Bootlegs. Personal recordings. What should just be a harmless memento for a fan... but often can become a real irritation for the artiste!
We've obviously spoken at length before about Japan, and Osaka. Someone, somewhere, somehow made a bootleg recording of the performance, and this particular bootleg has been on Jessica's radar for a long time. Sage advice was given to us by a Source At Apple who said basically if the material wasn't on their books, we should go ahead and plunder any and all bootlegs and get them under our label as soon as possible.
So we did. And in doing so, we issue the following challenge - anyone who wants to make a claim against us for taking these recordings as our own...bring it on. Show us your contract with Mary or her agents. Show us where you have been paying performance royalties. Show us where you have been paying PRS (or equivalent PRO) to the song copyright holders. Show us how you have tried to stop the bootleggers. So there.
On the first listen to these, you realise it's a great shame that these weren't professionally recorded at the time. Like Live at The Royal Festival Hall 1972, this performance is a great snapshot in time, a truly world class singer at her peak with a breathtaking performance. There are the lovely moments when she introduces and is joined by Benny Gallagher and Graham Lyle.
You'll notice that I took out the playback track of 'Let It Be' - the audio on that is just awful - so if you want to recreate the vibe of being there, cue up your own copy! It will just sound so much better than anything I could possibly have done to the track from the performance! I have cleaned up the interviews to make them a bit clearer and hopefully more enjoyable.
The bootleg recording is what it is - it's fairly poorly recorded, the mix balance is wrong, and there's often no separation between the backing and Mary's vocal. Technically there are multiple issues - there's a constant 15kHz tone throughout, as owners of the bootleg CD will know there are static-like intermittent bursts throughout, there's a huge amount of hiss, and very often the instrumentation and vocals are distorted.
Up until fairly recently, this would have been the best we could have got, because all of those issues I describe above are baked in to the two tracks.
We don't have a multitrack recording of the performance. We also don't have the AI facilities that Peter Jackson and Giles Martin had available to them, but we do have a fairly extensive array of modern audio repair tools. So I put on my big boy pants, and got stuck in to the tracks.
I say we don't have the AI that Jackson et al., have access to, but we do have some tools that have rudimentary AI. These can learn the sounds you are trying to repair and as long as you don't go too far with them, they'll keep most of the original track intact apart from the bit you don't want.
I used a lot of Spectral Repair in this process - in fact it was my very first step on all the tracks. I used this in this first process to remove around 3dB of hiss from the recordings. This generally had the effect of making the recordings much warmer and easier to listen to.
There are times of course where it's very difficult to remove entirely - and I apologise for this if it detracts from your listening pleasure. It's noticeable in the end of for instance ‘Plaisir D'Amour' - I could have dehissed this section harder but it would have pulled all of the life out of the guitar, and some of the sparkle from Mary's voice.
I ran this as a whole across the track - I used this to control all of the other frequencies in the track and to artificially boost the vocal to try and give more clarity. The vocal - due to the dubious nature of the recording medium - is very honky around those critical frequencies between 1 and 3kHz . That's an exceptionally wide spectrum to try and carve frequencies from. It's also critical because our ear is particularly attuned to these frequencies. Mary has a natural strong resonance around 2.7kHz which I always treat accordingly in our studio recordings. Which leads to process 3.
Another stage of nascent AI control, this stage involves the rounding and smoothing of the high frequencies. In this case, we really want to hear Mary's vocal of course, but we don't want the EQ peaking that the recording medium is putting on to it - as described above. We're also competing with other features - the strings and in some cases, that blooming xylophone. So there's a stage of processing here that is attempting to sweeten the vocals, whilst making them easier to listen to, and also filling out the sound. If you have the "original" bootleg, you'll know the low and mid frequencies come and go. It's very hard to nail this down - if the frequencies aren't there, I can't put them back in, so I've tried to get a "somewhere in between" happy medium.
Over the last few years, EQ has changed - one of the brilliant pieces of technology in audio software is that you can now easily use an EQ dynamically. What does that mean? Well, it's a bit like - or in fact exactly like - having a compressor on a specific frequency. This means, for instance, that that blooming xylophone has high frequency fundamentals all over the place. Every time it's hit, it pings out massively, there are a couple of hits at around 4kHz that are like having a pencil jabbed in your ear...only without the benefit of gaining a pencil.
So, using a dynamic EQ means I can dial in the ear-jabbing frequency and set the dynamic element so when it gets really loud, the dynamic EQ tramples it down and keeps it to mildly eye-watering. The problem is you don't want to take too much of Mary's vocal out in those frequencies, and the beauty of a dynamic EQ is that instead of having to have an EQ on all the time (or having to automate an EQ to be on and off and on and off etc), the EQ rides it naturally, and only kicks in when it gets too loud.
I apologise again for those that do pierce through. I can only say - it's not as bad as the terrible original.
This is the bit I couldn't really automate.
All the way through there are artefacts that transpire from the rubbish nature of the recording. There's that irritating static thing that happens, and there's distortion...oh boy is there distortion, all the way through.
I've got no idea what the static is, other than it being static! But I managed to remove that with a clever bit of software, after a fair amount of trial and error. There are times when this has gone really deep across the track, and sometimes the sound drops out because of it, or there's a thump, which my clever tech can't remove. I hope what I have done is less annoying.
The distortion is another thing entirely. It's difficult to say exactly where the overloading happened - I doubt it was on the original broadcast, but of course it might have been, in which case, there's nothing that we could have done with that. Or it was done by someone who didn't know how to set levels and it overloaded either the mic, or the tape they were recording onto. Or it's a combination, who knows?
It's immensely frustrating - I can take some of the clipping away, the bits that make it sound like Mary is singing through a kazoo. It's still a bit crispy, but there's nothing I can do with that. In some cases I've been successful with it. In others (end of 'Rising Sun') I don't think there's any more I could have done.
Frustrating sums it up really. Even with the best tools available to me right now, this is probably the best I can do. A slightly improved bootleg.
It will be interesting to revisit this when the AI tools are a bit further down the line. I'm not an advocate of AI for its own sake but in doing audio restoration of projects like this...well, if it means we can get Mary's recording to somewhere near her magnificent performance, and available for us all to enjoy without having pencils jabbed ear'olewards, then it's got to be A Good Thing. Right?
The other release I have worked on - When Will I See You Again - is a bit more straightforward. This is an EP of songs featuring, or written by Benny Gallagher and Mary. The title track is a cowrite and duet by Mary and Benny, and is taken from a demo of Benny's. Track 2 is a live performance of Benny and Graham Lyle's 'First Leaves of Autumn'. This was taken from Benny's live album, and Mary added her vocal from her studio!
The third track is another Gallagher and Lyle track, 'Fields of St Etienne', but this is an intimate, stripped back version with just Mary's vocal and guitar, and is taken from an old Portastudio recording.
Another Gallagher and Lyle track - 'Breakaway' - finishes the EP. This is a bit of a curio - it was recorded by Mary in her music room, rehearsing with Benny and Christian Marsac, in preparation for the concert in Largs. I've done what I can with it, but it will only ever be a rehearsal tape!
These releases are all live right now, you can go and listen and buy them on all the usual reputable sources.
MHM017 - Mary Hopkin Live at Osaka Expo '70
MHM018 - Mary Hopkin and Benny Gallagher - When Will I See You Again
We hope you enjoy the Mary Hopkin Radio Show with Benny Gallagher, and equally hope that these new releases will bring you some joy and happiness.
This will probably be the last blog of the year, so from all of us, Mary, Jessica, Morgan and Chris, we wish you a very Merry Christmas and a Happy New Year!
Christian Thomas is Production Director at Space Studios and the recording and mix engineer at Mary Hopkin Music