- Jun 19, 2014
You might be interested in this book, “Classical Recording A Practical Guide in the Decca Tradition By Caroline Haigh, John Dunkerley, Mark Rogers”On Friday 4th March my son Daniel asked me whether I wanted to help record a practice concert, happening on the next day, with his music theory teacher. He was a bit vague on the details (like when the actual concert was), but I thought this would be a great opportunity. Whenever you listen to a recording, one has no idea of the provenance of that recording - the sound of the performance, the acoustic, mikes, recording gear etc. How can we as audiophiles know how really transparent our gear is when we have a huge part of the chain outside of our experience? So I jumped at the chance, as it meant that I could evaluate the whole chain. It also meant that we could learn about the issues of recording orchestras - something my son (who is studying music production) hadn't done before. I thought that this would be just a bunch of kids learning music - but it turned out to be a specially commissioned piece with a capable amateur orchestra. And this was just a practice session, and Alida Watters (Daniel's music teacher) wanted us to record the full concert on Saturday 19th March at Gorseinon South Wales.
Readers of this blog will be aware of that some time ago I bought an AEA R88 stereo ribbon mike for doing Blumlein recordings. This is a very pure technique, as you simply use two mikes with a figure of 8 pattern - this means off axis sounds the mike has no output, on axis full output. You can read about it here. The technique has huge potential for transparency, as no mixing EQ or editing is involved - in our case the mikes go straight into the Focusrite Scarlett 18i8 ADC connected to a laptop which bit-perfectly records the digital data from the ADC. Monitoring is via USB to a Hugo 2 with DCA Stealths.
I must admit to having some trepidation about this. We have never recorded anything like this before - would we make a mess of it? How well would the Blumlein technique work in practice? On paper it's the most accurate and transparent technique available - but I don't have a single recording where I know that the technique is used. Most recordings are multi mics, with huge mixing consoles (and if you have ever heard the OP of a mixing desk direct it's not good too). The simplest technique used is the Decca tree, but that uses a simple mixer as it uses 3 mics. So why isn't Blumlein used more often? Perhaps it doesn't work too well...
Anyway, we set up the system (all battery powered to eliminate mains RF noise), and started monitoring. I used Hugo 2 and DCA Stealth, with my hands clamped hard against the ear cups; matched the volume, and compared the headphone sound to the actual sound, with me listening directly under the ribbon mics. First surprise - digital monitoring sounded very close to the actual sound. We took 5 takes, and subtlety adjusted the positioning, such that listening direct and with headphones gave no difference at all in the precision of lateral (left to right) placement - no detectable difference in angle, or how focussed the instrument was placed. That was a pretty stunning result.
Daniel in the foreground running the recording. Note the player on the left with his hood up - the noise from the heating system meant they had to practice in the cold!
But when we got home we checked the takes, and realised that listening at your leisure gave very different results to listening live. Without stress, ones sensitivity improves dramatically; also, you can go back and forth and compare the different takes much more thoroughly. And it turned out that the first take (take 1), albeit not ideal, was the best for depth; the other takes had more accurate lateral placement, with take 1 being slightly too wide. We figured out the reason for this - subsequent takes had the rear lobe of the mic pointing more to the ceiling, thus suppressing the perception of depth. For the actual concert (which will be joined by full wind and 4 soloists) Daniel will adjust the mic position from these lessons learned and hopefully we will get the best hybrid of the takes - that is good depth and accurate lateral placement.
I decided to publish take 1. Thanks to @Kentajalli for hosting this. If you listen on headphones use cross-feed set to max. Loudspeakers will be fine with no cross-feed.
So how did the takes sound compared to the live performance? In short I was pleasantly surprised as to how close it came, but with some extremely important differences:
1. The tonal balance - the recordings played back via speakers (something I can't talk about plus TT2 driving B and W 803D3) is quite a lot warmer than reality. Played back via (something I can't talk about and Dave with DCA Stealth) is still considerably warmer than reality, but more accurate in tonal balance than the loudspeakers. Having said that, the Church used for the tests is quite bright - most audiophiles would find it way too bright (but then many audiophiles don't listen to live non amplified music either and some audiophiles wouldn't know good sound if they were hit over the head with it as they listen by brand or wallet). Via the Stealth the balance is to my taste; but it exposes the real reason I went to the trouble of doing the lossless (or completely transparent) EQ on Mojo 2 - so that the system can go onto the ADC. Gentle EQ is appropriate (so long as it's completely transparent) for transducer and acoustic problems.
A quick word here about the Stealth - this is a headphone that impresses me the more I hear it. In particular the HF - as it has a very impressive purity and lack of resonances, and although the overall balance of the recording is warmer, it's nonetheless very accurate.
2. Depth. So on the live versus headphone digital feed by far the biggest change was the suppression of instrument placement depth. So the lead Cello is around 5 metres away from the mic; but listening on the headphones it immediately dropped by about 50%, to say 2.5m impression of depth. Listening to the takes again back at home reinforced that impression - Hugo 2/Stealth around 50%, x Scaler/Dave gave better impression around 66% of the depth (perhaps better but no more than 75%). On my 803, we felt it was pretty accurate sitting in the listening position - but of course this is artificially adding 2m anyway (listening position to loudspeakers).
3. Softness of transients. All the takes had a softening to transients - you can't perceive the starting of notes as accurately as live. This was apparent with live against headphones, and listening back home.
4. Timbre variation. Noting that everything is generally warmer than reality, there was one exception, the piano on the hard right side. This was very warm live (soft and warm like many English pianos and when used and the lid was down) but on the takes was brighter than reality, indicating that timbre variation is being suppressed.
5. Instrument separation and focus. This was pretty good, but not as good as live - it's not so easy to follow the lead violin when the cello plays.
Now the above deficiencies were entirely expected and I am convinced due to the ADC. Depth will be vastly improved by using the pulse array noise shapers, as it will resolve small signals much more accurately. I also know that aliasing - as I have been doing lots of listening tests recently - is a huge problem, and it directly affects the perception of the starting and stopping of transients. To stop this, you need to completely eliminate any trace of aliasing - to ridiculously small levels - and ADCs today have huge levels of aliasing. So I am confident that this deficiency can be overcome, or at least markedly improved upon. And careful readers will know from my posts that timbre variation and instrument separation and focus is down to noise floor modulation - something for which modern ADCs have large amounts of.
The experience also taught me that the Blumlein technique doesn't have any fundamental flaws. I was worried that it might over emphasise reverb - but the reverb you hear on the take is exactly as per the live acoustic, as the reverb in the church was a little dry compared to a concert hall. The technique in terms of lateral imagery (left to right) is nigh on perfect; placement of instruments laterally is the same as live whilst using the Stealth or the 803 D3. Moreover, instrument separation and focus, albeit not as good as live, I think is better than any recording I have. I wish all classical recordings used this simple technique.
I was also left with the impression how good the transducers (the R88, Stealth and 803) are - and also how good Hugo 2 is (and if I had used Mojo 2 I no doubt would be saying the same) compared to the live music. It was very much closer than I expected.
But my big and overwhelming excitement was hearing the deficiencies, and realising that these problems can, I think, be solved with the upcoming pulse array ADC. It's given me a huge stimulus to get the technical difficulties with the project overcome.
Just think - being able to close the gap from live non amplified sound to reproduced is a very exciting prospect indeed.
“Classical Recording: A Practical Guide in the Decca Tradition is the authoritative guide to all aspects of recording acoustic classical music. Offering detailed descriptions, diagrams, and photographs of fundamental recording techniques such as the Decca tree, this book offers a comprehensive overview of the essential skills involved in successfully producing a classical recording. Written by engineers with years of experience working for Decca and Abbey Road Studios and as freelancers, Classical Recording equips the student, the interested amateur, and the practising professional with the required knowledge and confidence to tackle everything from solo piano to opera.”