BACK in the DAY of LP, it was *necessary* to sum the LF or else the cutter could not do the cut in the master. That was the only reason that bass was summed for mono. For CD/digital and tape for that matter summing the bass is/was not required, but may be a hold over. Also it may have benefits in compression algorithms and for FM broadcasting (not sure here).
The issue of having "enough" LF energy on classical recordings is highly dependent upon the piece and the way it is recorded imo.
The issue of having "enough" LF energy on classical recordings is highly dependent upon the piece and the way it is recorded imo.
This is the partial analysis of two recordings from the late 90s.
The horizontal axis is time, vertical is frequency. The depth of the colour indicates the level. Top is left channel, lower is right.
The first recording is Sarah Maclachlan's 'Angel', the studio version. The passage displayed is the two opening piano notes, with backing bass. I am able to zoom right up to see LF content (further zoom is also possible) and there are clear differences between left and right.
The second file is a tom roll from James Taylor's 'Gaia' from the 1997 album 'Hourglass'. Unfortunately there is very heavy LF content on the roll and I am unable to zoom in further as it becomes one red mess. This is a topographical view, which shows clear, if small, differences between LF between the two channels. Part of the differences will be due to arrival times, but the drum kit is clearly recorded in stereo and the bass is part of stereo field and whatever post production was done, mono summing of LF was quite clearly not.
My apologies for the image size. These are captured off a 2560x1600 monitor.
The horizontal axis is time, vertical is frequency. The depth of the colour indicates the level. Top is left channel, lower is right.
The first recording is Sarah Maclachlan's 'Angel', the studio version. The passage displayed is the two opening piano notes, with backing bass. I am able to zoom right up to see LF content (further zoom is also possible) and there are clear differences between left and right.
An externally hosted image should be here but it was not working when we last tested it.
The second file is a tom roll from James Taylor's 'Gaia' from the 1997 album 'Hourglass'. Unfortunately there is very heavy LF content on the roll and I am unable to zoom in further as it becomes one red mess. This is a topographical view, which shows clear, if small, differences between LF between the two channels. Part of the differences will be due to arrival times, but the drum kit is clearly recorded in stereo and the bass is part of stereo field and whatever post production was done, mono summing of LF was quite clearly not.
An externally hosted image should be here but it was not working when we last tested it.
My apologies for the image size. These are captured off a 2560x1600 monitor.
Last edited:
There are two issues at play here, IMHO:
1. Do recordings have stereo bass?
2. Can people hear bass in stereo or is localisation dependent on HF components?
The recordings are from 1997, both are analog recordings (though I used sections from the CD print). That was 16 years ago, and it makes me feel old because I have heard very little of more recent music.
I'm guessing there are enough examples of both stereo and mono bass in recordings, engineers will sum to mono because of various reasons, or not because of various reasons. Vinyl is a big decider, quite obviously, but there are other use cases such as FM radio which is often heard in mono.
As to audibility of stereo bass, that's the reason this is in 'Lounge'?
1. Do recordings have stereo bass?
2. Can people hear bass in stereo or is localisation dependent on HF components?
The recordings are from 1997, both are analog recordings (though I used sections from the CD print). That was 16 years ago, and it makes me feel old because I have heard very little of more recent music.
I'm guessing there are enough examples of both stereo and mono bass in recordings, engineers will sum to mono because of various reasons, or not because of various reasons. Vinyl is a big decider, quite obviously, but there are other use cases such as FM radio which is often heard in mono.
As to audibility of stereo bass, that's the reason this is in 'Lounge'?
I wish for longer edit times 
As a totally random selection, I picked a recording from the 70s - this is Peter Green 'Slabo Day'. In my earlier graphs I posted the timescale as bars - my bad. This should be easier to read - each X-axis marker is 50ms.
The bass is different between the two channels in this one as well. Differences are much, much smaller, which would happen if you panned the recorded bass to center. It would still be a little different due to pan laws.
If you suggest a song and I have it, I will be happy to analyse it as per your specification (timescales, frequency step etc) for the purpose of my continuing education on this topic.
As a totally random selection, I picked a recording from the 70s - this is Peter Green 'Slabo Day'. In my earlier graphs I posted the timescale as bars - my bad. This should be easier to read - each X-axis marker is 50ms.
An externally hosted image should be here but it was not working when we last tested it.
The bass is different between the two channels in this one as well. Differences are much, much smaller, which would happen if you panned the recorded bass to center. It would still be a little different due to pan laws.
If you suggest a song and I have it, I will be happy to analyse it as per your specification (timescales, frequency step etc) for the purpose of my continuing education on this topic.
1 dB of separation?
How about it?
If you take an instrument that has been recorded mono and pan-pot it 1dB, will the apparent position move?
Wait, let's take two instruments, like two violins or maybe a viola playing high and a violin playing low (overlapping on the musical staff). How much pan-potted difference will make it audible that there are two instruments, separated by apparent left-right space in the mix?
How many dB will make the instrument appear to be virtually hard left or right?
Calculation of the direction of phantom sources by interchannel level difference and time difference stereophony stereo time of arrival - localization curves - sengpielaudio Sengpiel Berlin
Thanks Earl. I had to do the best I could with the tools I have. Would love to do some better cross-correlation and have some math whiz friends helping me with it.
Makes sense, but I was surprised at how much stereo separation the classical recordings have across the audible spectrum. I have not tested Jazz, it could be all over the place.
I would say No. Tho I can hear the shift at 1dB, it's hardly worth the trouble for stereo subs. Getting beyond that, it starts to be worth it. A sliding scale.
The point of this sidetrack was to find out if most recordings have mono summed bass. If they do, then there would be no good reason to prefer stereo subs. If we have established that many recordings do have stereo bass, then we can get back to the question of whether of not we can hear it. I've been able to in blind tests, but others I've tested weren't very good at it. Perhaps it varies a lot from person to person.
...and how to reproduce it. In acoustically small rooms the direct signal is completely swamped by the room. Directional information at low frequencies is heavily distorted.
...or how to keep what is being reproduced.
From this we can surmise that the problems exist in the rooms we are listening in and should be addressed as such.
Maybe we should define the frequency range we're talking about. What kind of test signals are you using? What type of room and speaker?
There's an interesting paper by Martens et al (Audio Engineering Society Convention Paper 6229):
"ABSTRACT
This investigation addressed two primary questions relating to the use of subwoofers in reverberant reproduction
environments, the first being whether listeners are able to discriminate between the auditory images
resulting from correlated and decorrelated low-frequency signals, and the second being whether decorrelation
between drivers produced identifiably greater listener envelopment. For the experiments reported in this paper,
octave-band noise samples with center frequencies ranging in third-octave steps from 31.5Hz to 125Hz
were presented via a multichannel loudspeaker array. These low-frequency signals could be either perfectly
correlated (drivers receiving identical signals) or maximally decorrelated between selected pairs of five loudspeakers
positioned according to the ITU standard configuration. Even in a small and highly reverberant
listening room, decorrelated signals with center frequencies greater than 50Hz were both discriminably and
identifiably different from correlated signals, but only when such low-frequency signals were reproduced via
the left and right surround channel drivers."
Toole comments in "Sound reproduction":
"Another recent investigation concludes that the audible effects benefi ting
from channel separation relate to frequencies above about 80 Hz (Martens et
al., 2004). In their conclusion, the authors identify a “cutoff-frequency boundary
between 50 Hz and 63 Hz,” these being the center frequencies of the octave
bands of noise used as signals. However, when the upper-frequency limits of the
bands are taken into account, the numbers change to about 71 Hz and 89 Hz,
the average of which is 80 Hz. This means, in essence, that it is a “stereo upperbass”
issue, and the surround channels (which typically operate down to 80 Hz)
are already “stereo” and placed at the sides for maximum benefi t. Enough
said."
I use sharply low passed pink noise. With pure tones I have a very hard time, same as with midrange and treble. I've done tests in my room with my modified Altec 838 bass cabinets and 416 woofers. Have also done tests in the warehouse at work with Meyer Sound subs. I do remember at point at which it became difficult to tell, but I don't remember how low that was. Think I posted it a few years back.
Once the wavelength is greater then your rooms longest dimension, then it wouldn't matter.
Given the statement; "Even in a small and highly reverberant listening room, decorrelated signals with center frequencies greater than 50Hz were both discriminably and identifiably different from correlated signals" they were in a room which has ~10' as one of it's dimensions.
The wavelength of 50 hertz is ~20' the asumption being that a half wavelength would be enough to distinguish it.
Given the statement; "Even in a small and highly reverberant listening room, decorrelated signals with center frequencies greater than 50Hz were both discriminably and identifiably different from correlated signals" they were in a room which has ~10' as one of it's dimensions.
The wavelength of 50 hertz is ~20' the asumption being that a half wavelength would be enough to distinguish it.
I thought we were talking about "normal" listening rooms not warehouses?
Could you upload those test signals? How deep does your "modified Altec 838 bass cabinets and 416 woofers" go? Did you also do in-room measurements along those listening tests?
I don't think the warehouse matters, we were within 3 meters of the speakers. My listening room is typical - to small.
The -3dB of the Altec cab is somewhere around 35Hz. I think the same for the Meyer subs.
I'll look for the signals, not sure I saved them. But I could make more. Let me know what cut-of frequencies you'd like.
The -3dB of the Altec cab is somewhere around 35Hz. I think the same for the Meyer subs.
I'll look for the signals, not sure I saved them. But I could make more. Let me know what cut-of frequencies you'd like.
At this point I think we would need to do free field listening tests to determine the lower cutoff for low frequency localization.
Then we would need to perform listening tests in rooms of different sizes in order to understand how room interaction influence localization and/or perception of envelopment.
Then we would need to perform listening tests in rooms of different sizes in order to understand how room interaction influence localization and/or perception of envelopment.
Sound Forge 10 Visual Phase Analysis
I checked a 1968 album called Super Session ripped using the original LP with a Shure V15 Type II cartridge and M-Audio Audiophile Firewire audio interface to Sound Forge 10. A Mastering Effects filter was applied with heavy roll-off above 200Hz. There's lots of out-of-phase bass present as displayed in the SF 10 Phase Meter Lissajous pattern. I also converted the two filtered stereo tracks to mono using "difference mode," which should have cancelled any in-phase bass and it still had plenty of amplitude.
Here's a screenshot
https://www.hightail.com/download/elNJb242Zy9LVlZqQThUQw
Years ago when Quadraphonic sound was introduced many people (self inculded) connected two rear mounted speakers together with the rear - (negative) speaker inputs connected together and the + (positive) to the right and left channel outputs. If the bass was truly monophonic then it should have been cancelled in the rear speakers, but that was not the case with many recordings. The solution was to use small speakers that had poor bass response, which allowed hearing the out-of-phase material as an added "ambienence."
I checked a 1968 album called Super Session ripped using the original LP with a Shure V15 Type II cartridge and M-Audio Audiophile Firewire audio interface to Sound Forge 10. A Mastering Effects filter was applied with heavy roll-off above 200Hz. There's lots of out-of-phase bass present as displayed in the SF 10 Phase Meter Lissajous pattern. I also converted the two filtered stereo tracks to mono using "difference mode," which should have cancelled any in-phase bass and it still had plenty of amplitude.
Here's a screenshot
https://www.hightail.com/download/elNJb242Zy9LVlZqQThUQw
Years ago when Quadraphonic sound was introduced many people (self inculded) connected two rear mounted speakers together with the rear - (negative) speaker inputs connected together and the + (positive) to the right and left channel outputs. If the bass was truly monophonic then it should have been cancelled in the rear speakers, but that was not the case with many recordings. The solution was to use small speakers that had poor bass response, which allowed hearing the out-of-phase material as an added "ambienence."
Last edited:
I think that room size is the dominating factor. The larger the room the less modes are filtered out, D/R, reverberation, etc.
40Hz and up in 20Hz steps?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.