Clipping on glottal stops in audiobook recording

[Crystaljuggler]

I'm recording audiobooks at the moment, and encountering a problem with some readers, specifically female readers, who produce hard glottal stops at the start of words like "I" and "Arthur" and "Oliver". These hard sounds send a big clip through the recording chain and I'm at a loss as to how to stop it happening. The setup is:

Condenser mic -> mixing desk -> compressor -> de-esser/expander -> back to insert point on desk and out to an Mbox via the desk's main SPDIF output.

There are no attack or release settings on the compressor or vocal processor, and the peak just seems to go right through the chain and clip on re-entry to the desk, as it's still there if I take the main output down a tad. The problem is compounded by the fact that I can't reproduce the problem with my own voice, and I can't really ask one of the flighty actor-types to make nasty sounds for me while I fiddle about for a solution. Has anyone encountered anything like this, any ideas on fixing it?

[Steve Hill]

I'd like to hear a clip or see a waveform before opining further, but my quick thoughts are to take the de-esser out of the chain (use in post if needs be), and back off the compressor a lot from wherever it now is - again, you can add more in post if needs be.

Is there a popshield on the mic?

Get the actor/s to back off from the mic a bit, and/or have them address it slightly off-axis. With VO work I often have the mic slightly over their heads, knowing they will additionally be looking down at a script.

And (assuming budget permits) don't be afraid to take some time doing test recordings and making sure everyone is happy. They want to sound their best too, and should be happy to help achieve this.

[muzines]

What average levels are you recording at? With 24-bit recording, there should be plenty of dynamic range to get a level for the voice that doesn't clip even on the loud parts.

If you want to tame the peaks pre-tape, you'll need a compressor or limiter with a very fast attack time - it sound as if you don't have the necessary tools here, so I'd just drop the recording level and deal with the transients afterwords in software.

What's clipping - the preamp, or elsewhere in the chain?

Edit: Plus of course, forgot to mention *mic technique* too... d'oh...

[James Perrett]

Lose the compressor and de-esser and turn the gain down on the mixer until it doesn't clip. Process as necessary afterwards in software.

James.

[Hugh Robjohns]

+1 Wot James' said!

Hugh

[Neil C]

and I can't really ask one of the flighty actor-types to make nasty sounds for me while I fiddle about for a solution.

If they are professionals I would have thought they should understand why you would ask them to do that. And if they are not professional then I expect they have some aspiration to be and they should try acting like one. Unless they are irreplacebly unique, a VO person can't afford to be unnecessarily difficult to work with.
I'd be prepared to make any amount of silly noises if you were paying me.
It wouldn't take at all long to make some sounds to see if you have a suitable gain level.

[andy cross]

Good VOs typically cost more than studio time (mine, anyway...) so you shouldn't be wasting their time helping solve your technical issues.

So, keep the recording chain as simple and as short as possible, do no processing to tape, allow plenty of headroom, and do everything else required in post.

[Crystaljuggler]

Sorry I haven't been back earlier to say thanks for the replies. All some very good advice, and I think I've narrowed it down to the mic input gain being too high and then being battled downwards with successive items in the chain, rather than being set at a sensible level and then raised if necessary. I think just putting it down in text rather than trying to work it out while recording actually helped me think a bit straighter. Some good advice on taking time to work with the talent - but about the last thing I want to do is criticize their vocal technique. Glottics are something that got drummed out of me when I was at Drama school, and they are quite hard to reproduce on demand, and often we're a good chapter into the recording before the problem rears its head and it is really too late to start again.

We're recording in 16-bit, on ancient eMacs using firewire drives - I did ask about recording in 24-bit for greater headroom, but the Powers That Be have said that will take too much disk space. I'm very much the new guy coming in and asking silly questions like "Why has it always been done this way?" This, for instance, is one of those problems that has always been seen as "some readers just clip" and I'm trying to find a way of approaching the management and getting them to adjust the standard practices. Time will tell.

Thanks again for the good advice, it really is appreciated.

[zenguitar]

I think it's about time you started on your negotiating stance with the management with some cost/benefit analysis.

You might, for example, point out just how cheap hard drives are these days, and how affordable modern 24 bit interfaces are too. Contrast that small one off cost with the on going costs of the time it takes to fix every recording in the mix. And then sweeten the pill more by pointing out that those affordable improvements will be seen in every product across the board; getting the very best out of the expensive talent, improving the brand, and increasing customer loyalty.

Remember, you deserve new toys and it's the management's responsibility to supply you with them

Andy

[planetnine]

You're going to have to learn how to say things like "Nobody records in 16Bit, that's a playback resolution once things have been mixed and mastered" and "If people knew we did this they'd shake their heads or laugh at us" and "people haven't worked like this since 1995" erm, diplomatically.

Or let us know and we'll record several hundred people laughing if the diplomacy fails. Do you want that job that badly?

There's got to be a real reason, like some mastering machine can't read 24-Bit, surely...

Sorry

>