Instructions on how to remove breath in audacity (reduce volume)

[gauguin]

Hi Friends,

I have been playing around with options on how to deal with the sound of my breathing in audacity.
One technique of course - if you are really good at it - is to turn your head away from the mic when breathing but that will alter the sound as you turn away/back to the mic as if the sound fades off and in.

The other is to deal with it in audacity. This is a great video for it (which is to reduce the volume of the breath sound which is more natural than to silence or cut it):

https://www.youtube.com/watch?v=eYYypOQfxIM

another video by the same guy on the "philosophy" of breath in voice over:

https://www.youtube.com/watch?v=BEJuA9TPMKs

Just wanted to share this as I found it useful and to the point.

Gauguin

[TriciaG]

Yeah, pretty much the advice we give as well:

- Breathing is natural. So unless you breathe like an asthmatic rhinoceros, don't worry about it. (I liken breaths in recordings to the squeak of guitar strings when an acoustic guitarist changes chords. The listener - or reader - doesn't notice it, unless someone points it out. Then they obsess over it. Ha!)

- Don't just delete them; it messes with your natural speech flow. Don't replace them with silence, or it'll sound worse because it doesn't have your ambient room noise.

- If you have to do something with them, reduce them using the Amplify effect. But really, unless you're that rhino, it's not worth the time.

[gauguin]

Totally agree but for beginners there are the odd occasions when the breathing is rather awkward and in the mic and re-recording seems to big of an effort (for example one problem I have is that I breath out heavily through my nose after I read a title/header).

Breathing is natural. I hear everybody does it these days!

[Cori]

The breath can be part of the character too. It might seem a little odd to say -- I've no idea if the difference is audible -- but I felt very different breathing while reading Dr John Watson, to Victor Frankenstein, for example.

I do quieten down some breaths, because o̶n̶ ̶t̶h̶e̶ ̶i̶n̶t̶e̶r̶n̶e̶t̶,̶ ̶n̶o̶-̶o̶n̶e̶ ̶k̶n̶o̶w̶s̶ ̶I̶'̶m̶ ̶a̶n̶ ̶a̶s̶t̶h̶m̶a̶t̶i̶c̶ ̶r̶h̶i̶n̶o̶c̶e̶r̶o̶s̶ I record relatively close to my mic, but as Tricia says, never ever replace with silence or room noise. Some professional audiobooks have had that done by some over-zealous editor, and they actually make me a little anxious while listening, because breathing is such a natural part of speech. I think maybe I end up holding my own breath, waiting. For myself, if it's a horrible breath even when quieted, I might actually paste a nice breath from somewhere else. This is why my recordings take approximately a million hours to edit, though, so it's not recommended to anyone who doesn't have strong anal-retentive tendencies.

And for what it's worth, I've had several chapters (even whole books) edited by other people, who I don't think touched the breathing ... and not a single piece of feedback (on YouTube, where people are pretty mean about any possible flaw) has EVER mentioned them. I probably just need to get over it!

[gauguin]

- Don't just delete them; it messes with your natural speech flow. Don't replace them with silence, or it'll sound worse because it doesn't have your ambient room noise.

- If you have to do something with them, reduce them using the Amplify effect. But really, unless you're that rhino, it's not worth the time.

The breath can be part of the character too. It might seem a little odd to say -- I've no idea if the difference is audible -- but I felt very different breathing while reading Dr John Watson, to Victor Frankenstein, for example.

Just for a little comparison fun here is a little audio I just did to illustrate the effect of breath and its treatment with audacity.
Notes:
- I deliberately exhaled before the start of the recording forcing me to breath in early on.
- I did not apply any other effects (e.g. noise reduction) to the audio and I have quite a noisy recording environment (HVAC)
- I read this sentence three times (can you guess what I read without googling )

This is what I did to the audio.
First sentence: I left the breath sound in raw.
Second sentence: I reduced the volume of breath with the amplify effect in audacity (select the breath section, then go to Effects - Amplify - and shift the slider to the left approx half way but you need to play around with the correct level).
Third sentence: I replaced the breath sound with room noise.

Now you can hear the differences and can judge for yourself what you like best. Clearly it depends on the case and text as to what will work best.

https://librivox.org/uploads/tests/breath_audacity.mp3

Greetings from Philadelphia,

Gauguin

[TriciaG]

For myself, if it's a horrible breath even when quieted, I might actually paste a nice breath from somewhere else.

Come to think of it, I've done that sometimes, too. I think I do this more than I try reducing the volume on bad breaths.

[Cori]

No googling ... Middlemarch! It does depend on the scenario, for sure ... though I'd not say any of those mid-sentence breaths would be noticeable in context.

[CliveCatterall]

As other have said, I find that a recording without *any* breath sounds is a bit weird.

I know you're not looking for tips on reducing the sounds of your breath, but on reducing sounds with audacity (and I'm not setting myself up as an expert). But here is what I have found anyway...

If you open your mouth fully when breathing it helps to reduce the noise. Easier to do that turning to the side (I never managed to learn that).

I also find that breathing as slowly as possible and using the whole of the gap between sentences makes the most difference. Don't pause in the gaps and then gasp quickly, or wait to start a paragraph and inhale at the last minute. Start to breathe in as soon as you finish the sentence. And don't leave breathing too late. Break long sentences: it's much easier to breathe quietly and quickly if you haven't completely emptied your lungs.

If you can't find a place to break the sentence sometimes you have to take little "sips" of breath to get you to the end of the sentence. It never sounds that good (no rhythm or structure to the sentence), but what can you do if the writer insists on 174 word sentences as in "Studies in he Art of Rat Catching"?

Writers. What can you do?

Clive

[gauguin]

If you open your mouth fully when breathing it helps to reduce the noise. Easier to do that turning to the side (I never managed to learn that).

I also find that breathing as slowly as possible and using the whole of the gap between sentences makes the most difference. Don't pause in the gaps and then gasp quickly, or wait to start a paragraph and inhale at the last minute. Start to breathe in as soon as you finish the sentence. And don't leave breathing too late. Break long sentences: it's much easier to breathe quietly and quickly if you haven't completely emptied your lungs.

If you can't find a place to break the sentence sometimes you have to take little "sips" of breath to get you to the end of the sentence. It never sounds that good (no rhythm or structure to the sentence), but what can you do if the writer insists on 174 word sentences as in "Studies in he Art of Rat Catching"?

Clive,

thank you for these great comments and suggestions. These are all great options to try to see which of those are the most natural to build in. I think for us hobbyists usually the more natural a technique the better the overall recording.

[msfry]

Hi Gauguin. Interesting discussion here. I've been recording now for almost 5 years. I used to de-amplify almost every deep breath, cut out every lip smack or tongue click, but after listening to so many audiobooks, I've come to appreciate recordings that don't sound like an automaton has read it. I love connecting with a reader's authenticity, and the brain adjusts, as it does in live speech. I still edit some of mine out, but not as much. I also have the problem of ear-pearcingly shrill "s's" on some days (not others), and those I do cut the peaks out of. I sometimes alter my pacing of an entire file with the Tempo Effect -1 or -2 (I tend to talk too fast), and sometimes I alter the tempo on particular phrases up to -6.

Seconding Cori, editing takes me way more time than recording. I listen thru headphones at my desk as I edit, then listen through the whole thing. Upload, wait a few days when I've distanced myself from the minutia of the project, then listen through my bluetooth while driving or cooking, and/or listen again thru my phone at night when I'm almost asleep (no earbuds), which gives a very different perspective.

[iBeScotty]

This is really interesting and I have just been experimenting this week with trying to tame breath sounds. Obviously, the occasional, distracting gasp can be edited but who would want to do that for everything in long-form narration?

My issue is that I usually use a little downward compression for peaks and overall sound consistency and then amplify to get the overall volume back up. This then also amplifies the breaths, unfortunately. Now, instead of straight amplification, though, I am finding that an upward expander (increasing the volume of the louder parts more than the softer parts) seems to help. I don't use Audacity, though, but the same procedure should be possible.

As an extreme test/example, I jumped rope for a minute (wow, am I out of shape) then jumped on the mic to say a phrase. I took the same phrase and duplicated it twice, adding a typical compression and amplification, then on the second duplicate, instead of amplification, using an expander.

Here it is at 26 seconds total and you can really tell how annoying the breath is in the second iteration:

https://librivox.org/uploads/tests/test_expander_scot.mp3

The measurements for each part:
Raw:
vol./loudness: 89.1 dB/-20.2 LUFS
range: 3.9 dB
peak: -2.4 dB

Compressed and amplified:
vol./loudness: 89.1 dB/-20.4 LUFS
range: 2.2 dB
peak: -4.9 dB

Compressed and expanded:
vol./loudness: 89.4 dB/-20.4 LUFS
range: 3.8 dB
peak: -4.4 dB

The expander took some fiddling and I am not experienced enough to really tell of the sound is decent but it seems to help.

[msfry]

Interesting! The Expander (third iteration) was far better no doubt, but I haven't the slightest idea what all those numbers mean or how to apply it to Audacity. Maybe one of our Audacity guru's will come along and explain it to me. Thanks for sharing.

[iBeScotty]

The numbers were just to show that the average volume was nearly the same after each adjustment and show how the peaks and dynamic range changed. The compression seemed to really squash the dynamics and the expander brought them back up.

[mightyfelix]

Pardon my ignorance, but your description of what the expander does just sounds like an anti-compressor to me. It seems like you're squishing the spikes and then undoing that by just stretching them back out again. What am I not getting here?

[iBeScotty]

Yes, it is like an anti-compressor (I am no expert for sure ) I guess but the benefit is that in these two stages, different thresholds can be used to massage where the compression and anticompression happen to give you the desired result, i.e. a lowering of peak levels and raising of overall level without raising low-level noise/breath.