The title being an homage to Star Trek TOS.
Conventional audio wisdom says that in most stereo recordings, the low frequencies are mixed to mono. Usually this is stated as anything under 100Hz has been mixed down to mono, therefor there is no need for stereophonic subwoofers. But is this true? I have found that it isn't, and in this thread I will attempt to show that many recordings, be they CD, MP3, or LP vinyl - do have significant stereo under 100Hz.
For a bit of history on this discussion, you can see these two threads:
https://www.diyaudio.com/community/...-ear-really-localize-bass.247583/post-3748187
https://www.diyaudio.com/community/...-subwoofer-system-provide.411881/post-7670252
Using common audio software and common recordings, I'm going to show the stereo content - or lack there of - in many recordings across several genres.
The software used will be Audacity with an M/S plugin, Goldwave for levels and REW for frequency plots. Other software should give very similar results.
Conventional audio wisdom says that in most stereo recordings, the low frequencies are mixed to mono. Usually this is stated as anything under 100Hz has been mixed down to mono, therefor there is no need for stereophonic subwoofers. But is this true? I have found that it isn't, and in this thread I will attempt to show that many recordings, be they CD, MP3, or LP vinyl - do have significant stereo under 100Hz.
For a bit of history on this discussion, you can see these two threads:
https://www.diyaudio.com/community/...-ear-really-localize-bass.247583/post-3748187
https://www.diyaudio.com/community/...-subwoofer-system-provide.411881/post-7670252
Using common audio software and common recordings, I'm going to show the stereo content - or lack there of - in many recordings across several genres.
The software used will be Audacity with an M/S plugin, Goldwave for levels and REW for frequency plots. Other software should give very similar results.
Here is an analysis of the Bee Gees song "How Can you Mend a Broken Heart?"
The file is converted to Mid-Sid format, then run through frequency analysis to plot the FR of the mid and the side channels. For those of you not familiar with Mid-Side encoding, it's a format where in channel 1 you have the mono information, that is the information that is common to both left and right in a stereo recording. In channel 2 is the Side info, which is all that is NOT common between left and right. There is much more info available on the web about this format if you care to search.
E.G. https://en.wikipedia.org/wiki/Joint_encoding
The basic idea is to show the mono part of the signal (mid) vs the stereo part of the signal (sides) plotted over frequency.
Below is a frequency response plot of "How Can you Mend a Broken Heart?" showing the levels of mono (mid) vs stereo (sides). You will see that in almost all recordings the stereo portion of the recording is lower in volume than the mono portion. At least for most of the spectrum. This is shown below.
Below is a normalized view of the same file. On this plot the mid and the side channels have been matched in RMS value. For me this makes it easier to see the relative levels of mono vs stereo. We can see that stereo seperation does drop about 4dB under 100Hz, but that seperation rises above 5kHz. A good argument for not using mono tweeters!
How would you prefer to see the mono vs stereo difference plotted? Actual levels as in the first plot, or the two signals normalized as in the second plot?
The file is converted to Mid-Sid format, then run through frequency analysis to plot the FR of the mid and the side channels. For those of you not familiar with Mid-Side encoding, it's a format where in channel 1 you have the mono information, that is the information that is common to both left and right in a stereo recording. In channel 2 is the Side info, which is all that is NOT common between left and right. There is much more info available on the web about this format if you care to search.
E.G. https://en.wikipedia.org/wiki/Joint_encoding
The basic idea is to show the mono part of the signal (mid) vs the stereo part of the signal (sides) plotted over frequency.
Below is a frequency response plot of "How Can you Mend a Broken Heart?" showing the levels of mono (mid) vs stereo (sides). You will see that in almost all recordings the stereo portion of the recording is lower in volume than the mono portion. At least for most of the spectrum. This is shown below.
Below is a normalized view of the same file. On this plot the mid and the side channels have been matched in RMS value. For me this makes it easier to see the relative levels of mono vs stereo. We can see that stereo seperation does drop about 4dB under 100Hz, but that seperation rises above 5kHz. A good argument for not using mono tweeters!
How would you prefer to see the mono vs stereo difference plotted? Actual levels as in the first plot, or the two signals normalized as in the second plot?
Last edited:
The phase and amplitude of bass frequencies usualy are less importance than in higher part of the spectrum. Although the power involved is greater in lows than in highs, bass frequencies are more similar to plane wave while higher are more spherical, if my memory doesn't fail to me.
Usually, LPs are recorded with mono bass to minimize the vertical stylus movement, so the stylus only moves horizontally for bass content.
Not necessary for any other sources. From various circuit experimentation, I've found that bass phase response is quite audible, at least to me.
Not necessary for any other sources. From various circuit experimentation, I've found that bass phase response is quite audible, at least to me.
I played a bit with Light My fire, as mentioned in the other thread.
I made two files, with an FFT high cut filter.
The first one starts to cut at 75Hz and steadily rolls off to 150Hz
The second one is a sharp cut at 250Hz.
I included two minutes audio of both files and screenshots of the waveform and FFT analysis.
You will hear a nice distorted left channel in the 250Hz (bass12m.zip) file
The bass guitar is on both channels, the bassdrum and snare on the left.
This is the CD version, not the LP.
Hugo
I made two files, with an FFT high cut filter.
The first one starts to cut at 75Hz and steadily rolls off to 150Hz
The second one is a sharp cut at 250Hz.
I included two minutes audio of both files and screenshots of the waveform and FFT analysis.
You will hear a nice distorted left channel in the 250Hz (bass12m.zip) file
The bass guitar is on both channels, the bassdrum and snare on the left.
This is the CD version, not the LP.
Hugo
It's tempting to suggest plotting a difference curve, even though perhaps we can't be sure the content of the mid/sides naturally has the same response balance before considering the separation factor, I don't think that changes.
I've been looking at a lot of files and it seems best just to plot the mid vs side at their normal levels. The reason is that it shows how much, or how little stereo effect there is in the recording. Let's look at two examples from Pink Floyd. Below is "Us and Them" from DSOTM. You can see that the stereo effect is very strong across the spectrum.
Next we see "Wish You Were Here" which has less stereo effect and not much stereo in under 100Hz. Definitely a near mono bass there.
Next we see "Wish You Were Here" which has less stereo effect and not much stereo in under 100Hz. Definitely a near mono bass there.
I don’t have a lot of vinyl rips to test this, but have found that some mix the bass to mono, others don’t. I haven’t yet been able to predict which do or don’t.
That’s a current trend, but for me the LF stereo effect is more interesting.
What are we looking at here? Looking at the waveform to see any differences in the LF?
Have you tried null testing?
I'm using a MiniDSP HTx as my active crossover, so I selected left and right channels on every output, and inverted the right input. Now when I play a song, I only hear what is different between the two channels.
There's basically no low frequencies coming through, which indicates there isn't low bass in one channel that isn't in the other channel. I suppose subtle differences between left and right would sound like noise?
I tried your example, "How Can You Mend a Broken Heart". With null test engaged, there is very little that comes through. Some chorus parts, and a hint of the lead singer. It is a strange experience. It's like most of the song has been magically deleted, but bits and pieces remain and sound perfectly normal! And no, no low frequency content is passing through.
I'm using a MiniDSP HTx as my active crossover, so I selected left and right channels on every output, and inverted the right input. Now when I play a song, I only hear what is different between the two channels.
There's basically no low frequencies coming through, which indicates there isn't low bass in one channel that isn't in the other channel. I suppose subtle differences between left and right would sound like noise?
I tried your example, "How Can You Mend a Broken Heart". With null test engaged, there is very little that comes through. Some chorus parts, and a hint of the lead singer. It is a strange experience. It's like most of the song has been magically deleted, but bits and pieces remain and sound perfectly normal! And no, no low frequency content is passing through.
Analyze "The Kraken" by Hans Zimmer. (I'm playing it from Spotify)
I'm hearing what seems like the majority of the song even with the null test. It sounds much more impressive in normal stereo, though. I'm not sure how they managed to escape the null test? But this might be better case for stereo subs. The context where I first heard about stereo subwoofers was reproducing pipe organ music.
I hear a little sub frequencies, but mostly it is above 100hz according to the RTA.
I'm hearing what seems like the majority of the song even with the null test. It sounds much more impressive in normal stereo, though. I'm not sure how they managed to escape the null test? But this might be better case for stereo subs. The context where I first heard about stereo subwoofers was reproducing pipe organ music.
I hear a little sub frequencies, but mostly it is above 100hz according to the RTA.
One last one, "Three Basses" by The Fearless Flyers (Spotify). It is a trio of bass guitars, so it would be a prime opportunity to pan low frequencies. It is a modern recording, so no technology limitation. But no low frequencies come through, so it appears it was a mixing/ mastering decision.
I'm calling it "inconclusive".
I'm calling it "inconclusive".
It's only the bass from both channels of the song. First only what normally goes to the subwoofer, second the full bass frequency bands up to 250Hz.
Kick and snare are panned to the left, bass guitar (or synth) a bit more to the right. It only confirms your findings about that particular song that the bass is not centered.
Yes, that’s how I started out doing the tests. Low pass at 100Hz, then do the null test. It makes it obvious to hear what’s going on.
From there I moved to the Mid-Side technique because it provides a track that is exclusively the stereo difference. With that we can easily see what is not common to left and right