[71 dB]

Does it split vocals off to the center? That is one thing I love about the Yamaha.

Don't know what you mean by splitting vocals off to the center, but whatever is mixed center (mono) stays centered, because left & right speakers play the same content as center speaker, only at lower level.

 

[bigshot]

The Yamaha takes centered content out of the mains and moves it to the center channel.

 

[71 dB]

The Yamaha takes centered content out of the mains and moves it to the center channel.

It would be strange if it didn't. When you spread stereo content into more speakers including the center speaker you have to get the "stuff" somewhere.

 

[bigshot]

I wasn't clear... It removes the centered content from the left and right and plays it discretely only out of the center speaker. It essentially creates three discrete channels up front.

The rears have more complicated processing going on. Not just time modifications, but pulling certain frequency bands and moving them to rear left or rear right. I haven't figured out how it is doing that yet, but I think it has something to do with moving frequencies that are highly directional and only on the one channel of the stereo spread to both the front and rear in that channel. It pulls that sound into the room. Sometimes it seems to be more forward, and other times more towards the rear. Not sure what it is doing, but it sounds great.

 

[71 dB]

I tested my "less mono" plugin on Mozart's Violin Sonata in F major, K. 376 ripped from Naxos 8.110988. This is from 1938. Yehudi Menuhin Violin, Hephzibah Menuhin, Piano.

This recording has a hefty amount of background noise. The plugin makes it more noticeable, spatial. I ended up reducing noise by 4 dB before the plugin and this resulted perceivable noise level similar to the original mono track without messing too much with the music. Also, I noticed, that this track has some weird stuff going on in the infrasonic range which caused minor clipping when processed with my plugin. So, I filtered out frequencies below 10 Hz and the clipping disappeared. I will add this filtering to my plugin.

After a little bit of noise reduction and filtering out infrasonic crap my plugin produced a very nice result. Far from high fidelity, but something that is much more pleasant and natural with headphones. The sound is not a point in the middle of my head. It is a "cloud" and my head is inside the "cloud". I even perceive it as a miniature soundstage! I compared it to the original track soloing the tracks in Audacity and the change is very similar to when stereo track is forced to mono. I am happy about the result. I am pleased with myself. Years of doing "crossfeed stuff" has given me such a good understanding of headphone spatiality that I was able to accomplish this within a few days.

I exported the resulting "less mono" version to a VBR 145-185 kbps mp3 file. That should be high enough bitrate for historical recordings...

 

[bigshot]

One thing I've noticed is a lot of transfer engineers leave the transfer in stereo when they master it to CD. This ends up making the noise stereo and the signal mono. If you flatten the whole track to mono, it greatly attenuates the noise. You might try making your DSP flatten to true mono before creating the stereo effect.

You'll want to listen to a bunch of different historical recordings before committing to the levels. It can vary widely from recording to recording. The kinds of noise can vary a lot too... from rumble to swishing to flurries of tiny impulse clicks. Different kinds of noise reduction are necessary with different kinds of noise.

As for data rate in lossy, historical recordings can actually be harder to compress than full response tracks. The surface crackle can force the data rate up, and if you compress too far, the clicks and crackles start turning into little bumps and thumps.

 

[71 dB]

I wasn't clear... It removes the centered content from the left and right and plays it discretely only out of the center speaker. It essentially creates three discrete channels up front.

The rears have more complicated processing going on. Not just time modifications, but pulling certain frequency bands and moving them to rear left or rear right. I haven't figured out how it is doing that yet, but I think it has something to do with moving frequencies that are highly directional and only on the one channel of the stereo spread to both the front and rear in that channel. It pulls that sound into the room. Sometimes it seems to be more forward, and other times more towards the rear. Not sure what it is doing, but it sounds great.

In stereo you have:

L (channel) => Left speaker
R (channel) => Right speaker

When you add center speaker, you can do something like this:

L - 0.7*R => Left speaker
0.7*L + 0.7*R => Center speaker
R - 0.7*L => Right speaker

When you add up all that 0.7*X stuff nothing has changed signal-wise, but now you have a center speaker playing at level -3 dB. You also have left and right speakers playing more "wide" stuff and less of the mono content.

This is nothing revolutionary. It is how matrix stereo works and it has been done in various forms for decades. However, it is important to understand those "new" channels are not "discrete" because they depend mathematically on each other. Multichannel sound with non-discrete channels can sound really nice, but for real discrete multichannel stuff the channels need to be separate from recording to speakers. So, a 7.1 soundtrack from a Blu-ray is (most probably) discrete channel stuff, but when you use your Yamaha DSP to make a stereo track into 7.1, it isn't discrete channel. It is typical, that rear channels are made of higher frequency content.

 

[71 dB]

One thing I've noticed is a lot of transfer engineers leave the transfer in stereo when they master it to CD. This ends up making the noise stereo and the signal mono. If you flatten the whole track to mono, it greatly attenuates the noise. You might try making your DSP flatten to true mono before creating the stereo effect.

This Mozart track I was working on is completely mono. When I do the null test on left and right channel I get zero signal. My plugin takes the left channel as the "mid" channel, copies it to "side" channel and starts to mold it so that it becomes something reminiscent of what the "side" channel of a stereo track would be.

You are correct about the fact that if the music is mono and the noise is stereo, flattening to mono attenuates the noise. A few decibels mathematically, but even more perceptually.

You'll want to listen to a bunch of different historical recordings before committing to the levels.

What levels are you referring to ? Noise reduction levels? Noise reduction is not part of the plugin and is done before the plugin and of course the level of noise reduction depents on the recording itself. Better quality recordings are best left as they are, but with this Mozart track the noise level is somewhat high to begin with and my plugin makes it more perceptual so that I felt some noise reduction is beneficial despite it messing up a little bit with the music signal.

It can vary widely from recording to recording. The kinds of noise can vary a lot too... from rumble to swishing to flurries of tiny impulse clicks. Different kinds of noise reduction are necessary with different kinds of noise.

Obviously. That's why the noise reduction must be done manually beforehand if any is needed.

As for data rate in lossy, historical recordings can actually be harder to compress than full response tracks. The surface crackle can force the data rate up, and if you compress too far, the clicks and crackles start turning into little bumps and thumps.

You are correct about noise being difficult for lossy codecs, but the VBR 145-185 kbps setting seems to be good enough. It sounds fine to my ears and null test didn't expose anything alarming.

 

[bigshot]

The center channel isn’t a sum of left and right, it’s just the info that is common to both, and that common info is subtracted from left and right. Is that clearer?

For some reason CDs of historical material aren’t always flattened to mono. If you want a DSP that works with all historical releases, flatten to mono first.

The Yamaha DSPs reduce noise, they don’t increase it. Your reverb and reflection should be taming harshness in the upper miss.

 

[71 dB]

The center channel isn’t a sum of left and right, it’s just the info that is common to both, and that common info is subtracted from left and right. Is that clearer?

The way to get "what is common to both channels" without ugly granulation is to calculate their sum and scale the result, typically dividing by 2. If left channel has a value of 0.5 (in the range of -1 to 1) and right channel has a value of 0.3, what is common to both channels is (0.5+0.3)/2 = 0.4. If you limit the common part to 0.3, the channels granulate (distort) each other depending on which one has a smaller absolute value. Calculating the arithmetic mean is a linear operation and doesn't create distortion.

For some reason CDs of historical material aren’t always flattened to mono. If you want a DSP that works with all historical releases, flatten to mono first.

Yes, that is very simple to implement.

The Yamaha DSPs reduce noise, they don’t increase it. Your reverb and reflection should be taming harshness in the upper miss.

Noise is always much more noticeable on headphones than speakers.

 

[71 dB]

I tried my plugin on Albert Einstein's speech (very dry recording) and it didn't work, because it sounded like Einstein was speaking in a large hall. So, there has to be adjustment for original material type (speech, music, etc.) After implementing this, the plugin works well for dry speech keeping it dry and easy to undertand, but adding the spatial feel.

 

[bigshot]

Yeah, hall ambience needs to be adjustable. My AVR has three different halls. Berlin is the driest. Vienna is the most echoey. I find that I use different ones for different things. Some recordings are dry and can use more, and others already have hall ambience and need less. If you can make your DSP user adjustable, that would be a good variable to have a setting for.

 

[71 dB]

Yeah, hall ambience needs to be adjustable. My AVR has three different halls. Berlin is the driest. Vienna is the most echoey. I find that I use different ones for different things. Some recordings are dry and can use more, and others already have hall ambience and need less. If you can make your DSP user adjustable, that would be a good variable to have a setting for.

I made 5 processing modes to choose from based on the original material. They go from totally dry (Anechoic chamber) to echoey (Orchestral music). Most of the reverberation come from the original mono track. This mode selector just adjusts the processing to fit to the original signal type. So, a mono recording made in a very reverberant hall processed using mode "Anechoic chamber" sounds reverberant, but "stupid", because mode "Orchestral music" is the best fit. A very dry recording (such as the Albert Einstein clip) sounds good prosessed with modes "Anechoic chamber" or "Small room". It is possible church music benefits from another mode with even more reverberation than "Orchestral music", but that is easily added if needed.

 

[bigshot]

I figured you might like to try your DSP on a variety of different historical recordings. Here are some links to some of my transfers for you.

https://animationresources.org/xfers/2-01handel_concertogrosso05.mp3
https://animationresources.org/xfers/gluckorfeoedeuridice.mp3
https://animationresources.org/xfers/mendelssohnreformation1948.mp3
https://animationresources.org/xfers/schnabeldiabellis.mp3
https://animationresources.org/xfers/soundingsilence04.mp3
https://animationresources.org/xfers/walkureact1walter1935.mp3
https://animationresources.org/xfers/mahler9walter1938.mp3

If you'd like more or want to try it with other types of music, let me know.

 

[71 dB]

I figured you might like to try your DSP on a variety of different historical recordings. Here are some links to some of my transfers for you.

https://animationresources.org/xfers/2-01handel_concertogrosso05.mp3
https://animationresources.org/xfers/gluckorfeoedeuridice.mp3
https://animationresources.org/xfers/mendelssohnreformation1948.mp3
https://animationresources.org/xfers/schnabeldiabellis.mp3
https://animationresources.org/xfers/soundingsilence04.mp3
https://animationresources.org/xfers/walkureact1walter1935.mp3
https://animationresources.org/xfers/mahler9walter1938.mp3

If you'd like more or want to try it with other types of music, let me know.

Thanks! That's thoughtful of you! I'm testing out my own historical recordings, but more material doesn't hurt.