July 9, 2009
In the last 2 weeks, I’ve recorded several interviews at 2 different trade shows. Those interviews have ended up in episodes of both Shutters Inc and Sine Lanugage.
In the wake of this, I’ve had a couple of enquiries from listeners as to my technique for recording interviews on flash-based field recorders like the Zoom H2.
These listeners have commented that they never seem to achieve the same level of results as I have managed, and have asked for some insights.
So, I shall endeavour to outline the pitfalls as I understand them.
First off, tempting though it may be, do not record to mp3!
Remember, mp3 is a lossy format, and we don’t ever want to save production audio (clips which still need further work before release) in a lossy format.
And for that matter, don’t record at 16 bit wav either.
No, your preferred option is to record 24 bit wav.
Yes, it will chew up your memory cards quicker, but memory cards are really not that expensive these days, so carrying a couple of extras shouldn’t represent too much of a burden, either physically or financially.
If your recorder of choice does not offer 24 bit wav, then fall back to recording 16 bit wav instead.
Secondly, your recorder SHOULD offer a choice of microphone gain sensitivities.
The Zoom H2 offers low, medium and high.
Low will turn the sensitivity down (useful for really loud sources), medium is what it sounds like, and high turns the sensitivity up (for really quiet sound sources).
I have found that the medium setting usually works well for these trade show interviews, but obviously, judge each on a case by case basis.
Remember, we are recording at 24 bit, so we don’t NEED to peg the meters at zero!
Peaks of -20dB to -12dB are just fine!
Third, if your recorder of choice has a headphone output (I don’t imagine there’d be any which do not, but you never know), then absolutely have some form of monitoring with you when you are recording.
This may be a set of lightweight street headphones, or even a decent set of earbuds.
Me? I use my trusty old Sennheiser CX300′s, with just one earbud stuck in one ear.
The reason for that is that through that ear, I can hear what the microphone is picking up, and through my other ear, I’m hearing the world around me.
Now, because you are monitoring (via your earbud) what the microphone of the recorder is hearing, you are able to move the recorder around as necessary througout the interview to make sure the talent stays ‘on mic’.
Now, you might be thinking that people aren’t going to like having a flash recorder stuck in (and moving around in front of) their face.
I would contend that if they have agreed to do an interview, then they are probably going to be ok with it.
My technique is to hold the recorder at chest height between myself and the talent.
That way, you SHOULDN’T get any plosives (pops), but the mic should be able to hear the talent fairly well, while keeping the ambient noise reasonably under control.
If you talent is a very soft speaker, then you may have to move the recorder closer toward them, and that may feel a little uncomfortable at first.
If the talent keeps backing away from the mic, stop the interview, explain to them that you NEED the mic that close in order to hear what they are saying without being drowned by background noise, then recommence the interview.
Thing is, MOST of the time, the person you’ll be interviewing is from the marketing department or the sales team and they generally don’t speak that quietly!!
OK, so now you’re back at your desktop (or in your hotel room working on your laptop) and ready to edit and mix.
Drag the files into your DAW of choice.
DO NOT go and normalise the waveforms!
Remember, they’re 24 bit files, so it’s all good.
In your multitrack (which is also mixing at 24 bits or higher, right? RIGHT??), lay up your interviews where you want them.
Adjust the gain so you’ve got peaks around -20dBFS to -15dBFS off each channel. At this point, you should have NO processing on your master output.
Put in some per channel automation to keep each interview roughly in the bacllpark in terms of output level. You don’t have to get too finicky with it, just ‘in the ballpark’ will be good enough at this stage.
Now, if your final audio piece is going to feature other pieces of audio as well, I’d suggest setting up a submix (buss) for just the interviews to go through.
Then, slap a peak limiter across that buss with an output level set for -15dBFS, and the threshold set so that you’re getting about 4-6dB of gain reduction on that peak limiter.
Then, AFTER the peak limiter, put a compressor with a moderate attack (~20-30ms), moderate release (~100-150ms), a medium ratio (3:1-5:1) and again, enough threshold to give you another 3-6dB of gain reduction.
Your interview submix should now be exhibiting tightly controlled dynmaics, but not sounding squashed.
Go ahead and mix it in with all your other audio bits so that everything sounds roughly equal in apparent volume.
Slap a peak limiter across your master output, and you should be cookin’ with gas!
October 5, 2008
This week, a ton of e-mail to answer from the last month or so,
including some reminiscing about the transition from analogue to digital within the radio industry through the late ’80′s and early ’90′s,
Ron Eastwood tried recording quasi-binaural on a boom box,
Jim Addie sent in a link to an interesting article on the merits, or lack thereof, of recording at higher sample rates and longer wordlengths (when I recorded this episode, I commented that I hadn’t read the entire article. I now have, and have the feeling that at some point in the past, I’ve been pointed to it, and have actually read it. Still, it was good to read it again!),
Greg Andreson (who is all “Bruced” out) wrote to tell me about his Zooms (the H2 and the H4!) and how much he likes them, and to comment on the wildly different standards of audio production that exist within the podcaster community,
and Pascal asked about sidechain compression.
One free VST plugin that I know of which does sidechain compression is Sidekick.
And finally, a bit of a chat about what is in store for episode 100.
August 17, 2008
This week, so many e-mails to answer, I couldn’t get to them all!
Jim Addie chimed in with some follow up on DC offset (thanks, Jim!),
Steve Mayfield wondered if I’d got my HDMI issues sorted (Yes! Thanks, Steve).
He offered a link to another site that reviewed some hand held field recorders, registered his ‘vote of confidence’ for a video podcast for SL100, and expressed his appreciation of the new audio2u.com podcast imaging.
Then, we heard from Ike Tamigian who owns the Tascam 122L audio interface and reckons that “for the money you can’t go wrong”. Thanks Ike!
Then, Noel Payne asked about what eq filters (high pass, low pass, notch, band pass etc) to use under what circumstances. And after I finished editing the podcast, I realised there’s more I need to add to this topic, next week. Stay tuned, Noel!
And finally, Ron Eastwood wrote in with a photography analogy regarding sample rate and bit resolution, and asked what is the most important factor (with regards audio quality)?
June 29, 2008
Hold on to your hats, people…
In episode 92, a listener (who wishes to remain anonymous) commented on the possibility of hearing damage inflicted by driving with the windows down, and the possibility of same being caused from riding your pushbike in busy traffic. The earplugs he’s using while riding are these.
Of course, my regular listeners know that I love my Sennheiser CX300′s… damn, I oughtta be on commission!
And that led to a discussion of noise-isolating vs noise-cancelling earbuds and headphones.
All of which led me to speculate on telephone handsets and whether or not they do us more harm than good.
Then, Jim Wesihorn checked in with news of a new pdf from the good folks at Izotope (makers of the VST/DirectX Ozone mastering plugin).
Seems they’ve turned out a whopping great 75 page document on audio restoration.
Going to have to check that out some time soon!
Then, Steve Mayfield sent me a link about a dummy head binaural mic for your video camera!
Crazy looking thing.
One would hope it works well given the price tag!
Another long time listener who wishes to remain anonymous wrote to assure me that no, there is no distorion in my podcasts!
Phew! What a relief!
I didn’t think there was, mind you.
He also went on to provide some interesting real-world feedback on sample rate conversion.
Then, Matt sent me this screen shot of the Pro Tools manual where apparently, even Digidesign doesn’t know the difference between “bit rate” and “bit resolution”.
Why do I feel like I’m fighting an up hill battle here?
Then, Luke asked about vocal training.
I suggested googling some of these words:
‘voice’, ‘vocal’, ‘coach’, ‘training’, ‘tuition’, ‘diction’, ‘eloqution’, plus your local area
or check your local phone book.
Next, E. Bernhard Warg sent me an mp3 of a shoot out between a Coles 4104 ribbon lip mic and a standard lav mic, with regard to ambient noise suppression.
Then, Jim Addie sent in some great information on how the Red Book standard was established.
And finally, there were some listener comments on audio2u about my disaster last week.
October 14, 2007
This week, I read an absolutely amazing thread on prosoundweb about digital levels.
Some of the ideas fly a little bit in the face of some of the things I’ve suggested in the past (particularly with regard to tracking as hot as possible).
Having now finished reading it (I was still half way through it at the time I recorded this episode), the main gist of what was being discussed (and there were some pretty big industry heavyweights in there) was that if you are tracking in 24 bit, there is no need to aim for the hottest possible level to disc. These guys were advocating tracking with average levels around -20dBFS (and peaks around -12dBFS)! Now, if anyone had tried to sell me on that idea a week ago, I probably would have held on to my existing position and disagreed.
But no, the theory makes a whole lot of sense.
In a nutshell, the idea is this:
At 24 bit resolution, we’ve got 144dB of S/N ratio to play with (or you could refer to it as dynamic range if you wanted to).
NOTHING (Shall I repeat that? NOTHING) that you are going to record into your DAW has that kind of dynamic range.
Nothin’, zip, nada, nil.
For podcasters, the widest dynamic you’re likely to deal with will be your own voice, and even if you’re REALLY inconsistent with levels (and don’t own an outboard compressor), the most you’ll have to contend with might be in the vicinity of 50dB (from the quietest passage to the loudest passage)…. but even that is unlikely.
So, if you track with peaks at -12dB, AND you happen to have a soft passage 50dB below that (-62dBFS), you’ve still got 82dB of S/N below ya!
But why would you need to track that low?
Well, the theory (and according to these heavyweights who are tracking and mixing this way, the practise) suggests that if you track too close to full scale, you might introduce clipping when you start running plugins (EQ, compressors, peak limiters, whatever) that don’t operate at higher bitrates.
Look, it’s not a thread for the faint hearted, but if you have an interest in getting the best possible quality out of your digital audio setup, it’s worth the time and effort.
It took me 4 days to get through it because I kept on re-reading lines and paragraphs to make sure I understood it.
Also, the real meat starts about 7 pages in when Paul Frindle weighs in on the discussion.
OK, enough of that.
This week in the podcast, part 4 in the series on constructing your own promo: some tips on track laying.
July 22, 2007
This week’s episode was inspired by a comment on the Zoom Handy H2 portable recorder blog post.
Joseph (from English Mojo) asked about what to look for, and what to avoid, in a field recorder.
After running through a checklist, I moved on to a discussion of how to calculate data storage requirements based on given sampling rates, bit resolutions and channels.
Here are some of my notes:
1hr @ 44k 16 bit stereo = 10MB/min or 605MB/hr
1hr @ 44k 32 bit stereo = 20MB/min or 1210MB/hr
1hr @ 48k 16 bit stereo = 10.9MB/min or 659MB/hr
1hr @ 48k 32 bit stereo = 21.9MB/min or 1318MB/hr
1hr @ 44k 16 bit mono = 5MB/min or 302MB/hr
1hr @ 32k 16 bit mono = 3.6MB/min or 219MB/hr
1hr @ 22k 16 bit mono = 2.5MB/min or 151MB/hr
MP3 (CBR 320)
1hr @ 44k 16 bit stereo = 2.29MB/min or 137.3MB/hr
July 4, 2005
In episode 2, Bruce describes how to set up both Windows and Audition, so you can record your voice from a microphone into Audition.