wesayso, now that your system is well tuned, you are exposing how bad some recordings, mixes, and masters really are 
Given that we have similar tunings, I echo your comments.
In case you have not seen it, Bob Katz has a decent explanation as to what went wrong here: Loudness War: Peace is Almost Here! - Digital Domain: CD Mastering | Mastered for iTunes | Audio Mastering | Blu-Ray Mastering It's a 20 minute piece, but if you scroll down a bit, there is a good 2 minute video that you can download and try it on your system. Amazing to see/hear the difference.
I remember in the early 80's getting really excited on how well recorded and dynamic some of the early digital recordings, mixes and masters were. Examples like The Police, Synchronicity - Murder By Numbers with a DR of 18 and Stewart Copland's drums on the outro just kicking with dynamics. Same with Stevie Ray Vaughan and the song Tin Pan Alley another DR 18, Dire Straits, So Far Away DR20, Peter Gabriel, Security are a few examples. There are countless others
But then the loudness war starting taking hold as well explained in Bob's video above.
Then there are the remasters... and as you have already found out, not only hit n miss, but it seems that in the majority of cases, the dynamics have been squashed even more on the remasters than the original. This is a tough pill to swallow, especially if one has paid a premium price for it.
I am sure you know about the DR database: Album list - Dynamic Range Database DR is not the end all be all for determining audio quality, but it is a good place to start when hunting for you favorite version/master of a particular artist/group.
Transcribing vinyl to digital can be fairly easy as I describe in this article using Audacity (free): Computer Audiophile - Guide to Converting Analog Vinyl To Digital Files Using Windows
Like you, and I am sure many others, some of my vinyl has a lot of surface noise, with pops and clicks. If you scroll down the towards the end of the article, I compare a CD rip to a vinyl recording of the same master called "which is which". Here is the link to the download: http://audio.computeraudiophile.com/2012/1029/Which-one-is-which.wav
That vinyl was in horrible shape. But with some judicious use of declick and removing the noise, it is almost indistinguishable from the CD. Give it a listen to see if you can tell which is which.
If you have the means and time, it may very well be worth transferring your vinyl to digital, especially if the CD version(s) does not compare... I ended up doing that with several albums where the CD version was not as well mastered as the vinyl.
Given that we have similar tunings, I echo your comments.In case you have not seen it, Bob Katz has a decent explanation as to what went wrong here: Loudness War: Peace is Almost Here! - Digital Domain: CD Mastering | Mastered for iTunes | Audio Mastering | Blu-Ray Mastering It's a 20 minute piece, but if you scroll down a bit, there is a good 2 minute video that you can download and try it on your system. Amazing to see/hear the difference.
I remember in the early 80's getting really excited on how well recorded and dynamic some of the early digital recordings, mixes and masters were. Examples like The Police, Synchronicity - Murder By Numbers with a DR of 18 and Stewart Copland's drums on the outro just kicking with dynamics. Same with Stevie Ray Vaughan and the song Tin Pan Alley another DR 18, Dire Straits, So Far Away DR20, Peter Gabriel, Security are a few examples. There are countless others
But then the loudness war starting taking hold as well explained in Bob's video above.
Then there are the remasters... and as you have already found out, not only hit n miss, but it seems that in the majority of cases, the dynamics have been squashed even more on the remasters than the original. This is a tough pill to swallow, especially if one has paid a premium price for it.
I am sure you know about the DR database: Album list - Dynamic Range Database DR is not the end all be all for determining audio quality, but it is a good place to start when hunting for you favorite version/master of a particular artist/group.
Transcribing vinyl to digital can be fairly easy as I describe in this article using Audacity (free): Computer Audiophile - Guide to Converting Analog Vinyl To Digital Files Using Windows
Like you, and I am sure many others, some of my vinyl has a lot of surface noise, with pops and clicks. If you scroll down the towards the end of the article, I compare a CD rip to a vinyl recording of the same master called "which is which". Here is the link to the download: http://audio.computeraudiophile.com/2012/1029/Which-one-is-which.wav
That vinyl was in horrible shape. But with some judicious use of declick and removing the noise, it is almost indistinguishable from the CD. Give it a listen to see if you can tell which is which.
If you have the means and time, it may very well be worth transferring your vinyl to digital, especially if the CD version(s) does not compare... I ended up doing that with several albums where the CD version was not as well mastered as the vinyl.
In general I have to agree. All genres seem to sound really good making it clear that a good speaker really can play all genres.
But when searching for my childhood favourites I have heard the same song from the same band in (sometimes up to 3) different dressings (different pressing CD's) that are as far removed from each other as can be. Maybe I'll upload a sample. But I have thrown a lot of garbage CD examples from my HD already. This is no small difference I'm talking about. Some of them are a mess. A lot of them in the genre I'm looking for right now, sadly.
Through the years I should have skipped buying those remasters on sight. I was still playing the vinyl at the time.
They were able to come up with great dynamics and great recordings in the "old" 70's-80's days, even though they were so limited in gear and number of tracks allowed.
These days, it's unlimited everything, and most feel obligated to use ALL the tools they have at the same time, just to justify the expense. Same applies to these so-called "remasters".
One of the reason why I moved away from the DAW monsters that Logic, ProTools, etc have become and discovered MixBus by Harrison. Simple and to the point.
These days, it's unlimited everything, and most feel obligated to use ALL the tools they have at the same time, just to justify the expense. Same applies to these so-called "remasters".
One of the reason why I moved away from the DAW monsters that Logic, ProTools, etc have become and discovered MixBus by Harrison. Simple and to the point.
I am a huge Led Zeppelin fan too. I have all their major work. Some on vinyl, all of it on digital. Yeah, some recordings can suck, but hey they're sooo good that I can listen past the poor recordings. If you can post some samples of poor v good recordings of the same song, I'd be interested.
Also, regarding vinyl playback, there is so much good music on vinyl that it would be a waste to give up on the medium. I mostly listen to digital nowadays, but I do have a bunch of LPs that I listen to often. I use a Technics SL1200 with DL103 into Pass Pearl II. But then it goes back into my PC and comes out the DAC because I use convolution to do EQ and crossover. The distortion is clearly higher on vinyl, but there is so much good music that I am not going to give it up yet.
Last edited:
I have thought long and hard how I could supply examples for what I mean.
It isn't that easy because I'd have to supply it as AAC encoded samples or my server would fill up fast.
The 2 main problems I encounter with stuff I know well are dynamic range compression and balance.
I have songs I can still enjoy with only a DR of about 7. But if you really crank them the faults of the over use of compression become irritating. While at more sane levels they still can sound decent enough to fully enjoy them.
But I also have stumbled over samples with DR numbers of 16, but no good balance in the mix. It will sound pretty good when cranked but due to the balance it doesn't work for me at my preferred play back level.
What I'm getting at is there is a huge variety in the mastering from the same music. The hardest part is to find the good balance that works, sounding like I remember my preferred songs. I'm not claiming superiority of vinyl, just a lack of consistency in CD's.
As Mitch pointed out the effects of the Loudness Wars are very real. But sadly that doesn't mean that a recording with higher DR numbers automatically sounds better. A lot of it seems to depend on the SPL level used when creating the songs.
Even the Led Zeppelin Remasters still have faults and warts. I'm glad they didn't go overboard and polish them. But at least the mastering seems more like how I remember my old vinyl versions. There is a good sense of depth, it is pleasing to listen to and it sucks me into the song. The CD's I tried before always sounded bright (sometimes harsh) in comparison.
Kudo's to you ra7, if you're not plagued by this. But to me, remembering the music as I have played it all my life (probably a more romantic view than an accurate one) did not convey on CD with some of them. It's like someone messed up my tone controls while I wasn't looking. I do have a lot of CD's where this isn't a problem. It's mostly the old stuff I remember from my vinyl days.
I will never get rid of my vinyl collection, but with my current family life I have no physical space to implement it into our living room. When I was living on my own the CD and vinyl collection alone pretty much dominated my room. I cannot ask that of my family members. Part of this project was to win more floor space, getting rid of my semi PA speakers that took up a lot of space. Visually these towers might dominate even more, but they take very little space otherwise and still sound great. Way better than what I had before.
If I didn't have size constraints I might not even went this route. Right now I am just glad I build them. But unlike most members here I'm not thinking of the next speakers to build. Unless it is commissioned by someone else.
It isn't that easy because I'd have to supply it as AAC encoded samples or my server would fill up fast.
The 2 main problems I encounter with stuff I know well are dynamic range compression and balance.
I have songs I can still enjoy with only a DR of about 7. But if you really crank them the faults of the over use of compression become irritating. While at more sane levels they still can sound decent enough to fully enjoy them.
But I also have stumbled over samples with DR numbers of 16, but no good balance in the mix. It will sound pretty good when cranked but due to the balance it doesn't work for me at my preferred play back level.
What I'm getting at is there is a huge variety in the mastering from the same music. The hardest part is to find the good balance that works, sounding like I remember my preferred songs. I'm not claiming superiority of vinyl, just a lack of consistency in CD's.
As Mitch pointed out the effects of the Loudness Wars are very real. But sadly that doesn't mean that a recording with higher DR numbers automatically sounds better. A lot of it seems to depend on the SPL level used when creating the songs.
Even the Led Zeppelin Remasters still have faults and warts. I'm glad they didn't go overboard and polish them. But at least the mastering seems more like how I remember my old vinyl versions. There is a good sense of depth, it is pleasing to listen to and it sucks me into the song. The CD's I tried before always sounded bright (sometimes harsh) in comparison.
Kudo's to you ra7, if you're not plagued by this. But to me, remembering the music as I have played it all my life (probably a more romantic view than an accurate one) did not convey on CD with some of them. It's like someone messed up my tone controls while I wasn't looking. I do have a lot of CD's where this isn't a problem. It's mostly the old stuff I remember from my vinyl days.
I will never get rid of my vinyl collection, but with my current family life I have no physical space to implement it into our living room. When I was living on my own the CD and vinyl collection alone pretty much dominated my room. I cannot ask that of my family members
. Part of this project was to win more floor space, getting rid of my semi PA speakers that took up a lot of space. Visually these towers might dominate even more, but they take very little space otherwise and still sound great. Way better than what I had before.If I didn't have size constraints I might not even went this route. Right now I am just glad I build them. But unlike most members here I'm not thinking of the next speakers to build. Unless it is commissioned by someone else.
JRiver's effect Surround Field, Medium enhancement
As I've treated every bit of my project with care and dedication to find advancement in my playback chain there was still a loose end.
I've stated in the past that I like and use JRiver's Effect: Surround Field and set that to Medium enhancement. Even after all of the little and big changes I made, I still like to have that effect included in my DSP settings. It makes everything more engaging, more realism if you will.
But I had tested everything to advance my knowledge and results and always just accepted that little [v] check mark in Effects, without knowing what it did exactly. I even complimented the effect with Voxengo's free MSED to get it to perform to my liking.
I decided that although I think I have a pretty good handle on what it does I really should test it to be sure.
So here's what I found:
I generated a simple test file with a couple of random tones. Turned it into a "Stereo" track and cut the first two tones from the right channel and the last 3 from the left channel while keeping 3 in unaltered "stereo".
Here's how it looks:
Now if I run that file trough JRiver with the Surround Field effect set to Medium enhancement I get this (output set to diskwriter):
It performs exactly like I thought it would, it raises the SPL level on the "Side" signals, leaves the true stereo signal untouched and due to the mid/side processing it introduces a negative version of the SPL difference in the opposing channel. Acting a lot like cross talk cancelation, but without the delay.
A close up view where you see the negative version in the opposing channel:
So what is it that I do with MSED, you may wonder. I use MSED to get the volume of the mid and sides back to normal. I hand picked a level that seemed right. Boosting the mid SPL by 2.2 dB. Lets see how that looks:
Not bad, after setting it by ear! It does what I aimed to do, making the volume difference between mid and side virtually non existent (it may need a hair more), so the only real difference is the negative copy in the opposing channel, the cross talk cancelation as I like to look at it.
I do not know if this effect works with just any type of speaker. But I still like it in my DSP chain. I remember reading John Atkinson's explanation of the BSG Technologies qøl Signal Completion Stage. Even stating that vinyl records may have something similar going on due to their nature.
Anyway, I finally showed myself what it does. It will remain an acceptable part in my chain. And now I know why my early MSED experiments didn't do as well, I boosted the sides in MSED by 1.6 dB max at the time. It seems all I had to do is boost a bit more and I would have created an exact copy of that JRiver effect. Good to know.
** adding this to my original series ©: know what it is you're listening to **
As I've treated every bit of my project with care and dedication to find advancement in my playback chain there was still a loose end.
I've stated in the past that I like and use JRiver's Effect: Surround Field and set that to Medium enhancement. Even after all of the little and big changes I made, I still like to have that effect included in my DSP settings. It makes everything more engaging, more realism if you will.
But I had tested everything to advance my knowledge and results and always just accepted that little [v] check mark in Effects, without knowing what it did exactly. I even complimented the effect with Voxengo's free MSED to get it to perform to my liking.
I decided that although I think I have a pretty good handle on what it does I really should test it to be sure.
So here's what I found:
I generated a simple test file with a couple of random tones. Turned it into a "Stereo" track and cut the first two tones from the right channel and the last 3 from the left channel while keeping 3 in unaltered "stereo".
Here's how it looks:
Now if I run that file trough JRiver with the Surround Field effect set to Medium enhancement I get this (output set to diskwriter):
It performs exactly like I thought it would, it raises the SPL level on the "Side" signals, leaves the true stereo signal untouched and due to the mid/side processing it introduces a negative version of the SPL difference in the opposing channel. Acting a lot like cross talk cancelation, but without the delay.
A close up view where you see the negative version in the opposing channel:
So what is it that I do with MSED, you may wonder. I use MSED to get the volume of the mid and sides back to normal. I hand picked a level that seemed right. Boosting the mid SPL by 2.2 dB. Lets see how that looks:
Not bad, after setting it by ear! It does what I aimed to do, making the volume difference between mid and side virtually non existent (it may need a hair more), so the only real difference is the negative copy in the opposing channel, the cross talk cancelation as I like to look at it.
I do not know if this effect works with just any type of speaker. But I still like it in my DSP chain. I remember reading John Atkinson's explanation of the BSG Technologies qøl Signal Completion Stage. Even stating that vinyl records may have something similar going on due to their nature.
Anyway, I finally showed myself what it does. It will remain an acceptable part in my chain. And now I know why my early MSED experiments didn't do as well, I boosted the sides in MSED by 1.6 dB max at the time. It seems all I had to do is boost a bit more and I would have created an exact copy of that JRiver effect. Good to know.
** adding this to my original series ©: know what it is you're listening to **
As long as I'm verifying everything, just did the same test with MSED, boosting only the sides with 2.3 dB gives the same results as JRiver's Surround Field effect. But, an that's a big but, boosting the mid in the same MSED chain negates the negative copy in the opposing channel. It turns into a simple volume boost instead.
I noticed in the graph above I had "Flip Phase 180" on in MSED, not in the usual JRiver setup but in this offline config. It is also evident even boosting the "Mid" content with MSED eats away the reversed phase crosstalk signal that the JRiver effect introduced.
I experimented with setting up something similar by mixing channels in JRiver:
The resulting signal trough JRiver:
The little boost in the end of the chain is for direct comparison to the other method, making the final SPL the same.
This one obviously also alters the "Mid" signal, contrary to the above. At first glance it has similar results, but this one is completely configurable. I'll have to play a bit more to see if I like it enough. There's delay in this one, although that will only be ideal in the sweet spot. As you move to the sides the path lengths change. I didn't notice this effect though. It seems very similar to the Surround Field effect, with the exception of the phantom centre vocal. Somehow I get the idea of an improved intelligibility. Might be in my head though. Need more time to see if this one works out. I did notice a slight difference in width perception. Not sure yet on depth, too much variables to play with.
All in all fun to experiment with. It really does seem the working ingredient of the Surround Field effect was the phase inverted copy in the opposing channel. Making instruments and voices stand out more as separate identities.
I experimented with setting up something similar by mixing channels in JRiver:
The resulting signal trough JRiver:
The little boost in the end of the chain is for direct comparison to the other method, making the final SPL the same.
This one obviously also alters the "Mid" signal, contrary to the above. At first glance it has similar results, but this one is completely configurable. I'll have to play a bit more to see if I like it enough. There's delay in this one, although that will only be ideal in the sweet spot. As you move to the sides the path lengths change. I didn't notice this effect though. It seems very similar to the Surround Field effect, with the exception of the phantom centre vocal. Somehow I get the idea of an improved intelligibility. Might be in my head though
. Need more time to see if this one works out. I did notice a slight difference in width perception. Not sure yet on depth, too much variables to play with.All in all fun to experiment with. It really does seem the working ingredient of the Surround Field effect was the phase inverted copy in the opposing channel. Making instruments and voices stand out more as separate identities.
Last edited:
Try reading this paper on cross talk cancelation,
https://www.princeton.edu/3D3A/Publications/BACCHPaperV4d.pdf
I'd call that humbling, but interesting non the less .
https://www.princeton.edu/3D3A/Publications/BACCHPaperV4d.pdf
I'd call that humbling
, but interesting non the less .
IMO cross talk elimination is based on a fundamental flaw in understanding the human auditory system.
In real life, when an object generates a sound, both ears will perceive the sound. From intensity and timing differences between both ears, the auditory system is able to localize the source of the sound. With loudspeaker it is exactly the same thing. For an optimal stereo image, both ears need to perceive sound from both speakers. It is easy to understand that in the case of full signal to left, or right, the most natural way of perceiving the location of the sound is to keep it as it is. Any attempt to modify the sound so that cross talk is reduced, that is, to limit the perception of one of the speakers by one of the ears, will lead to an unnatural situation.
Of course, if one wants to emulate the sound of a pair of headphones - a stereo image like a band is playing right in the middle of your head - tricks are necessary.
In real life, when an object generates a sound, both ears will perceive the sound. From intensity and timing differences between both ears, the auditory system is able to localize the source of the sound. With loudspeaker it is exactly the same thing. For an optimal stereo image, both ears need to perceive sound from both speakers. It is easy to understand that in the case of full signal to left, or right, the most natural way of perceiving the location of the sound is to keep it as it is. Any attempt to modify the sound so that cross talk is reduced, that is, to limit the perception of one of the speakers by one of the ears, will lead to an unnatural situation.
Of course, if one wants to emulate the sound of a pair of headphones - a stereo image like a band is playing right in the middle of your head - tricks are necessary.
IMHO it is very difficult to fool the ear-brain system.
Stereo is a flawed sytstem to start with.
The microphones picks up the sound in one acoustik enviroment and the loudspeakers produce the sound in quite another acoustik.
The sound you hear in a concert hall is not only from in front of you. Back and back-side sound from the concert hall is picked up by the microphone (omni mic) and reproduced as a sound coming from the loudspeakers in front of you. In the case of cardio mic. it is attenuated and missing in the repruduction.
If a crosstalk cancellation works in fooling the ear brain well why not use it?
There is no such thing as a pure correct path from the 3D acoustik enviroment to a stereo system.
The only theoretical right (well almost) is bin aural with microphones place in a artificial head where the ears are and reproduced by a set headphones placed the excact same place in your ears.
But everyone that has heard such a system, knows it cannot fool the ear-brain. Very limited sense of depth and no fysical stimulation for starters.
Koldby
Stereo is a flawed sytstem to start with.
The microphones picks up the sound in one acoustik enviroment and the loudspeakers produce the sound in quite another acoustik.
The sound you hear in a concert hall is not only from in front of you. Back and back-side sound from the concert hall is picked up by the microphone (omni mic) and reproduced as a sound coming from the loudspeakers in front of you. In the case of cardio mic. it is attenuated and missing in the repruduction.
If a crosstalk cancellation works in fooling the ear brain well why not use it?
There is no such thing as a pure correct path from the 3D acoustik enviroment to a stereo system.
The only theoretical right (well almost) is bin aural with microphones place in a artificial head where the ears are and reproduced by a set headphones placed the excact same place in your ears.
But everyone that has heard such a system, knows it cannot fool the ear-brain. Very limited sense of depth and no fysical stimulation for starters.
Koldby
Last edited:
Your theory here would make a lot more sense if we had a separate speaker for every instrument or voice. Sadly, we don't have that (yet). And due to having two speakers that are acting like the first reflection of an acoustical event we totally missed, it is a fun, but flawed representation.
While I don't have the intention to go as far as the paper I linked above, it is a fun read and intriguing view on the principles of crosstalk.
I noticed after correcting my speakers with FIR filtering, that with the correction came the sense of a somewhat duller/less lively representation of the music I was listening to. It was as if the processing had shrunk my speakers. Up to that point the speakers were impressing me with the lively presentation. Not correct in many ways but very engaging in it's own way.
As noted much earlier in this thread, the JRiver effect: Surround Field gave me back some of that missing sense of listening to actual voices and instruments. It presented the stage with more depth and every instrument and voice had more presence of it's own as a separate identity within the stage. At that time I tried to find clues as to why I experienced it like that. I've had it on and off many times. But always came back to have at least some of it included. Together with the MSED plugin I could determine how much of it I wanted to keep. Turn on the JRiver effect and the sides would come to life. But many times overpowering the phantom centre. Boosting the phantom centre to overcome that problem works, but only up to a point.
With more experimentation it resulted in me thinking the attenuated anti phase copy of pure left channel sounds in the right channel and vice versa is the key ingredient responsible for the more lifelike experience.
Back ground vocals came to life, sound way more like actual voices instead of the thinner representation usually heard. And they image where the mix put them. Sometimes in front of the speakers and at other times only wider or more to the back. Some of that had to do with reflections and was the reason for my ambient channel project. But the other ingredient still clearly had to do with whatever processing was included in that JRiver Surround Field effect. Basically an effect that Blumlein had already described long ago to widen the stereo field. But it wasn't the widening that struck me most or why I kept using it. The width was still the same, but the definition of the side content changed (for the better I.M.H.O.). As did depth perception.
With the progression of my room and the processing I used, the difference between having the effect on or off did become smaller. But there's still a noticeable difference. That's why I still want to spend time to play with this. I do not think it is an effect that will work for every speaker every time. But with the arrays it definitely works for me.
This is again a little study/experiment to see what it can bring me. I won't place the speakers in a 20 degree angle and start running full blown crosstalk cancelation or put a matrass in front of my nose. I've played with some of that a long time ago and could not get past the side effects the processing samples I tried gave me. But with the tiny amounts the JRiver effect gave me as a side effect, it does change my perception in a good way. The manual way I showed earlier is even better in the phantom centre. Not sure yet on the sides but that could as well be because I'm not boosting the sides anymore. Even with running MSED as a means to control the effect the plot I showed the sides as a tiny bit louder.
But I'm definitely not trying to mimic headphones here. (*)
If I close my eyes I want to drown in what I hear, being caught by the music. Step by step I do get closer. I've been abducted by the music many times
.Forget about everything except the music. To be transported to the event... as best as possible anyways. The "you are there" as far as the music allows it.
I totally agree. I rather listen to my speakers than Headphones. Way more engaging and sense of space plus the physical senses that are triggered. The vibrations in the floor and couch help too as well as in your body.
(*) If I were I would have build concave arrays.
Last edited:
Mid and Side as in the Stereo signal split into all content that's the same for both channels (mid) and the content that's unique in the left and right channel (side).
A plugin like Voxengo's free MSED splits the Stereo into mid and side and lets you control their volume separately. It can also mute each of the signals. Most EQ plugins also have mid/side processing ability.
The ideas mentioned here are largely influenced by the Blumlein Shuffling. Though they do have a slight resemblance to the cross talk principles. That resemblance is based on what actually happens once you boost either mid or side signal separately.
See: http://www.audiosignal.co.uk/Resources/Stereo_shuffling_A4.pdf for a bit of background on it.
Most will use the principles to widen the Stereo sound. I am more interested in the side effect, clearer individual sounds from the separate instruments and voices.
Something that was claimed for an analog device called the BSG qøl Signal Completion Stage as well. It turns out that this device works very similar to JRiver's Surround Field processing. Basically just a boost of the "Side" signal level. The thing I've wondered about from the start was: what is more effective, the relative SPL difference between the mid and side signal or the fact that it introduces a negative phase copy of itself in the opposing channel (the crosstalk part).
For the Width part the volume difference plays a role. But for the clearer perception I think the phase inverted copy plays an important part.
I got interested in it because it worked for me to enhance the Stereo separation. But I do not want the relative SPL level of the sides to be higher than the phantom centre (mid).
Disregard this part if you will: My experiments with Ambiophonics showed me there are a lot of unpleasant side effects in most digital forms of crosstalk cancelation. The linked BACCH paper shows to my knowledge, which is limited, one of the most effective ways to implement it.
I tried some sample material a long time ago that was enhanced with that particular method and it definitely did something. I still got the files somewhere. It was taken from a radio show where Professor Edgar Choueiri explained it's basic principles and played samples processed for a standard stereo triangle setup. In those days I was still playing with Car Audio and bought a plugin from the Ambiosonics website to experiment. I'm kind of revisiting that due to the effectiveness of the more basic Blumlein Shuffling. I can't afford to play with Professor Edgar Choueiri's version.
A plugin like Voxengo's free MSED splits the Stereo into mid and side and lets you control their volume separately. It can also mute each of the signals. Most EQ plugins also have mid/side processing ability.
The ideas mentioned here are largely influenced by the Blumlein Shuffling. Though they do have a slight resemblance to the cross talk principles. That resemblance is based on what actually happens once you boost either mid or side signal separately.
See: http://www.audiosignal.co.uk/Resources/Stereo_shuffling_A4.pdf for a bit of background on it.
Most will use the principles to widen the Stereo sound. I am more interested in the side effect, clearer individual sounds from the separate instruments and voices.
Something that was claimed for an analog device called the BSG qøl Signal Completion Stage as well. It turns out that this device works very similar to JRiver's Surround Field processing. Basically just a boost of the "Side" signal level. The thing I've wondered about from the start was: what is more effective, the relative SPL difference between the mid and side signal or the fact that it introduces a negative phase copy of itself in the opposing channel (the crosstalk part).
For the Width part the volume difference plays a role. But for the clearer perception I think the phase inverted copy plays an important part.
I got interested in it because it worked for me to enhance the Stereo separation. But I do not want the relative SPL level of the sides to be higher than the phantom centre (mid).
Disregard this part if you will: My experiments with Ambiophonics showed me there are a lot of unpleasant side effects in most digital forms of crosstalk cancelation. The linked BACCH paper shows to my knowledge, which is limited, one of the most effective ways to implement it.
I tried some sample material a long time ago that was enhanced with that particular method and it definitely did something. I still got the files somewhere. It was taken from a radio show where Professor Edgar Choueiri explained it's basic principles and played samples processed for a standard stereo triangle setup. In those days I was still playing with Car Audio and bought a plugin from the Ambiosonics website to experiment. I'm kind of revisiting that due to the effectiveness of the more basic Blumlein Shuffling. I can't afford to play with Professor Edgar Choueiri's version.
Last edited:
This crosstalk cancelation is way more interesting than I expected it to be. Though my posted JRiver method isn't the most effective way to implement it. I was able to improve on it by summing Left plus Right and subtract that at -15 dB from the main signal. Doing it like that has the inverted phase side signal down by at least 3 dB compared to the mid. Even without my ambient speakers I was completely taken by surprise how this effect influenced the perception of imaging. A complete wrap around that had a strong likeness to my tests with the BACCH radio show. At that time I was still listening to my old speakers, the big semi-PA Jamo's.
If you don't correct the side cancelation signal level the sides will wrap around you at the listening position. Fun, but not the way it is supposed to be. This definitely improves phantom centre intelligibility. Now I start to wonder if you could mimic this behaviour by some select cuts and boosts in the "mid" signal only. As a loop-back shows dips at ~3.6 and ~7.2 KHz for the combined signal. Bringing in a bit of the side boost (with it's inverted phase copy in the opposite side) does give a very believable stage. Fun stuff! If you overdo it the phantom voices really shine. The actual delay is very critical though. So far 0.270 ms seems to work really well in my setup. Of coarse this is related to speaker separation and distance to the listening position as well as the size of the listener's head.
At least now I know why the crosstalk setup has the speakers about 20 degree apart. It is way more powerful than I ever imagined! Did not expect a signal at -15 dB to have this big of an impact.
Thinking it trough a bit more, the Line Array seems a perfect candidate for this type of processing. You only deal with side to side positioning as up and down keeps the important distances equal. Even listening off axis it still sounds better than without it. I use a bit of EQ to keep tonality in check. That's one of the draw backs, if any.
I still can think of a few other ways to implement this. I need a bit more side cancelation but without the side boost. But things get complicated fast!
If you don't correct the side cancelation signal level the sides will wrap around you at the listening position. Fun, but not the way it is supposed to be. This definitely improves phantom centre intelligibility. Now I start to wonder if you could mimic this behaviour by some select cuts and boosts in the "mid" signal only. As a loop-back shows dips at ~3.6 and ~7.2 KHz for the combined signal. Bringing in a bit of the side boost (with it's inverted phase copy in the opposite side) does give a very believable stage. Fun stuff! If you overdo it the phantom voices really shine. The actual delay is very critical though. So far 0.270 ms seems to work really well in my setup. Of coarse this is related to speaker separation and distance to the listening position as well as the size of the listener's head.
At least now I know why the crosstalk setup has the speakers about 20 degree apart. It is way more powerful than I ever imagined! Did not expect a signal at -15 dB to have this big of an impact.
Thinking it trough a bit more, the Line Array seems a perfect candidate for this type of processing. You only deal with side to side positioning as up and down keeps the important distances equal. Even listening off axis it still sounds better than without it. I use a bit of EQ to keep tonality in check. That's one of the draw backs, if any.
I still can think of a few other ways to implement this. I need a bit more side cancelation but without the side boost. But things get complicated fast!
Last edited:
I have to add... SPL level of playback is of HUGE importance as well.
A few dB can make a world of difference how the perception of this effect works. For the first time I've had singers right in my face with the convolution corrected playback. Controllable with SPL level.
I may need to point out the absence of early reflections is of utter most importance as well, I don't expect any of it to work on a less controlled frequency and phase setup.
A few dB can make a world of difference how the perception of this effect works. For the first time I've had singers right in my face with the convolution corrected playback. Controllable with SPL level.
I may need to point out the absence of early reflections is of utter most importance as well, I don't expect any of it to work on a less controlled frequency and phase setup.
Last edited:
Thanks for the response and the links. I tried the JRiver medium enhancement and it is a nice effect. It brought the sound out from between the speakers to outside the speakers. But it leaves a hole to the left and right of the center phantom image. Sort of like someone took the image and moved nearer to the speakers and away from the center. I don't know if I like it or not, but definitely interesting.