soundonsopund.com/sos/jul99/articles/cdrmastering.htm?session=eeaf16b0e2
5. Watkinson, J., The Art of Digital
Audio, Focal Press, Oxford, 1999
(2nd edition reprinted 1994).
Ref:
0201ds2 problem
Dunelm
Records: Investigation of review copy
of DRD0201:
McLachlan plays the 32 Beethoven Sonatas:
Volume 1 sent to Classical Musicweb:
Examination of the review copy of the
CD by Mr. Nigel Winchcombe of Precision
Electronic Production: Interim report
on progress
Introduction
The
work done by Jim Pattison of Dunelm
Records to June 20th, 2004,
has been reported in the document 0201ds
problem.
In that
document there is a section entitled:
Addressing the problem (5): Independent
examination of the Review CD-R.
Whilst Jim Pattison was on holiday
(June 26th to July 3rd, 2004),
Mr. Nigel Winchcombe, of Precision Electronic
Production – formerly one of the BBC’s*
Sound Engineers with 30 years’ experience
of sound mixing – has had an opportunity
to examine the returned review copy
of the CD as well as another copy made
from the Original CD Master but on different
stock.
By arrangement,
Mr. Winchcombe and Jim Pattison held
a discussion on Sunday afternoon, July
4th, 2004 and the following
interim report was tabled.
1. Mr Winchcombe observed that,
of the entire piano repertoire, Beethoven’s
"Pathétique" Sonata
is probably the most unforgiving work
to a recording medium. He had, therefore
concentrated his examination at this
stage on this sonata.
2. He had played the review CD
on six different CD players listening
to the playback from an equidistant
position between two BBC Monitoring
Loudspeakers. He had found that there
were very slight "clicks"
– somewhat similar to the low-level
clicks on a vinyl LP but nothing like
as bad as these.
3. Each of the six CD players
give very slight variations in the number
and quality of the "clicks"
as heard at very high playback levels
on the speakers.
4. Mr. Winchcombe stressed that
the "clicks" he had heard
were at a very low level and could not
be said to match (or even come near)
the description of the defect described
by the reviewer. This could be explained
by postulating that the error-correction
facility on the reviewer’s CD player
may not be as sophisticated as that
which exists on the players used in
this test.
5. Jim Pattison was asked to
describe, in some detail, the set-up
of the four AKG B-ULS condenser microphones
used at the recording session, their
recording patterns, heights and lineal
distances from the piano. Except for
the advice to tilt the centre stereo
pair a little more towards the strings
of the piano, the set-up described was
recognised as an approved one.
Further tests
The discussion then turned to devising
a further series of tests after Jim
Pattison had made certain variants of
CD preparation from the 1:1 DAT Safety
Copy of the waveform at 16 bit.
These variants
would be prepared whilst Mr. Winchcombe’s
was absent on a recording project for
the next 14 days.
Based on the results of these tests,
Mr. Winchcombe would decide on the next
step in the process of " Addressing
the problem".
Jim Pattison
July 4th, 2004.
* British
Broadcasting Corporation
on adds
I did experience some tracking problems
on my oldest CD player and discussed
the situation with Mr. Pattison of Dunelm
Records who asked that I return the
disc to him for extensive testing. The
testing was accomplished, and no tracking
problems were found. Mr. Pattison did
comment that the disc is a CD-R and
might have tracking problems on older
and less sophisticated audio equipment.
For my part, I can assure readers that
the disc played well on the three more
modern CD players that I own.