Extracting surround sound from a stereo recording

Archived from groups: rec.audio.pro (More info?)

In his (rather excellent) book, Mastering Audio, Bob Katz discusses a
process originally described by Dolby Laboratories called "Magic
Surround" in which surround sound information is extracted from a
stereo recording via delays. (Note he was writing two-mic true stereo
recordings, not multitracked panned mono)

What does anyone know about this? I know when I listen to my OSS
recordings with headphones I hear "surround". It seems one should be
able to achieve the same thing through loudspeakers. I don't have any
surround sound processing gear so I've never fiddled around with any
of this stuff, short of sitting directly between my monitors with the
drivers facing my ears (Not entirely satisfying.)

Thoughts, or experiences, anyone?

Kelly Dueck
19 answers Last reply
More about extracting surround sound stereo recording
  1. Archived from groups: rec.audio.pro (More info?)

    > In his (rather excellent) book, Mastering Audio, Bob Katz discusses
    > a process originally described by Dolby Laboratories called "Magic
    > Surround" in which surround sound information is extracted from a
    > stereo recording via delays. (Note he was writing two-mic true stereo
    > recordings, not multitracked panned mono)

    > What does anyone know about this? I know when I listen to my OSS
    > recordings with headphones I hear "surround". It seems one should be
    > able to achieve the same thing through loudspeakers. I don't have any
    > surround sound processing gear so I've never fiddled around with any
    > of this stuff, short of sitting directly between my monitors with the
    > drivers facing my ears (Not entirely satisfying.)

    I've been involved with surround sound for almost 35 years -- listening to it,
    recording it, reviewing it, and writing about it. The following is a reasonably
    complete and accurate explanation (unlike the other responses posted).

    All encoded surround systems use greater or lesser degrees of
    "out-of-phase-ness" to encode rear and side information. Simply subtracting R
    from L produces a signal in which mono components are cancelled, and front
    components are attenuated in proportion to how closely they're panned to
    center-front. The out-of-phase components are strengthened in proportion to how
    closely they are to being 180 degrees out of phase. The net result is that,
    over-all, out-of-phase trumps in-phase. Okay?

    The "direct" components in a recording have a fixed phase relationship. The
    "ambient" components are of continuously varying phase. This means that taking
    L-R even from a recording that isn't explicitly encoded produces much the same
    result as if the recording were encoded -- the ambient components are
    strengthened, the direct weakened. This system is called "Dynaquad," and it's
    been around since 1970. It can be added to almost any system simply by adding
    two speakers wired in series across the "hot" L and R amplifier terminals, no
    amps or decoders required.

    A similar effect occurs in active decoders. My experience has been that
    Ambisonic UHJ decoders produce the best results, SQ the worst, with QS falling
    in-between (but closer to UHJ).

    As for delay... The Haas (or precedence) effect states that delayed sounds
    arriving within 5 to 20ms of the initial sound are not heard as separate sound
    sources, regardless of where they come from. (We're talking identical sounds, of
    course.)

    So... If I play the ordinary L and R signals, delayed, through speakers set to
    the sides, the direct components from the side speakers are masked by the direct
    sounds from the front. But the ambient components are of varying phase, and not
    subject to Haas masking. So, in effect, the ambient components are "unmasked"
    and now audible around the listener.
  2. Archived from groups: rec.audio.pro (More info?)

    Someone can correct me if I am wrong. As I understand surround from stereo
    via delays it is simply calling a goose a duck.
    If you want a two mic surround as I understand it, you have to use two
    figure 8's to code as two ms arrays or an equiv.

    Rich

    "Kelly Dueck" <kellyd@escape.ca> wrote in message
    news:cd189750.0409011345.5bdbd5b7@posting.google.com...
    > In his (rather excellent) book, Mastering Audio, Bob Katz discusses a
    > process originally described by Dolby Laboratories called "Magic
    > Surround" in which surround sound information is extracted from a
    > stereo recording via delays. (Note he was writing two-mic true stereo
    > recordings, not multitracked panned mono)
    >
    > What does anyone know about this?
  3. Archived from groups: rec.audio.pro (More info?)

    Rich Peet wrote:

    > Someone can correct me if I am wrong. As I understand surround from stereo
    > via delays it is simply calling a goose a duck.
    > If you want a two mic surround as I understand it, you have to use two
    > figure 8's to code as two ms arrays or an equiv.

    Exzctly right. It just creates a plausible illusion.


    Bob
    --

    "Things should be described as simply as possible, but no
    simpler."

    A. Einstein
  4. Archived from groups: rec.audio.pro (More info?)

    Your "complete response" appears to make a number of assumptions regarding
    the mic configuration and its attitude toward phase.
    Use of 100' wide separation omni's, vs 3" spaced omni's within a 1:1
    parabolic, vs sass, ms, ortf, binaural, etc. etc.
    You are telling me that each decodes phase to rear the same? And that no one
    inserts, pans, or artificially places a subject in a channel?

    It sounds to me as you are telling me that MS is the only microphone config
    allowed for virtual surround.
    I guess I just don't understand enough and will just go make surround
    "stuff" by how I can comprehend.

    Rich

    The following is a reasonably
    > complete and accurate explanation (unlike the other responses posted).
    >
    > All encoded surround systems use greater or lesser degrees of
    > "out-of-phase-ness" to encode rear and side information. Simply
    subtracting R
    > from L produces a signal in which mono components are cancelled, and front
    > components are attenuated in proportion to how closely they're panned to
    > center-front. The out-of-phase components are strengthened in proportion
    to how
    > closely they are to being 180 degrees out of phase. The net result is
    that,
    > over-all, out-of-phase trumps in-phase. Okay?
  5. Archived from groups: rec.audio.pro (More info?)

    > Your "complete response" appears to make a number of assumptions
    > regarding the mic configuration and its attitude toward phase.
    > Use of 100' wide separation omni's, vs 3" spaced omni's within a 1:1
    > parabolic, vs sass, ms, ortf, binaural, etc. etc.
    > You are telling me that each decodes phase to rear the same? And
    > that no one inserts, pans, or artificially places a subject in a channel?

    > It sounds to me as you are telling me that MS is the only microphone
    > config allowed for virtual surround.
    > I guess I just don't understand enough and will just go make surround
    > "stuff" by how I can comprehend.

    It makes NO SUCH assumptions. The explanation is generic, because the ability to
    extract surround information described DOES INDEED WORK "generically" IN THE WAY
    DESCRIBED for any stereo recording that has ambient or random-phase components.

    Instead of reading what I actually wrote and thinking about it, you have
    projected your own preconceptions on it and completely distorted it.

    It is indeed correct that you don't understand enough. You not only don't
    understand much about surround sound, you understand next to nothing about
    recording or acoustics.

    What I wrote was and is conceptually correct and complete. Had I written it out
    in extreme detail for someone who knew nothing whatever about the subject (eg,
    you), it would have been three to four times as long.

    Why do I waste time writing for people who don't bother to think about what they
    read?
  6. Archived from groups: rec.audio.pro (More info?)

    "William Sommerwerck" <williams@nwlink.com> wrote in message news:<10jd33d98hus9a0@corp.supernews.com>...
    >
    > As for delay... The Haas (or precedence) effect states that delayed sounds
    > arriving within 5 to 20ms of the initial sound are not heard as separate sound
    > sources, regardless of where they come from. (We're talking identical sounds, of
    > course.)
    >
    > So... If I play the ordinary L and R signals, delayed, through speakers set to
    > the sides, the direct components from the side speakers are masked by the direct
    > sounds from the front. But the ambient components are of varying phase, and not
    > subject to Haas masking. So, in effect, the ambient components are "unmasked"
    > and now audible around the listener.

    This is more, or less, how Bob Katz explains it in Mastering Audio,
    although he doesn't go into as much detail about phase, which would
    have been helpful. Do you have a suggested starting for delay times?
    (Around 20 ms, for example) Or should one just "fiddle" with the delay
    time to the side speakers until it sounds convincing?
  7. Archived from groups: rec.audio.pro (More info?)

    > This is more or less, how Bob Katz explains it in Mastering Audio,
    > although he doesn't go into as much detail about phase, which would
    > have been helpful. Do you have a suggested starting for delay times?
    > (Around 20 ms, for example) Or should one just "fiddle" with the delay
    > time to the side speakers until it sounds convincing?

    10 to 20 ms would be good starting values. Fiddle away.

    Why is it that people want an "exact" answer to something that can easily be
    determined by simple experimentation? It's not like you're going to damage
    something, for heaven's sake.
  8. Archived from groups: rec.audio.pro (More info?)

    alrighty then
  9. Archived from groups: rec.audio.pro (More info?)

    "Rich Peet" <RichPeet@comcast.net> wrote in message
    news:0sFZc.14232$3l3.10770@attbi_s03...
    > alrighty then
    >
    >
    http://home.comcast.net/~richpeet/tones.mp3
  10. Archived from groups: rec.audio.pro (More info?)

    In article <10jeh9pnuth17e0@corp.supernews.com> williams@nwlink.com writes:

    > 10 to 20 ms would be good starting values. Fiddle away.
    >
    > Why is it that people want an "exact" answer to something that can easily be
    > determined by simple experimentation?

    Well, in this case, he did ask for a starting place. It may not be
    obviousl that it's in this ballpark and not 50 or 100 msec. But I
    agree that asking for "the settings to EQ a kick drum" is pretty
    useless.


    --
    I'm really Mike Rivers (mrivers@d-and-d.com)
    However, until the spam goes away or Hell freezes over,
    lots of IP addresses are blocked from this system. If
    you e-mail me and it bounces, use your secret decoder ring
    and reach me here: double-m-eleven-double-zero at yahoo
  11. Archived from groups: rec.audio.pro (More info?)

    So ya see billie.
    That is if you still have various virtual surround equipment, that there is
    a relationship between time of arrival, pitch, and phase. Therefore only
    stereo mics with overlapping elements will maintain the phase.

    Rich
    PS I will stop talking Minnesotian if you will stop being a grumpy old man.

    "Rich Peet" <RichPeet@hotmail.com> wrote in message
    news:AzHZc.358284$%_6.186125@attbi_s01...
    >
    > "Rich Peet" <RichPeet@comcast.net> wrote in message
    > news:0sFZc.14232$3l3.10770@attbi_s03...
    > > alrighty then
    > >
    > >
    > http://home.comcast.net/~richpeet/tones.mp3
    >
    >
  12. Archived from groups: rec.audio.pro (More info?)

    > So ya see billie.
    > That is if you still have various virtual surround equipment, that there is
    > a relationship between time of arrival, pitch, and phase. Therefore only
    > stereo mics with overlapping elements will maintain the phase.

    > PS I will stop talking Minnesotan if you will stop being a grumpy old man.

    I AM A GRUMPY OLD MAN! (Geezer is more like it.) I've been one since early
    childhood.

    Briefly... Regardless of how you mic a recording, the direct sounds from an
    instrument maintain a fixed amplitude/phase/time relationship with each other
    (assuming the instrument doesn't move). Subtracting R from L in playback tends
    to cancel these, with greater cancellation occurring towards the center.

    But the ambient components represent repeated reflections and their multiple
    mixings -- again, regardless of how you mic. The result is that there are many
    anti-phase (and near-anti-phase) components that are _not_ cancelled when R is
    subtracted from L. L-R is therefore a signal in which the direct sounds are
    selectively attenuated, and the ambient sounds preferentially enhanced. You can
    hear this in a moment by listening to the difference signal of any orchestral
    recording, regardless of how it was miked (including pan-potted multi-miking).

    I can't explain it any more-simply than that.
  13. Archived from groups: rec.audio.pro (More info?)

    what ever

    I think my sound file proved that not to be the case.

    Rich Peet
    deep in the swamp muck and going deeper

    "William Sommerwerck" <williams@nwlink.com> wrote in message
    news:10jf9acn5bq4v8c@corp.supernews.com...
    > > So ya see billie.
    > > That is if you still have various virtual surround equipment, that there
    is
    > > a relationship between time of arrival, pitch, and phase. Therefore only
    > > stereo mics with overlapping elements will maintain the phase.
    >
    > > PS I will stop talking Minnesotan if you will stop being a grumpy old
    man.
    >
    > I AM A GRUMPY OLD MAN! (Geezer is more like it.) I've been one since early
    > childhood.
    >
    > Briefly... Regardless of how you mic a recording, the direct sounds from
    an
    > instrument maintain a fixed amplitude/phase/time relationship with each
    other
    > (assuming the instrument doesn't move). Subtracting R from L in playback
    tends
    > to cancel these, with greater cancellation occurring towards the center.
    >
    > But the ambient components represent repeated reflections and their
    multiple
    > mixings -- again, regardless of how you mic. The result is that there are
    many
    > anti-phase (and near-anti-phase) components that are _not_ cancelled when
    R is
    > subtracted from L. L-R is therefore a signal in which the direct sounds
    are
    > selectively attenuated, and the ambient sounds preferentially enhanced.
    You can
    > hear this in a moment by listening to the difference signal of any
    orchestral
    > recording, regardless of how it was miked (including pan-potted
    multi-miking).
    >
    > I can't explain it any more-simply than that.
    >
  14. Archived from groups: rec.audio.pro (More info?)

    I don't want to beat this into the ground (this will be my last posting), but
    what I described (L-R ambience extraction) was first suggested by the late David
    Hafler in 1970. Many people (myself included) adopted it and found it provided
    excellent enhancement with virtually any recording that contained ambience.
    Given "normal" miking techniques, you would have to deliberately engineer the
    recording to make it not work (ie, by grossly suppressing the ambience).

    The fact is, it "looks good" on paper, and works in practice. The addition of
    delay makes it even better. Sony made a nice little unit that provided
    difference/delay without screwing up the main channels. I reviewed it in the
    late '80s for Stereophile. It sounded good, and worked very well.
  15. Archived from groups: rec.audio.pro (More info?)

    OK my last post as well.
    You seem to think that I am working in theory and something in my mind.
    You were wrong in more than a couple places as I would not post in a pro
    audio group without some basis.
    Turn on the scope if you don't want to connect the equipment.

    Rich

    "William Sommerwerck" <williams@nwlink.com> wrote in message
    news:10jfgflq1ag84b2@corp.supernews.com...
    > I don't want to beat this into the ground (this will be my last posting),
    but
    > what I described (L-R ambience extraction) was first suggested by the late
    David
    > Hafler in 1970. Many people (myself included) adopted it and found it
    provided
    > excellent enhancement with virtually any recording that contained
    ambience.
    > Given "normal" miking techniques, you would have to deliberately engineer
    the
    > recording to make it not work (ie, by grossly suppressing the ambience).
    >
    > The fact is, it "looks good" on paper, and works in practice. The addition
    of
    > delay makes it even better. Sony made a nice little unit that provided
    > difference/delay without screwing up the main channels. I reviewed it in
    the
    > late '80s for Stereophile. It sounded good, and worked very well.
    >
  16. Archived from groups: rec.audio.pro (More info?)

    In addition to what Mr. Sommerwerck says elsewhere, I should add that the
    problem with listening to a stereo recording through Dolby Pro-Logic surround
    decoding is that you get unpredictable results.

    It's predictable in that it sounds the same every time you do it, but it's
    UN-predictable as far as getting a result that can be radically different
    from what the recording artist originally intended for their audience to
    hear.

    It can be entertaining to listen to stereo recordings this way, but at least
    with Pop & Rock recordings, you sometimes get some weird stuff coming out of
    the surround (aka rear) speakers. For that reason, it's not too kosher to
    me.

    --MFW
    [remove the extra M above for email]
  17. Archived from groups: rec.audio.pro (More info?)

    Marc Wielage <mfw@mmusictrax.com> wrote in message news:<0001HW.BD5E8F310005B01BF05095B0@news-server.socal.rr.com>...
    > In addition to what Mr. Sommerwerck says elsewhere, I should add that the
    > problem with listening to a stereo recording through Dolby Pro-Logic surround
    > decoding is that you get unpredictable results.
    >
    > It's predictable in that it sounds the same every time you do it, but it's
    > UN-predictable as far as getting a result that can be radically different
    > from what the recording artist originally intended for their audience to
    > hear.
    >
    > It can be entertaining to listen to stereo recordings this way, but at least
    > with Pop & Rock recordings, you sometimes get some weird stuff coming out of
    > the surround (aka rear) speakers. For that reason, it's not too kosher to
    > me.
    >
    > --MFW
    > [remove the extra M above for email]

    Well I wouldn't expect to get predictable results with a pop recording
    that was close miked and multi-tracked in a relatively dead studio and
    made use of a multitude of artificial reverbs and incompatible
    "spaces".

    Imagine listening to a stereo recording of a large scale rock
    production featuring lead vocals with a 50's style slap echo, dry
    guitars, drums tracked in a "drum room" and overdubbed strings
    swimming in "Large Hall." Mixed to surround it might sound as if you
    were sitting in a bathroom inside a drum room inside a concert hall
    with the guitarist on your lap. Not a pretty picture.
  18. Archived from groups: rec.audio.pro (More info?)

    > It can be entertaining to listen to stereo recordings this way,
    > but at least with pop & rock recordings, you sometimes get
    > some weird stuff coming out of the surround (aka rear) speakers.
    > For that reason, it's not too kosher to me.

    Some of the "system" decoders (SQ, QS) included circuitry that pre-processed
    regular stereo signals so that they were wrapped into a horseshoe from LR around
    to RR after being run through the decoder. This produced consistent results with
    few or no side-effects.

    Kosher or not, you can try and see what happens. Don't like the results? Switch
    out the decoder.
  19. Archived from groups: rec.audio.pro (More info?)

    On Sep 4, 2004, William Sommerwerck <williams@nwlink.com> commented:

    > Some of the "system" decoders (SQ, QS) included circuitry that pre-processed
    > regular stereo signals so that they were wrapped into a horseshoe from LR
    > around
    > to RR after being run through the decoder. This produced consistent results
    > with
    > few or no side-effects.
    >--------------------------------snip----------------------------------<

    You should play around with the Lexicon MC-12. Lotta very interesting
    surround modes on this thing. Pro Logic II is particularly interesting,
    though the derived stereo surround is a little phasey to me.

    But I'm still not happy about the idea of listening to an artist's work in a
    way that they would really dislike.

    --MFW
    [remove the extra M above for email]
Ask a new question

Read More

Pro Audio Audio