audio engineering

(See also the first post about reverb)

If you have a second tool in your toolbox for live sound at a 24 hour kirtan (also known as a "24 hour continuous soundcheck"), get a Graphic EQ.

If your kirtan is in a temple, then odds are on that it has a resonant frequency that is going to affect the sound.

A room's resonant frequency is related to the distance between the walls. Most sound waves will bounce or penetrate the walls in predictable fashions, and fall off in volume as they do so. However, the sound waves whose wavelength matches the distance between the walls will instead amplify as they bounce back and forth between the walls. This effect is most pronounced in the bass frequencies (<200Hz).

The effect is like this: you set your bass guitar volume at a good level for the kirtan, and play along; but there is one note - when you hit it, the whole room lights up: "Boom!"

Equal sound energy output from the amplifier = more volume in the room whenever that note is sounded.

Another example: you take two mrdangas whose bottom ends are tuned to different notes. You play them outside, and with equal force the volume of each drum is equal. Take them inside the room, however, and suddenly the drum that is tuned to the resonant frequency of the room sounds much louder when hit with the same force.

For more information on room acoustics and resonant frequency, check out this article: Room Acoustics.

So: use a Graphic Equalizer to notch that frequency down, to compensate for the increased energy of that frequency in the room.

You need to identify the room's resonant frequency. The only way I know to do this, at this point, is to observe the effect when playing an instrument - that's how I came across this phenomenon.

In the Sydney temple it's D. In Canberra at the recent 24 hour kirtan it was G. When I hit those notes in those venues on a bass or the bottom strings on a guitar: "boom!"

Here's a handy chart to convert notes into frequencies:

So for Canberra we'd notch 48.9Hz, 97.9Hz, and 196Hz, and possibly some immediately adjacent, depending on the Q-factor of the resonance. You can think of the Q-factor as "the width of the room's resonant frequency band". The bigger the Q-factor, the wider the spread of the resonant frequency of the room across the frequency spectrum. Parametric EQs sometimes have a Q-factor settings that allows you to widen or narrow the EQ range of a cut or boost. On a Graphic EQ the Q-factor is usually set, and you use adjacent frequencies to widen the spread further.

For Sydney temple we'd notch 36Hz, 73Hz, and 146Hz as the starting point.

Something we struggled with in Canberra was the resonance of mrdangas. I noticed that one mrdanga was tuned close to the resonant frequency of the room. Whenever someone played that mrdanga it was loud. The other thing was that near the wall, where the kirtan leader was sitting, it created a powerful booming sound. Kirtan leaders would request me to "turn down the mrdanga in the foldback", to which I would have to reply: "there's zero amplification on the mrdangas!". Also, that mrdanga would create problems with the vocal mics as the powerful resonant frequency entered them and caused a low frequency feedback loop in the amplification system.

Without a graphic EQ we dealt with it by reducing the bass frequencies on each of the the vocal mic channels. This is a less-than-ideal solution, however, as it affects the tonal quality of the singers. With a Graphic EQ we could have identified and notched specific frequencies across the mix. This would have resulted in a more precise and targeted reduction of feedback without such an effect on the singers.

With instruments like a bass or a guitar, which create the resonance through the amplification, a well-targeted Graphic EQ will eliminate the problem. For feedback loops like those created by the mrdangas, the same approach will reduce the impact. However, to fix the sound of the mrdangas in the room requires one of two things: retuning the mrdangas (not a reliable approach), or changing the structure of the room to reduce the reverberation of that frequency.

I'll be investigating that before next year's event in Canberra. I'm sure there are some low-cost, low-impact, high-effect measures that we can take. I'll let you know how it goes.

In the meantime, check out this video:

If you have only one tool in your sound toolbox for a 24 hour kirtan, it should be reverb.

In this video from the recent Canberra 24 hour kirtan I used my Zoom 1202 reverb unit to give Janardana's voice more "epic".

Reverb and perception - psycho-acoustics

The sound captured by the recording device differs from the sound in the room. In the recording the reverb effect is more pronounced. Try this experiment: next time you attend mangal-arati in a temple, record it, then listen to the recording. What you will notice is that in the recording you'll hear an echo-ey reverb sound that you didn't hear in the temple while you were there. The reason for this is that the brain interprets the live sound with reverb cancellation. It is there "colouring" the sound, but you don't consciously perceive it, unless you are specifically listening for it - and even then it takes some training to bring it to conscious awareness.

When you listen to a recording in a different space, however, your brain doesn't do the same cancellation. Consider this: as you listen to the recording of Janardana singing in this 24 hour kirtan you are consciously aware of the reverb on his voice, and it sounds artificial.

However, you are not aware of the reverb in the room that you are in - but it is there. The recording is playing, and your brain is psycho-acoustically cancelling out the reverb effect of the room that you are in. So for people in that room while Janardana was singing live, the reverb effect was perceived differently. If you were there, you'll know what I'm talking about. It didn't sound like that.

Compare Janardana's kirtan with this one, from 2:30am, where I was singing into a mic with no reverb:

A couple of things: there's "no obvious reverb effect" on my voice compared to Janardana's voice; and you're still not conscious of the reverb sound of the space you're listening in right now. Try making a recording of your voice in that room, and you'll notice it on playback.

Using reverb in a live mix

The idea with reverb is to be subtle. It's an effect - not a component. It should be felt, not heard - like a spice. Enough to modify the taste, but not as an overpowering taste in its own right. If you spice a meal right, people should say: "That tasted great!", not: "That tasted like paprika". Diners ask great chefs for their recipe, because it's not immediately apparent what has been done to make it taste so great; and so it should be for live mixing. A common give-away of an amateur is to crank up the reverb until it screams: "O Hai! I'm putting reverb on the vocal!"; at which point they feel satisfied that they've put enough reverb on it - because it now sounds like there is reverb on it.

The idea is actually to not make it sound like it has reverb on it, but to create a different spatial awareness in the minds of the audience; to use the reverb effect to transport the audience into another realm. It creates a mystical experience that makes the kirtan "larger than life".

A handy technique is to push the reverb up until it is consciously perceivable as "a reverb effect" in the room, then roll it back so that it disappears from conscious awareness as reverb, and instead appears as "epic".

The Zoom 1202 is a mid-90's reverb effect that I bought 30 minutes before the gig from a Cash Converter's, during Gaura Vani's 2009 Australian tour. On that tour we used it to amazing effect on an auxiliary channel, piping the flute and vocals into it.

Stereo panning is an obvious way to create separation between components in a mix (vocals, instruments, etc), but you can use reverb to achieve a similar effect when mixing in mono, by placing instruments and vocals in different "spaces" through reverb. If a sound has a different reverb setting, then the brain will perceive it as existing in a different space, thus separating it psycho-acoustically in the mix.

When using reverb in this way, nudge the reverb up on different components at various times to cause them to "feature" in the mix.

Rather than setting and forgetting, actively work the faders and the reverb sends to bring mix components to the fore. The best way to do it is to bring something up, either in volume or in reverb space, then fade it back down into the mix. The effect of this is to colour the mix. Once the brain has been made aware of a mix component, even when the mix component retreats back into the same space as the other mix components, the mind is still aware of it. So you don't need to get into the volume war, where you make the overall level of the mix progressively louder. Instead, bring things up, then fade them back.

The effect of doing this will be to make the whole experience more epic for the audience. Think of yourself as a tour guide pointing out notable features to the audience, then allowing them to fade into the distance as you pass them.

Remember: the effect of reverb should be felt; reverb should not be heard.

While travelling with Ekendra to the 24 hour kirtan in New Gokula, we listened to some rough mixes I've done for a live kirtan album I'm working on. Ekendra's van is totally pimped out with the massive subwoofer. I mean it has mad, ridiculous crazy bass.

I noticed that the mix I had done didn't cut it with the other tracks he put on. I don't have a sub in my mixing setup at home, so I didn't notice it before this.

I did notice, however, that the mrdanga sound wasn't good on the bottom end. The mrdangas that were used in the kirtans were not so good, so they didn't record well.

This triggered me thinking, and I remembered a technique that I'd used before - one I got from an RnB producer.

Here's what you do:

1. Copy the mrdanga bottom end track to another track. We'll call this track the "mrd sub" track.
2. Put a low pass filter on it, and roll off everything above 100Hz.
3. Switch it out of mix by assigning the output to null.
4. Create a new audio track. We'll call this the "sub bass" track.
5. Put a test oscillator (tone generator) on it, set to a sine wave.
6. Tune the test oscillator to the fundamental of the kirtan, using this chart of musical notes to frequencies. Obviously you should tune it down in the bass region, below 100Hz.
6. Put a noise gate on the oscillator channel.
7. Set the side chain on the noise gate to trigger from the mrd sub track.
8. Tune the noise gate threshold so that it only triggers when the bottom end of the mrdanga is struck.
9. Tune the release of the noise gate to give your new mrdanga bottom end the resonance you want.
10. Roll off the bottom end of the original mrdanga track and mix with the new sub bass track.

Voila, a new mrdanga bottom end - tuned to the kirtan and giving you lower end power.

I still don't have a sub, so I can't really mix the kirtan well for that kind of listening environment; but with this technique in hand again, I can make moar good the recording.

Postscript: You might notice that the Logic noise gate pictured above has a filter on it for the side chain. So you don't have to put a low pass filter on the mrd sub track if you're using Logic - just set the Noise Gate's "Side Chain High Cut" to 100Hz.

A quick remix with an electro vibe to go with the Autotune:

• Bachhu Tune Electro Remix (mp3, 3.3MB, 2:24)

The previously uploaded version was cut short by about a minute. This new one is complete.

I spent some time today doing a little recording. My goals today were to start to iron out a work flow for recording, in preparation for the World Sankirtan Party album recording slated for next month, to play with a tune that I really like, and also to play with auto tune.

Here's the result:

• Bachhu's Tune Demo (mp3, 2:24, 3.3MB)

There are three guitar tracks (one rhythm, one lead, one harmony), a bass guitar track, two keyboard (gospel organ) tracks, and a lead and backing vocal. I used autotune on both the lead and backing vocal.

A few observations:

• Autotune rocks, and is the next evolutionary level in music in general, and recorded kirtan in particular.
• Autotune works better when you give it the scale to work with, rather than allowing it to work chromatically.
• The main vocal, which I did in one take, could benefit from a retake to get rid of some of the "warbling" that autotune produces when the pitch is indecisive.
• Getting a good result with autotune requires a particular singing technique (with no vibrato) which I will have to work to master.
• This tune rocks. I heard it from Bachhu when he came up to Brisbane for Maha Kirtan 2.

Excellence is not accidental - it's intentional.

Video courtesy of Krishnapada.

Here are a few subtle highlights of the sound mixing in this video segment:

You can hear at various points shifts in the balance of instruments. For example, the top end of the tabla comes up at 1:14, around 2:31 on the "turnaround", and again at 4:37. Changing the balance of the instruments creates a dynamic feel to the kirtan. It renews interest and energy and keeps it fresh.

There is a chorus effect on the harmonium during Vraja Dhama's kirtan. It's most noticeable around 3:37 during the response.

The booming of the bottom end of the mrdanga around 3:38 was me mucking around with Krishnapada. I maxed out the 80Hz on the AKG D770 mic we had on the bottom end of the tabla, and this caused the 800W subwoofer we had hooked up to go to town.

In the kirtans overall you can feel and hear a strong presence in the elusive 80Hz-120Hz range, delivered with authority and precision by an RCF ART 705A 800W active sub with an 80Hz crossover. This is balanced by a reasonable amount of 12kHz-20kHz provided by a few select cartals and a tambourine played by competent players. The overall effect was a relatively high SPL (sound pressure level) that was as much "felt" as it was heard, and didn't leave you with ringing ears and hearing damage afterwards.

Heaps of cartals in the 16kHz-20kHz range will make you deaf. Heaps of bass in the 80Hz-120Hz range will make you shake your booty.

To deliver those frequencies you need Lots.Of.Power. Shifting the amount of air needed for low-frequency, long wavelength sound needs a powerful amp and a big driver. The RCF ART705A, with 800 watts of power amp and a 15" driver delivers this.

Two of these would be ideal to get an equal experience throughout the room, and Yes, we will get there; but in the meantime we had a 100W Ashdown bass amp near the musicians to provide bass guitar foldback, and put the subwoofer at the back of the room to minimize feedback through the mics.

Bhakti Yoga is the "Yoga of Entertainment". Entertainment, from the Latin tenere, is all about capturing and holding the attention. Bhakti Yoga is about focusing and directing the attention to the transcendental reality beyond the veil of Maya.

Here's the final, stand-up kirtan of the Sunday Feast, ably lead by Maha-mantra das.

It's a simple four mic affair: a Shure WH-30 condenser headset mic for the lead vocal, a Behringer C2 condenser for the room, and a couple of Shure SM58s - one for the backing vocal (Param Satya and Prahlad), and one for Sridhar's saxophone.

If I'd had time to set up (this is just straight after the "stage" kirtan), I would have set up two additional mics - another C2 for a stereo image, and an omnidirectional condenser for more of the room, which would have helped to put more drums in it - I could have taken all the top end off that one and boosted it up to put more bass in the mix.

I've mastered this one really hot, and left the cartals in their full glory (I usually put a low pass filter on the room mics to bring them down). The vocal is so high that it's still the most prominent element, and that's the most important thing - the chanting.

• Maha-mantra Sunday Feast Kirtan (mp3, 30MB, 21:48)

Enjoy!
- Sitapati "That's me playing the chimptas" das

Here's a recording of Radha-ramana-hari from last night.

A few of things about this recording:

1. The main vocal is a bit distorted. This is because I used an extreme microphone technique, super close to the mic. I did this because for the two preceding kirtans I was manning the desk (as you can see from the video above). When it came my turn to sing I was in a different position and it sounded as though my voice wasn't coming through the PA. I had set up a foldback monitor (the Roland Street Cube) - but since things were running late (let me just add that I was personally on time) there was no time to soundcheck. In fact my voice was coming through at a normal volume, and my close miking caused the signal to clip.

2. The first part of the kirtan kicks up quite a few bpm when the tabla comes in. We'll need to practice this for the July 3 gig at Fusion, including some time with a metronome for me, I think. I am also going to separate the two melodies, so the first Hare Krishna kirtan will get a part C and maybe D, and Radha-ramana-hari stays as it is with Parts A (Radha-ramana-haribol), B (Sri Krishna Govinda...), C (Hare Krishna low), D (Hare Krishna high), and E (Sri Krishna Govinda high).

3. Without the sound check I wasn't able to get things set up nicely, but I think the two things that I could have done to make the sound better would have been to EQ the room using a graphic equalizer on the whole mix, and move the mixing station further back. Doing that will require a multicore - I'm going to look into that soon.

• Radha-ramana-hari (mp3, 26MB, 18:53)