The DE555 de-esser plug-in combines advanced look-ahead dynamic range control technology for a truly transparent process for removing sibilants from vocals and dialog. But there are a few other uses for the DE555 too…
De-essing is the process of removing elements of a track referred to as sibilants, without adversely affecting the entire track. Examples of sibilants are the ‘sss’ and ‘tthh’ in a dialog or vocal track.
At the heart of a de-esser is an input key filter – a filter that isolates the sibilants from the rest of the audio signal. High pass and band pass filters are common in de-esser algorithms. The DE555 Key Filter section includes controls for filter frequency, type (both high pass and band pass options), and a listen mode to allow the auditioning of sibilants. The unique Focus control alters the shape of the filter to make it more sensitive to the audio signal frequency content close to the selected filter frequency. This enhances the selectivity of the filter, and hence the amount of isolation of sibilants.
The amount of de-essing is typically controlled by a range parameter. The DE555 gives an additional parameter called Ratio to adjust the gradation (or slope) of the de-essing curve. Just like a compressor, higher ratio control settings give stronger de-essing results, while lower ratio control settings provide more gentle (i.e. transparent) results.
Generally a fast attack and release time is the best choice when de-essing. The DE555 has a user controllable release time. Sometimes a slower release can make the de-esser sound more natural when used on vocals and dialog. Other de-essing applications discussed below can also benefit from the adjustable release time.
Finally, a high frequency only, or HF only option, is another common feature of de-essers. When a de-esser algorithm is triggered, it reduces the signal level of the entire audio signal, that is, all the audio signal frequencies evenly. A HF only mode causes the de-esser to affect only frequencies at or above the key filter frequency.
Why Not Use a Compressor?
Now that the basics of a de-esser have been covered, one might wonder why a simple compressor with a high-pass filter wouldn’t do the job just as easily. However upon further investigation, one will find this approach is not the best choice.
When de-essing, fast attack/release times are necessary for removing just the sibilance in a track. Unfortunately, fast compressor attack/release times can introduce distortion into the audio signal. But if attack and release times are not fast enough, the sibilant event may be missed (not reduced enough, if at all), or the rest of the dialog or vocal level is reduced too much. Furthermore, often the sibilance reduction does not sound natural if all signal frequencies are reduced during the sibilance event. On top of these problems, sibilance and vocal/dialog levels can vary considerably (even if pre-compressed), forcing the continuous adjustment (or automation) of a traditional compressor input threshold.
The DE555 de-esser addresses all these problems. Using technology from the award winning ML4000 mastering limiter and multi-band dynamics processor, the DE555 attack and release times are extremely fast, and yet do not introduce any distortion into the audio path. A HF Only mode is provided for instances when only reducing the high frequency signal content is needed during sibilance events. And finally, the input signal level is NOT used to trigger de-essing action. Instead, an advanced algorithm allows the DE555 to de-ess consistently at any signal level. For example, if a portion of the dialog level was at -24dB (a whisper), and later at -2dB (a shout), the DE555 would still only de-ess the amount given by its Range and Ratio controls. This advancement reduces time spent adjusting (or automating) threshold dependent controls found on compressors (and some other de-essers!).
Now that we’re all convinced the DE555 is the best thing since Pro Tools itself, let’s take a look at how to use the dang thing.
Vocal and Dialog De-essing
The main purpose of the DE555 de-esser is to reduce the amount of sibilance in dialog and vocal tracks. To do this, the user must first isolate the sibilance. Using the Key Filter section controls, sweep the Frequency control with the Listen button engaged to audition the filtered signal. Be sure to try the high pass and band pass filter options, as well as the Focus control amount (particularly when using the high pass filter). Usually dialog sibilance can be found between 3 and 10kHz, although sometimes an even higher filter frequency setting is useful.
Monitor the output levels in the output meter during when auditioning sibilance through the Key Filter. The goal is to see maximum signal level during the sibilant events, and a minimum signal level for the rest of the dialog/vocal. Be sure to have the Range control set to a reduction amount of 0dB so the de-esser is not triggering. You might wonder why you can still de-ess in the DE555 even when monitoring the filtered signal. Naturally I have an answer! Some folks find it useful to only monitor filtered sibilant events and evaluate sibilant reduction simultaneously. Try it for yourself. I also recommend green eggs and ham.
Once the sibilant events have been reasonably isolated, disengage the Listen button, and select the amount of de-essing using the Range control. Usually 6 to 9dB of de-essing is enough. Using a smaller Ratio (2:1 or so) can allow for a greater Range amount (12 to 15dB).
Finally experiment with the HF Only mode to see if high frequency only de-essing sounds more natural for your vocal or dialog track.
Reducing Sibilance in Background or Group Vocal Tracks
Just as the lead vocal track can require de-essing, the background vocals often have the same issues. Many singers belting out the same phrases can only further emphasize ‘essing’ events. And when the lead singer and background singers are not entirely in sync (or even when they are), the sibilance of the background vocals should not distract from the main vocal. More aggressive de-essing settings (higher Range, Ratio, and Release settings, 2x to 4x more than the settings on the lead vocal track) for background vocals allow the lead vocal to stand out more. The lead vocalist articulation comes out, while the background vocals remain in the…uh…background.
Like ‘essing’ events, a vocal or dialog track can contain the ‘inhales’ of the singer/speaker before each phrase is belted out. Often a gate, like the one on Channel G, is great for removing this unwanted audio. But sometimes this audio is to be included in the production. For example, if the actor in the movie takes a big breath before he kisses the girl goodnight, the audience will expect to hear that inhale. Use the DE555 to soften that sound if it proves to be a bit too loud.
Reducing Guitar Fret Squeaks
Vocals are not the only kind of audio that may contain unwanted high frequency events. Fret squeaks produced by the guitar player (electric or acoustic) can be reduced using a de-esser. The same tips apply as when de-essing a vocal or dialog track – too much can sound unnatural – so don’t overdo it.
Reverb Tail Manipulation
Got too much ‘sssssshhhhhh’ in that reverb tail? The DE555 comes to the rescue again. Reducing the high frequency portion of the reverberant signal can be accomplished using the HF Only mode of the DE555. Use the high pass filter mode in the Key Filter section with a frequency of 8kHz or higher. Apply a modest amount of reduction with a Range of 6 to 9dB, and a gentle Ratio of 2:1 to 3:1. A handy way to control the contour of a track that already has reverb ‘printed’ into it.
Clearly there are more uses for a de-esser than just the removal of dialog and vocal track sibilants. The DE555 is transparent enough to be used for other applications like those described in this article, and probably a few more! Thanks for reading another edition of Colin’s Corner.