Voice Post Production

Hi Forum,

As most of the beginners coming here I could really use some help with my recordings. I am trying to start recording reading short stories and post them on YT. The important part is that it is not going to be signing – just me reading text, maybe adding some comments at the beginning and end. I have tried to master my vocal as good as I can but I have this feeling that I am doing it all wrong. Part of it is probably working with my own voice makes me feel weird. However, I am not so sure about the plugins and setting I am using. At the moment I am using following settings, this is data raw from the chain I am trying to do:

Normalize: ApplyGain=yes RemoveDcOffset=yes Level=-1.000000 StereoIndependent=no
Compressor: Threshold=-15.000000 NoiseFloor=-40.000000 Ratio=2.500000 AttackTime=0.300000 DecayTime=1.000000 Normalize=no UsePeak=no
Bass and Treble: Bass=3.000000 Treble=8.000000 Level=1.000000 Normalize=no
Normalize: ApplyGain=yes RemoveDcOffset=yes Level=-1.000000 StereoIndependent=no
NoiseGate: funct=Gate stlink=“Link Stereo Tracks” hpf=No freq=0.000000 level-red=-12.000000 thresh=-35.200000 attack=250.000000

So basically, I normalize to -1, then compress, add bass and treble (+3 and +8), than normalize to -1 again, run NoiseGate with at -35db to get rid of any possible background sounds left in there and export. I have tried to manually equalize the track but I have failed miserably so I am using ‘bass and treble boost’ plugin. I have also included before and after sound tests. This is just a headlines from BBC News so please ignore the subject - this was the only one not talking about war or economy. Also, I am not a native English speaker so please ignore the accent.

My goal is simple – get good crisp sound without sounding weird and unnatural. I do understand that I am not a professional announcer and do not have particularly good voice. Also recordings are done in my bedroom with USB microphone which might also be an issue. I use Audio Technica AT2020USB+. I have tried to upload .wav files but they were over 1MB. I have files in MP3 will be good enought.

Thanks for your help.


The original “test_before.mp3” sounds perfectly clear to me.
My suggestion would be to add two Nyquist plugins to your copy of Audacity …

Limiter.ny ( to replace the native compressor effect, whose release-time is way too slow IMO )
and DeEsser.ny ( which reduces sibilance which will become more obvious after using a compressor/limiter)

So my suggested chain would be …

HighPassFilter...: rolloff="12 dB" frequency=100.000000
Normalize: ApplyGain=yes RemoveDcOffset=yes Level=-1.000000 StereoIndependent=no
Limiter...: thresh=-10.000000 hold=10.000000 makeup=1
Limiter...: thresh=-5.000000 hold=10.000000 makeup=1
De-Esser: control-action-choice="Apply changes" control-absolute-threshold=-18.000000 control-step-dur-ms=5.000000 control-frequency-bound1=3000.000000 control-frequency-bound2=10000.000000 control-n-bands=5 control-crossfade-interval=5.000000

T-Chain.txt (493 Bytes)
DeClicking could be incorporated into the chain, (another additional plugin), but including that increase the processing-time to a multiple of the playback-time, ( at least 3x).

Both the DeClicker and DeEsser take additional time; but when I started using them, the 15-20 minutes they added to my chain prompted me to adjust my routine and find something else productive to do. The time saved by them (especially the DeClicker–which I used to do by hand, one at a time) more than made up for the time I had to do something else with! :slight_smile: