If you do recording long enough, you’ll eventually be asked to record some sort of spoken word.

Maybe your great uncle Joe-Bob wants to narrate his life’s story for posterity, or perhaps you’re asked to record the sermons at your church. Or maybe you’re simply running live sound for some sort of speech.

It can be tricky.

Especially when you’re trying to make the voice nice and loud, easily heard, and at a consistent, steady volume.

Lots of speakers are very dynamic, meaning they have moments where they whisper and others where they yell.

It can be a nightmare to try to reign them in.

You COULD try to “ride the fader” the entire time, but that’s not very effective and isn’t realistic.

So? You reach for compression.

The problem with compression, especially when you’re dealing with a very dynamic source, is that it becomes easy to over-compress, to over-compensate.

And on a source like a vocal, over-compression is VERY obvious, and doesn’t sound that good.

I’ve found that using a very high ratio, combined with a high threshold and a lot of make-up gain, seems to be the perfect combination for taking care of that pesky dynamic speaker.

The quiet parts come through nice and clear.

The loud parts get squashed into submission.

And nothing sounds too over-compressed.

Of course, you’ve got to set it up right.

And that’s what I show you how to do in one of the new bonus videos inside of Understanding Compression.

‘Tis a lot of fun. Get started here:


  • Hillel

    Which video is the one you’re referring to when you write “And that’s what I show you how to do in one of the new bonus videos inside of Understanding Compression.”? I have some spoken word studio work coming up and I’d like to bone up on the technique before I start working

    • Hey Hillel,

      I’m talking about the Bonus videos inside of http://www.UnderstandingCompression.com


      • Hillel

        Hey Joe! ,

        I have Understanding Compression. The 3 Bonus Videos are 1.Maximum Volume,
        2.Understanding Gates and Expanders, and
        3. Understanding Limiters.,
        I’m not sure I get which one is geared towards spoken word…. Sorry for making you nuts

        • Maximum volume I believe. Also the Gates and Expanders video. The audio I use in those is spoken word.

          • Hillel

            awesome, Thanks! and btw I just became a VIP member! crazy excited!

            • So glad to have you!!!! Don’t forget to say hi in the forum.

          • Hillel

            awesome Thanks!

  • Frank Adrian

    My keys to a great spoken word recording?

    1) Pick the right mike. There are a variety of mikes that will work well for this kind of recording. You can find a variety of dynamic mikes for this under the “broadcast mike” category. The Electro-Voice RE20 is an example. The Shure SM7B is another. They’re dynamic and have internal pop screens. As such, they’re really good at cutting down extraneous noise. But, being dynamic and (again) having internal pop screens, they need a strong voice to operate well and you might not get enough air in the recording. As such you might want to use a nice condenser mike to capture a more natural sound. For more strident voices, you might use a ribbon mike to get additional smoothness. As usual, taking the time to find the right mike for the voice will pay off greatly in the quality of the recording. 

    2) Record in as dead a space as you can manage. Future compression will exacerbate natural reverb and change its tonality, making the vocal sound uneven. You may not be able to get the person in a perfect vocal booth, but you can probably get your recording space a little less reverberant. Put down a carpet and record them somewhere other than the bathroom.

    3) Get the subject as close to the mike as you can. Close miking increases the ratio of direct to reverberant sound. Your ability to compress transparently goes up concomitantly. That being said, most non-professionals don’t “work the mike” particularly well. So be ready with the following…

    4a) Use a pop screen. You may not be recording a singer, but it is still a human voice. And human voices have plosives. A pop screen will help the spoken word, just as it does a song’s vocal.
    4b) Compress going in. This is a hard one for me to recommend. I usually hate inbound compression because, if you get it wrong, you can’t undo it. Nevertheless, dynamics happen – you probably will need to compress to tame peaks. I like using a compressor at less than a 4:1 ratio with a gentle soft knee (about 6db). Make sure that the compressor is not doing anything most of the time. If you’re seeing gain reduction anywhere but where plosives and (perhaps) fricatives are being processed, you’re probably compressing too much. And the gain reduction shouldn’t be more than 3-6 dB when it does happen. Use a relatively short attack (5-10 ms) and release (40-75 ms). Watch for “t”s turning into “d”s and “s”s turning into “z”s, if you make the attack too short and the amount of compression too high.

    4c) Watch your gain structure. Most decent mikes and preamps (and you should use a preamp) won’t overload with spoken word material. See the above guidelines on getting the level into the compressor right and make sure you do a test read to make sure everything is set right.

    4d) When all else fails, use a safety limiter, too. Some folks don’t know when to stop. A  limiter with a hard 0db limit (especially when you don’t see it doing anything) can still be a lifesaver. Again, a soft knee will help with transparency.

    5) Don’t ignore editing. If the timing of a phrase is off a bit, fix it. Same way with getting rid of bumps, scrapes, page turns, etc. If you want to be really pro about it, replace anything you edit out with bits of the room noise taken from quiet places in the recording (yes, I am obsessive).

    6) Ride the fader using automation. Calm down those loud parts using automation envelopes. It will make a difference. Plus, you can decrease too loud breath sounds (don’t get rid of them entirely, though – real people breathe).

    7) Use de-essing. If you can get an automated de-esser working well enough, good on you. I usually can’t. You can also manually edit a volume automation envelope to decrease the volume of fricatives and smooth out the start of dental stops – but again, watch for consonant shifting.

    8) Use eq. You can add a bit of air and increase intelligibility (by boosting HF), add power (by boosting MF), cut vocal muddiness (by cutting LF), etc. Plus, people have varying formant frequencies – increasing or decreasing the level at the formant’s frequency can make a huge difference in vocal resonance.

    9) Use compression for post-processing. Slow down, cowboy! Just because you’ve compressed coming in doesn’t mean you’re done compressing. Or maybe it does – it depends. You still probably want some amount of compression to add a bit of heft to the vocal. Just don’t have it huffing and puffing like a steam locomotive, OK? Also watch the compression level carefully because it can undo everything you’ve already done with de-essing and breath removal. Just like in musical vocal recording, you might want the compressor and eq in the chain while you’re editing. If you’re completely obsessive, you might want to use a multi-band here – low frequencies (85-250 Hz) contain fundamentals but add more mud than tone – you may not want to compress these frequencies very much. Vowels (which power the voice) come in at 350-2KHz – if your reader is weak, compress this band to give it that “voice of God” power.The consonant sounds are dominant at 1.5-4KHz – compressing here will add a bit more intelligibility (but go easy – watch the dreaded t-to-d and s-to-z shift).

    10) Add a bit of reverb for sheen. Real spaces have reverb. A totally dead recording (which you should strive for initially) sounds sterile and cold. The point in recording so as to reduce the natural reverb was not to get rid of reverb, but to let you more easily control the same. Add a small bit of reverb to add the life of the room back in. Hint: Just as with vocals, sometimes a set of small, discrete echoes can sound better (and more intelligible) than an artificial reverb. Even better? If you can’t get rid of all of the reverb in the room (and you probably can’t), use a room mike when recording and add back a bit of the room’s natural reverb if it doesn’t sound awful.

    11) Use your judgment. As always, if it sounds good, it is good. Get it right going in. Don’t overcompress.

  • Jac Mandel

    Many times I also use several compressor plugins in series, each doing a little bit of the work, but in the end sounding quite transparent even though the resulting level is very even and loud.

    • Yeah. That’s a popular approach. I found that I end up not needing the multiple plugins. I usually don’t need compression on the quieter parts, so having a compressor compressing those parts doesn’t seem to be necessary. But that’s just my experience.