Stereo in Live Show Recordings

[E A Poe]

Should all tracks be in stereo, on the shows that have soundboards? Does stereo enhance it, or does it kill it for being forced?

[SteveZ98]

The soundboards, at least the front of house ones from '75 and '77 that Empress Valley typically releases, are stereo recordings and capture the way the instruments were presented to the audience. They seem like mono recordings most of the time because all of the instruments were pumped through the PA system together so the sound was equally distributed throughout the arena. The only time you hear that they are stereo (different sounds in one channel versus another) is during things like the violin bow section of Dazed when Jimmy intentionally has the sound bounce from side to side. I'm not aware of any way to add a stereo effect that puts Jimmy in one channel by himself during the "mono" parts. The problem with doing that is his guitar and Robert's voice overlap in the frequencies they cover, so they can't really be separated when the two of them are playing at the same time. There are fake stereo effects that can be added to a recording, but they won't be able to put Jimmy in one channel by himself. At best, they'll add some ambience to the recording, but there are other ways to do that that are more convincing.

 

 

 

 

[gibsonfan159]

3 hours ago, SteveZ98 said:

There are fake stereo effects that can be added to a recording, but they won't be able to put Jimmy in one channel by himself. At best, they'll add some ambience to the recording, but there are other ways to do that that are more convincing.

I've been thinking about trying this. Taking a soundboard and creating four different frequency settings that isolate each member as much as possible (like you said, there will still be overlapping) and remixing them separately. At the very least this would allow slightly more control over the instrument balance compared to just adjusting the EQ.

[Night_flight3030]

3 hours ago, gibsonfan159 said:

I've been thinking about trying this. Taking a soundboard and creating four different frequency settings that isolate each member as much as possible (like you said, there will still be overlapping) and remixing them separately. At the very least this would allow slightly more control over the instrument balance compared to just adjusting the EQ.

Your stereo must be old or you have a mixing board  frequency modulator because nowadays stereos already have built-in equalizers for instance my Denon has many sound options I don’t like I prefer the equalizer with the little knobs I could adjust but now they stereos don’t give you the option.

[gibsonfan159]

3 hours ago, Night_flight3030 said:

Your stereo must be old or you have a mixing board  frequency modulator because nowadays stereos already have built-in equalizers for instance my Denon has many sound options I don’t like I prefer the equalizer with the little knobs I could adjust but now they stereos don’t give you the option.

Different kind of "stereo", bud.

[Night_flight3030]

1 hour ago, gibsonfan159 said:

Different kind of "stereo", bud.

Care to explain? I have a mixing board.

[gibsonfan159]

32 minutes ago, Night_flight3030 said:

Care to explain? I have a mixing board.

Stereo as in separating the instruments in the balance. I was talking about creating a "false" multitrack of sorts by isolating frequencies.

[rm2551]

How far along is the tech? Sooner or later the meta data of the sound / frequencies should allow for a reconstruction, or synthetic recreation of a soundboard that is far superior than the original. i imagine voice would present the greatest challenge. But if you had billions, and access to the best tech and people as well as details of all of the original equipment used, soundstage, amps, the venue, instrumentation, and sampled each note and beat, individually recreated with applied atmosphere (what it should have sounded like), then applied powerful software to tracks that generated output that is a clean reconstruction (as much as possible) - you could potentially create authentic sounding (but arguably fake to a large extent) copy that could be many times cleaner than whatever source is used.

But again, how far along is the tech? Sooner or later, "deep fake" kind of recreations of SB's will be a reality. There is no doubt. Someone will write the code. Some studio will apply this initially to historically significant stuff to recreate scratchy recordings of important historical moments. Then it will just be out there to be used.

One day, Mike Millards recordings will be full multi-track sounding shows as good as you can imagine - or very, very close. One day.

Back to reality - Go for it Gibsonfan159! Good idea.

[Xolo1974]

They’ve already started doing this Rod...at Abbey Road studios. Check out their website 

[SteveZ98]

1 hour ago, Xolo1974 said:

They’ve already started doing this Rod...at Abbey Road studios. Check out their website 

I did, thanks for the heads up. Interesting work. Here's the link: https://www.abbeyroad.com/de-mix. And here's a video they put together of some work they did on early Stones recordings. At the 4:45 mark there's a demo of them using their tech on a mono version of Satisfaction. It definitely updates the sound, but it's not the full left\right stereo effect I think we're hoping to be able to do to Zep soundboards. However, there's no doubt this kind of processing will improve significantly over time. 

 

Edited July 7, 2019 by SteveZ98

[godzilla777]

There is a program that can do this. It is Izotope RX 7. It has this feature called music rebalance. That can separate the frequencies.

[SteveZ98]

3 hours ago, godzilla777 said:

There is a program that can do this. It is Izotope RX 7. It has this feature called music rebalance. That can separate the frequencies.

That's an interesting program. Here's a demo they put together. I'm not sure how well it will work on mono recordings, but they have a ten day working trial version so it's possible to try it out before spending the US$400 they're asking for the full version. 

 

[SteveZ98]

I downloaded RX 7 and have been trying out the Music Rebalance functionality. I don't see it being able to do what we hoped it would do, namely allowing a mono soundboard to be converted into a stereo recording with Jimmy in one channel instead of both. The program works decently on existing professionally recorded stereo tracks. For instance, you can mostly remove the vocals from a song. However, it didn't work nearly as well on front of house soundboards. I'll keep playing with it, but unless I completely misunderstand how to use it (which is unlikely, as the interface is pretty simple), I'm not holding out much hope. If anyone else wants to try it, they offer a free ten day trial version that has full functionality, except you can't record what you've done (which is why I didn't post any samples of my tests.)

[SteveZ98]

One of the members at Royal Orleans, ssellers, just posted some samples there in the "Future Technology and LZ Soundboard" thread of mono soundboard tracks he converted to stereo using an open source program called Spleeter. This seems to be exactly the kind of thing we were looking for. I'm in the process downloading it and getting it set up. For anyone who's interested, here's the link to the Spleeter site. 

https://github.com/deezer/spleeter

[zeplz71]

I've been following some YouTube videos for a few days showing the software. It's definitely interesting and I did some playing around on a few Zep boots. A similar high-end technology was used on the recent Beatles Live at the Hollywood Bowl live album.

Here's a good channel to follow:

 

 

 

 

[zeplz71]

sample of Robert Johnson. Vocals on the left, guitar on the right channel .

WeTransfer link :https://we.tl/t-A2mirIgkcJ

 

 

[SteveZ98]

I’ve used Spleeter for a couple of days and have some observations based on that experience. The first is that this is a complete game changer for people looking to turn mono recordings into stereo. The default settings do a stunning job of separating instruments and vocals into individual tracks, and the system can be trained to be even better. However, there are a few caveats that need to be mentioned:

1 - The parts are less than the sum of the whole. By themselves, the individual tracks (aka stems) don’t sound great. The bass and drum tracks in particular have a phasey\watery sound to them. While you can tell what’s being played, the sound isn’t nearly as good as if you had a real multi-track recording and could listen to the bass or drums tracks from that. However, what’s interesting is that when you mix all of the stems together, the phasey sound disappears and the song sounds like it did before it was broken into parts, assuming you mix all of the parts back together. If you wanted to create a recording of Zeppelin without Robert’s vocals, you’d hear some unpleasant artifacts. Likewise, if you wanted to capture just his vocals, they won’t sound as good you as might hope.

2 - Mixing is an art unto itself. Just having the individual stems is no guarantee that you’re going to create something listenable from them. The challenge, especially with the guitar, is to balance it so that it’s more prominent in one channel than the other but still audible in both (it sounds unnatural if it’s only in one channel.) That’s harder than it sounds, and takes a far amount of experimentation to get the blend correct.

3 - Bleed through happens. While the system is excellent at pulling songs apart, at times sound from one part of the song leaks into other. For instance, on a version of Dazed from the last night at Earls Court, Robert’s voice showed up very noticeably in the guitar track in a couple of places. While that could be fixed manually, things like that mean you cannot fully automate the conversion from mono to stereo.

4 - It may not be able to separate some instruments. There are only a few default settings, and the one most suitable for rock music is the one that breaks a song into four stems, drums, bass, vocals, and other. That works great on a song like Sick Again, where the guitar ends up as the only thing in the other stem. However, I’m not sure what it will do with something like Since I’ve Been Loving You where both the keyboards and the guitar might be placed in the other stem. Some experimentation will tell me how much of an issue that is, so it may not be as much of a problem as I think it will.

5 - Spleeter is a command line program, meaning there’s no graphical interface and to get it to work you have to enter commands into a terminal. Not everyone is comfortable doing that, which will limit the number of people who can use it.

6 - It runs slow when it uses your computer’s CPU to pull apart the songs. It took approximately 12-15 minutes for it to finish working on a live version of Sick Again, which is a relatively short song from a live Zeppelin perspective. The documentation says it runs up to 100 times faster if you tell it to use your GPU. If you’re a gamer and have a high end graphics card, that’d be great but it’s not much help for those of us using lowly integrated graphics cards.

Even with these issues, this program is amazing. Bravo to everyone involved in creating it.

[rm2551]

Since Spleeter is open source it will only get better and better. I've said it before and I'll say it again - one day - hopefully in my lifetime - there will be the technology to turn Mike the Mikes recordings, for example, into almost professional sounding multi-track recordings. Noise and static carefully suppressed and the band enhanced beyond what was previously thought possible.

Kashmir from 21 June '77 is going to sound crystal clear with the clarity of a pro recording - but still with the live ambience. Not to mention TSRTS and that drum intro.

Jeeeez I hope I am around for that......