Matrix(ed) recordings

[Billster]

Though I've been trading a good while, I only fairly recently became cognizant of "matrix" recordings.

My two questions are - are matrix recordings essentially 2 or more recordings of the same event, combined into a single stream?

And second question, what is the value of matrix recordings? My assumption/guess is that the point is to use each recording to enhance the strong points of the recording so as to create a more full sound (appropriate crowd noise, bass but not too much, drums but not too much, etc) than either recording might represent standing alone - is this correct? Is this the spirit behind matrix recordings, and are there other reasons beyond this one for creating them?

[brainysod]

My two questions are - are matrix recordings essentially 2 or more recordings of the same event, combined into a single stream?

Yes.

And second question, what is the value of matrix recordings? My assumption/guess is that the point is to use each recording to enhance the strong points of the recording so as to create a more full sound (appropriate crowd noise, bass but not too much, drums but not too much, etc) than either recording might represent standing alone - is this correct? Is this the spirit behind matrix recordings, and are there other reasons beyond this one for creating them?

Yes, mainly to polish a soundboard recording by adding in the audience and to make the instruments sound like they werent recorded in a room. And for the audience recording - to make the instruments clearer.

[ozzyzak]

I think that's right Bill....Aud/Sbd mix, I believe are so that you get the get instrument sound but without being sterile, so they might throw in a little aud for crowd noise/ambience. I'm not an expert but that'd be my guess.

[lucasweb]

Yea, the posts above are correct as far as i know, also i've seen matrixes done with multiple aud sources if there isn't a soundboard source available

[Billster]

Yea, the posts above are correct as far as i know, also i've seen matrixes done with multiple aud sources if there isn't a soundboard source available

Yes, I think that's what was done with Radiohead 5/1/04, isn't it? Combined 3 different audience sources (I am 90% sure) into one...I guess the idea is to highlight the strengths of each....I assume fades and relative volume levels are used appropriately for this reason?

The 5/1/04 3-source matrix sounds remarkably like a good soundboard.

If anyone knows and wants to speak to how these matrices are actually made, I think it'd be interesting reading. What I mean is, is it a lot of trial and error to get a particular matrix sounding good, or do guys have a pretty good idea what role each of the 2, 3, whatever, sources needs to play?

[wazoo2u]

:nono:
OK, here we go again. What are commonly (inaccurately) referred to as "Matrix" recordings are in fact "Blends" of independently recorded SBD and AUD sources of an event. In the audio world, a matrix recording is in fact:

Mid-side Matrix
One cardioid mic (M) points at subject (0°), and a figure-8 mic (S) picks up ambient sound by pointing at sides(90° and 270°).

The signals are then run through a decoder matrix (software or hardware) which creates two stereo channels:
n M+S = channel one
n M - S = channel two

One channel is a mix of M and S and the other be a mix of M and Sø (which means S is 180° out of phase)

n The more S, the stronger the separation.

MS Matrix recording is mostly used in acoustic concert settings to capture the true image of the hall.

One technique that Dan Healy perfected when mixing and recording the Grateful Dead, was to place a MS Matrixed microphone array at the mix position, and record it on a source that was locked to the same source as the SBD recording (assuring that the recordings were speed accurate to each other) What Dan called an "Ultra Matrix" was (as I understand it) actually a BLEND of a Matrix recording and a SBD recording. There are also time coincidence issues that are addressed when combining these sources.

Many "Matrix" recordings in circulation are simply blends of 2 different recordings and present a significant challenge in Post Production in terms of correctly time and pitch aligning the sources. The results of these efforts often sound really good, but are difficult to produce.

The cow has left the barn in terms of correcting the proper use of the term "Matrix", but in fact, it's nice to know the proper terms to apply to these very different methods.

[Billster]

Thanks wazoo, I didn't know any of that! Very interesting stuff.

[TheMamba]

Nice explanation. Thanks!

[feralicious]

wazoooooooooooooooooooooooooo...

doing his best to catch the cow. :)

[wazoo2u]

wazoooooooooooooooooooooooooo...

doing his best to catch the cow. :)
:lol :lol: :lol :lol::lol :lol::lol :lol::lol :lol::lol :lol::lol :lol:

[jcrab66]

a true matrix is also the combined output of all single channel inputs into a soundboard through 2 channels

[wazoo2u]

a true matrix is also the combined output of all single channel inputs into a soundboard through 2 channels

I've never heard the term applied to the configuration you describe. I would describe this (a regular soundboard recording) as a summed mix.

Actually, I would think that in concept, a matrix would exist within a soundboard as the relationship between all inputs and their ability to be selectively assigned to ALL outputs, at least, that's how I would commonly utilize any sort of routing or patching matrix, and that relationship is basically how a soundboard is defined ie: a rectangular array of elements (or entries) set out by rows and columns.

I believe that what we're discussing here is a recording technique that utilizes a specific kind of encoding, not routing. You can see the reference to MATRIX CO-EFFICIENTS in this description of MS technique: http://www.uneeda-audio.com/ms-mat.htm

[rerem]

:nono:
OK, here we go again. What are commonly (inaccurately) referred to as "Matrix" recordings are in fact "Blends" of independently recorded SBD and AUD sources of an event. In the audio world, a matrix recording is in fact:

Mid-side Matrix
One cardioid mic (M) points at subject (0°), and a figure-8 mic (S) picks up ambient sound by pointing at sides(90° and 270°).

The signals are then run through a decoder matrix (software or hardware) which creates two stereo channels:
n M+S = channel one
n M - S = channel two

One channel is a mix of M and S and the other be a mix of M and Sø (which means S is 180° out of phase)

n The more S, the stronger the separation.

MS Matrix recording is mostly used in acoustic concert settings to capture the true image of the hall.

One technique that Dan Healy perfected when mixing and recording the Grateful Dead, was to place a MS Matrixed microphone array at the mix position, and record it on a source that was locked to the same source as the SBD recording (assuring that the recordings were speed accurate to each other) What Dan called an "Ultra Matrix" was (as I understand it) actually a BLEND of a Matrix recording and a SBD recording. There are also time coincidence issues that are addressed when combining these sources.

Many "Matrix" recordings in circulation are simply blends of 2 different recordings and present a significant challenge in Post Production in terms of correctly time and pitch aligning the sources. The results of these efforts often sound really good, but are difficult to produce.

The cow has left the barn in terms of correcting the proper use of the term "Matrix", but in fact, it's nice to know the proper terms to apply to these very different methods.
I think that is a technically accurate explanation,though the term has come to mean a mixed recording with dual sources,such as a board feed and a mic or two,with some % of each getting recorded. If it is an after the fact blending,it is often called a merge. There is a Hendrix New Years eve show that was merged,in large part because the board and a couple of auds each had significant shortcomings but blending allowed for a more listenable and complete show. A Source might have low hiss,but very weak bass,another may have hi-freq hiss,but full freq response while another has good vocal presence but is incomplete. With enough time and software it is possible to assemble a recording that is a better representation of the show than any one source.
For a mixed matrix'or "matrix" its about the SBD clarity and detail +the aud ambience,audience reactions.
At one point Healy used Beta HI-Fi with the board going through a PCM unit and an AUD feed going onto the Beta HIFI. That enabled a synched 4 channel recording that later could be mixed to the best ratio. The board rec was digital,the aud,analog,but Beta HIFI-or for that matter,VHS hifi,gives a good 20-20k response with a real nice SN ratio,enough for the job.

[RJ Hythloday]

^^ I realize this is a very old thread, but I was wondering if there is a list of Healy Ultra Matrixes. I'm guessing a lot of the sbds that circulate from his era are U-mtx but I know it's not all of them.

[co9ol]

Hey guys, I was wondering if anyone out there would be interested in providing an in-depth tutorial on how to make a matrix/blend. I think it'd be really interesting and we could all learn a lot.

[Karst]

:nono:

[Five]

its not a beginner's task... here's something to get you started:

bass sounds best coming from the soundboard, not echoing around the room, so try to get most of the bass from the dry soundboard not from the audience source if at all possible.

it is very similar to replacing the audio on a DVD, except you blend the two audio sources

some info here (esp check Author85's post)
http://www.thetradersden.org/forums/showthread.php?t=37589

you'll need a smidge of eq to make them sound natural together (don't overdo it)
http://www.thetradersden.org/forums/showthread.php?t=3514

and if your primary source isn't running at the correct speed, you will need to do pitch correction... this will also come in handy when stretching/shortening the 2nd souce
Here's the method I came up with to speed correct shows. I hope I've written
it clearly enough. If not, flame me. Ask any questions you may have. It is
a work in progress so I will be updating/clarifying as we go.

Jim

-------------------------------------------------------------------------------
Steps to speed correcting using Audacity, EAC, and SHNTool (or equivalent):
-------------------------------------------------------------------------------

1) Determine speed correction needed for each track. (Details below in section
"Step 1 Detailed Description")

2) If the speed correction is not a constant for the entire show, determine
where you want to cut the sections for each speed. For example, tracks
1-4 need to be slowed down 3%. Track 5 only needs to be slowed down 1%.
Listen to the transition between tracks 4 and 5. Is there a tape cut? If
so, that's where you want to have the transition which changes speed
correction from -3% to -1%. If not, choose a location with minimal volume.
Do this for the entire show.

You can make the process easier if you simply use the existing track
boundaries as section transitions. This does not always produce the best
results though. Your choice.

3) Begin speed correcting sections you determined in Step 2. In Audacity,
appending WAVs is a little tedious. First open the first track, then open
the second track in a new window. Select all of the second track (CTRL A)
and copy (CTRL C). Now go back to the first track, select Edit->
Move Cursor->To Track End. Then paste (CTRL V). Close the window for the
second track.

Repeat this until the current section is complete. If the section boundary
you determined in Step 2 is not an existing track boundary, you will need
chop the last track of the current section at this boundary. Open the last
track of the current section, simply move cursor to the boundary point you
determined, select Edit->Select->Start To Cursor. Then do (CTRL X). Now
append the portion you just deleted to the end of the current section. Don't
close the window containing the remaining portion of the last track of the
section. It will be the first part of the next section. Finally, once the
section is build, go to Tools->Change Speed. Enter desired speed change.

Repeat this procedure for each section. You can keep a window open for each
completed section or save each. I like to keep the windows open to avoid
any change of corrupting data through the save and then re open process.

4) When all the sections are build and speed corrected, it's time to append the
sections together (see Step 3 above for appending process). Each new
section that gets appended could have a section transition problem. After
you append the section, you must zoom in to the transition boundary to make
sure there is not a discontinuity. This transition point is easy to find
because the newly appended section will be shaded darker than the rest of
the WAV. Zoom all the way until you can see the individual samples around
the section transition point. Look for a smooth waveform at the transition.
If not smooth, adjust samples around the discontinuity to make it smooth.
Once zoomed in, there's a tab in the upper left corner of the display to
change to sample edit mode. Once the signal looks smooth, listen to the
transition to make sure there is no pop or click. Repeat this for all
sections.

5) Once all sections have been appended, save off the complete show or disc.

6) To cut the new tracks, I use EAC. Using EAC's Process WAV feature, open the
WAV file of the entire disc or show created. Starting at last track, select
portion of WAV to be included, move mouse to select all of last track, then
go to File->Save Selection. Then delete this selected portion of the WAV.
Repeat this until you have saved off all the tracks and have only the first
track left of the huge WAV you started with.

7) Fix SBEs in tracks just cut. Use whatever you want to do this. I use
SHNTool.

That's it.

-------------------------------------------------------------------------------
Step 1 Detailed Description
-------------------------------------------------------------------------------

First open the track up in Audacity. Listen and try to locate points with
clear notes. For example, OTHAFA intro outro, STH start, BD or HB riff.
Select the desired portion of the WAV with the mouse.

Now go to View->Frequency Spectrum. A new window will come up with the
frequency spectrum of the selection. For the options on the bottom left side
of the window, select size 16384, rectangular window, frequency spectrum, and
logarithmic (log) scale.

Now maximize this window and you should see discrete spikes. These spikes will
correspond to tuning note frequencies. A, A#, B, B# etc. The key to
correcting is to determine where a note spike frequencies should be. Refer
to the include table below for concert pitch note frequencies. These are
the frequencies where the spikes should be assuming Zep was tuned to concert
pitch.

For example, you see A4 at 455 Hz. From the table you know it should be at
440 Hz. So the correction needed in percent is calculated by:
(1 - 455/440) * 100 = -3.4%.
The general equation is:
percent correction needed = 1 - (freq observed / freq expected) * 100

A positive percent means speed up and a negative percent means slow down.

It is important not to only look at one spike when doing this, look at several
and get a good average of what the correction needs to be. If a spike
frequency doesn't make sense, ignore it and look at others, or select a
different section of the WAV file.

Another possible gotcha is correcting to the wrong half step. For example,
you see a spike at 435 Hz. You assume it is A4 so you correct it to 440. It
is really G#3. So now the correction is a half step high (fast). I do a
sanity check with other known correct recordings to avoid doing this.

That's basically it.

-------------------------------------------------------------------------------
Concert Pitch Frequency Table
-------------------------------------------------------------------------------

-------A----Asharp--------B----------C----Csharp---------D----Dsharp--------E----------F-----Fsharp---------G-----Gsharp

------55----58.27---61.735---65.406---69.296---73.416---77.782---82.407---87.307---92.499---97.999---103.83

----110---116.54---123.47---130.81---138.59---146.83---155.56---164.81---174.61-------185-------196---207.65

----220---233.08---246.94---261.63---277.18---293.66---311.13---329.63---349.23---369.99-------392----415.3

----440---466.16---493.88---523.25---554.37---587.33---622.25---659.26---698.46---739.99---783.99---830.61

----880---932.33---987.77---1046.5---1108.7---1174.7---1244.5---1318.5---1396.9-----1480------1568---1661.2

---1760---1864.7---1975.5-----2093---2217.5---2349.3-----2489------2637---2793.8-----2960------3136---3322.4

---3520---3729.3---3951.1-----4186---4434.9---4698.6-----4978------5274---5587.7---5919.9---6271.9---6644.9





Fixed pitch font version. Copy to text editor and view with font set to Courier.

+-------+--------+--------+--------+--------+--------+--------+--------+--------+--------+--------+--------+
| A | Asharp | B | C | Csharp | D | Dsharp | E | F | Fsharp | G | Gsharp |
+-------+--------+--------+--------+--------+--------+--------+--------+--------+--------+--------+--------+
| 27.5 | 29.135 | 30.868 | 32.703 | 34.648 | 36.708 | 38.891 | 41.203 | 43.654 | 46.249 | 48.999 | 51.913 |
| | | | | | | | | | | | |
| 55 | 58.27 | 61.735 | 65.406 | 69.296 | 73.416 | 77.782 | 82.407 | 87.307 | 92.499 | 97.999 | 103.83 |
| | | | | | | | | | | | |
| 110 | 116.54 | 123.47 | 130.81 | 138.59 | 146.83 | 155.56 | 164.81 | 174.61 | 185 | 196 | 207.65 |
| | | | | | | | | | | | |
| 220 | 233.08 | 246.94 | 261.63 | 277.18 | 293.66 | 311.13 | 329.63 | 349.23 | 369.99 | 392 | 415.3 |
| | | | | | | | | | | | |
| 440 | 466.16 | 493.88 | 523.25 | 554.37 | 587.33 | 622.25 | 659.26 | 698.46 | 739.99 | 783.99 | 830.61 |
| | | | | | | | | | | | |
| 880 | 932.33 | 987.77 | 1046.5 | 1108.7 | 1174.7 | 1244.5 | 1318.5 | 1396.9 | 1480 | 1568 | 1661.2 |
| | | | | | | | | | | | |
| 1760 | 1864.7 | 1975.5 | 2093 | 2217.5 | 2349.3 | 2489 | 2637 | 2793.8 | 2960 | 3136 | 3322.4 |
| | | | | | | | | | | | |
| 3520 | 3729.3 | 3951.1 | 4186 | 4434.9 | 4698.6 | 4978 | 5274 | 5587.7 | 5919.9 | 6271.9 | 6644.9 |
| | | | | | | | | | | | |
| 7040 | 7458.6 | 7902.1 | 8372 | 8869.8 | 9397.3 | 9956.1 | 10548 | 11175 | 11840 | 12544 | 13290 |
| | | | | | | | | | | | |
| 14080 | 14917 | 15804 | 16744 | 17740 | 18795 | 19912 | 21096 | 22351 | 23680 | 25088 | 26580 |
+-------+--------+--------+--------+--------+--------+--------+--------+--------+--------+--------+--------+




Matlab Code to generate table:

ratio_per_note = 2^(1/12);
lowest_A = 440/8;
current_freq = lowest_A
for n=1:84
output_freqz(n) = current_freq;
current_freq = current_freq * ratio_per_note;
end

and of course the usual stuff, work from the two complete sources, not track by track, and be sure to cut on the sector boundaries at the end yada yada

take your time until its the best you can make it