It's only after this sort of processing that I'll start cleaning up the recording and dealing with the gaps of silence between spoken passages. A noise gate (or an off‑line 'strip‑silence' function) can be used to automatically mute sections that fall below a certain level. However, if not used carefully, these tools will clip the 'T's and 'P's off the ends of words, and shut out natural breathing sounds. Worse, if your recording is noisy, a noise gate will actually draw attention to the problem, since your client will be able to hear the difference between room tone and absolute muting. Another trick is to use a downward expander to reduce the noise floor of quiet sections, rather than cut out the noise completely.
Why the difference in approach? Lots of my e‑learning and corporate jobs are done for production houses that will be adding other voices to the mix. Ironically, if I record my voice via a great mic and preamp, and add some nice processing, it can end up standing out from the other voices, drawing attention to itself. The production house will see this as me causing them a problem, even though my recording sounds better than the others!
“Bunny Studio Voice has a very efficient process to help us identify the right voice actors that best meet each of our video’s needs. Their platform keeps us informed throughout the process, from the actor accepting the project all the way through when the recording is complete. I highly recommend them as a reliable, high quality and cost-effective voice over provider”
Next, run these files through any standard EQ high‑pass and low‑pass filters you use, zoom in on the waveform, and select a one‑second strip of silence from each recording. Save these strips as separate files. For instance, on my system, a one‑second strip of silence recorded at 16‑bit, 44.1kHz on a Neumann TLM103 would have this name: room_tone_neumann_tlm_103_44100_16_one_second.wav. Ideally, the noise floor of each strip will be lower than ‑60dB. Now, when you have to remove a bump or a squeak between words, you can paste in a piece of these strips instead of muting, which would draw attention to the edit.
Finally, as you complete more work, you can end up with a lot of different files from your projects, so good 'housekeeping' of your sessions and recordings is a must. Glancing at the filename simone_kliass_avast_virus_database_neumann_tlm_103_original.wav, I can easily tell that this is the original recording of my wife's "Your virus database has been updated” message in Portuguese for Avast! on a Neumann TLM103 mic. It doesn't matter what your system is, as long as it works and you can stick to it.
Even if you're mixing yourself, it's better to add EQ and compression only when you can hear how it will sit with the soundtrack or special effects. Ultimately, it comes down to knowing what your voice will be used for and making a judgement. For instance, if I'm recording a tagline for a TV or radio advert, I'll generally run a mic into a nice preamp (where I might add slight tube warmth, and subtle EQ or compression, just to give the recording a bit of 'body'), and I may do a little de‑essing using a plug‑in when performing any edits. However, for e‑learning jobs, corporate videos and so on, I'll tend not to add any creative processing while recording or editing, and will only use the de‑esser ever so slightly.