Hi guys, I recently recorded a 30 second advertisement and it sounds absolutely great. I’d love to learn more about audacity, but can’t because my machinima is on a dealine.
Thing is, there is few things here and there in the sound that could be made better. I have no idea how to use audacity other than one thing I did.
The one effect->noise removal->get noise profile remove noise thing.
With that being said, if someone experienced could please do the following…
Please see if you can somehow smooth the very small studdering in the word “become”.
Please remove any noise that is not my voice. (momentary silence is fine)
and anything else to spice it up that would make it better.
Obviously I don’t want to post my audio for the whole world to see so please pm me if you are very experienced and willing to do it.
I cleaned it up a little in Audacity and applied reverb, (Freeverb to be precise).
I used super (c) to produce the MP3 version, (MP3s made by Audacity are noticeably noisy in silent bits).
Really? That’s not happening here. Which version of Lame are you using? (there was a release of Lame for Audacity that was noisy in some situations but it should now be fixed).
An MP3 of total silence should be total silence, and that is exactly what you get from Lame-for-Audacity.
Q) So where’s that noise coming from?
A) Dither.
Audacity does all “mixing” in 32 bit and that includes Exporting.
When audio is encoded as MP3 the bit depth is dropped to 16bit and is subject to the Audacity dither settings (Edit menu > Preferences > Quality).
There are a few ways that you can avoid the dither noise:
Export as 32 bit WAV and encode using an external encoder that does not apply dither.
Temporarily switch off dither in Audacity Preferences,
Set dither to “Rectangle”
Which is the “best” solution?
Option 1 is a good solution provided that the external encoder is of equivalent quality as Lame, though it is perhaps a bit inconvenient.
Option 2 is a good solution when exporting a single 16 bit track - in this case there is no advantage to applying dither.
Option 3 is possibly the most convenient solution as it can be set once and left. It still has the advantages of using dither when downsampling 32bit tracks and/or mixing multiple tracks during Export, though the subjective effectiveness of rectangle dither is possibly not quite as good as triangle or shaped on non-silence.
Option 4 - ignore the noise and use triangle or shaped dither. In music recordings it is rare that there is absolute silence, and total silence is about the only time that the dither noise is evident.
Q) Would it be better if Audacity did not apply dither when Exporting a single 16 bit track?
A) Probably yes - though it is a “special case” that has been discussed before and the developers seem reluctant to change it.
