If you're trying to make your voice-over sound better, be sure to remove background noise with gating, use subtractive equalization to remove problem frequencies, and introduce subtle compression to control dynamics. Additionally, ensure that you've achieved the right levels for the distributor you're using.
When you’re trying to process your dialogue, it’s a good idea to remove background noise. You can do this manually, or you can use a gate and carefully alter the range, ratio, and threshold to achieve just the right amount of attenuation without making it sound noticeable.
I find a great starting point to be the Nice and Tight vocal preset within the FabFilter Gate.
A good starting point for your EQ is to start with a low-latency linear phase setting, attenuate the low frequencies, boost 200Hz subtly, cut 300Hz subtly, boost a little of 500Hz to accentuate vowels, and remove a little 700Hz to remove nasal tones. Then boost 1.5kHz to 2.5kHz for clarity.
These won’t be the exact values for every dialogue recording but it’s a great start.
Next, if you’re finding that your ess sounds or sibilance is becoming harsh, use a de-esser to attenuate these frequencies whenever they’re loud enough. I like to use a split band, single vocal setting when working on dialogue, and then reduce the sibilance by about 3 to 4dB.
For a super clean de-esser, try the Weiss DS1 MK3 compressor.
When compressing, the goal should be dynamic control and not to add a particular timbre or distortion to the dialogue - in other words, try not to change the sound but instead cause transparent dynamic control. Use a softer knee, 3:1 ratio, and achieve about 3 to 4dB of attenuation.
Also, set a quicker attack and make the release time about 100ms.
To make your vocal sound full but not too boomy or warm, try a little tape saturation - this will add some higher-order harmonics. Or if you want warmer dialogue, try tube saturation to create a strong second-order harmonic causing amplification of the lower frequencies.
Avoid more than 1% THD or higher settings.
When picking out a limiter to use for your dialogue, try one with transient shaping - this way you can limit the signal yet still adjust the attack, release, and overall effect the limiting has on your transients. Some settings will sound more musical, whereas others sound more transparent.
If you don’t have one of these limiters, use a release time of 50ms, and be sure it doesn’t distort the signal.
When you’re mastering dialogue for YouTube, make sure that the overall dialogue and whatever background music you might be using comes out to roughly -14 LUFS. A limiter like the TC Electronic Brickwall makes this easier by allowing you to set the amplification based on the desired loudness.
Additionally, lower the ceiling of your limiter to -2dBTP to avoid changes to the amplitude once the signal is encoded.
When mastering dialogue for a movie that’ll be on Netflix, be sure that the overall dialogue and whatever background music you come out to no greater than -27 LUFS. Set your ceiling or output to no greater than -2.2dBTP - since if the signal goes over -2dBTP, it won’t be accepted.
Whereas mastering for streaming is more of a recommendation, these metrics are more-or-less rules set by Netflix and other video distributors.
If you want to use oversampling; however, the plugins you’re using don’t offer it, you can use the Metaplugin to house all of your processing, and to increase the sampling rate for all plugins. This will reduce aliasing distortion, resulting in a cleaner and smoother high-frequency range.
Additionally, you can use the FIR setting to alter the timbre of the overall processing, which can be helpful in the vocal sounds a little too harsh.
