De-Essing For Podcasts
A de-esser is a podcaster’s best friend. It reduces the sibilance (words with “sss” or “shh” sounds, like ship, chip, sip, zip, or jump) of phrases so that they don’t jump out in listeners’ ears. It is literally a de “sss”er. Once you start listening for sibilance, you’ll never be able to unhear it, and I apologize for that, but it is all in an effort to help you make a better podcast!
A de-esser is a combination of the last two things we looked at, an equalizer and a compressor, which makes it a type of multiband compressor, which we’ll get to eventually. Essentially it is operates like a compressor, but we can specify which frequency bands it looks at, like an EQ. As sibilance almost always occurs within the 3 kHz – 10 kHz (if not higher), de-essers looks at these frequencies specifically and compresses based on only the information in this range.
From the compressor side of things, we have a threshold, or the point at which our de-esser will start compressing the signal, and either a ratio or range. A ratio, like a compressor, will be the amount of dB needed to move the signal in this frequency range up by 1 dB, while a range is the most amount of gain reduction the de-esser will use to keep the signal at the threshold. The latter is more simple, but the former is common. That being said, with either method it is very hard to over compress the signal for podcasting purposes, and it becomes very evident when you do. So just trust your ears on this one.
A de-esser is a combination of the last two things we looked at, an equalizer and a compressor, which makes it a type of multiband compressor, which we’ll get to eventually. Essentially it is operates like a compressor, but we can specify which frequency bands it looks at, like an EQ. As sibilance almost always occurs within the 3 kHz – 10 kHz (if not higher), de-essers looks at these frequencies specifically and compresses based on only the information in this range.
From the compressor side of things, we have a threshold, or the point at which our de-esser will start compressing the signal, and either a ratio or range. A ratio, like a compressor, will be the amount of dB needed to move the signal in this frequency range up by 1 dB, while a range is the most amount of gain reduction the de-esser will use to keep the signal at the threshold. The latter is more simple, but the former is common. That being said, with either method it is very hard to over compress the signal for podcasting purposes, and it becomes very evident when you do. So just trust your ears on this one.
Here we’re breaking down the de-esser into its separate parts. On the left we have an EQ that is only looking at a very narrow bandwidth, around 7-10 kHz, and a compressor set to a low threshold (-32 dB), a very short attack (0.01 ms with a 10 ms lookahead to make sure we catch all the sibilance) and a very short release to make sure our compressor is only affecting the sibilance. Not that make up gain is off.
The most important setting to choose on a de-esser, however, is whether to use Split or Wide Band modes. Split-band mode literally splits the incoming audio into the frequencies with sibilance in them, and those without. In split band mode the de-esser will only compress the frequencies you’ve set it to look at, and leave all other frequencies untouched. This is handy in music production, where you might want to keep the airy 10-15 kHz sounds to make the singer sound more present.
Wide band, conversely, will compress the whole signal, whenever it detects sibilance. This can be very useful to podcasters with less expensive microphones, where sibilant words have a tendency to sound extra harsh across a wider spectrum than a de-esser would necessarily detect. That being said, you do have to dial in your settings more precisely, as this method will reduce your whole signal, meaning there is more chance of accidentally compressing non-sibilant words. To help you strike the right balance, here are some tips to keep in mind:
Keep your threshold low. I mean real low, around -40 to -20 dB. Because sibilance cuts through so much harsher than other sounds, we want to start compressing it as soon as we detect any on the track.
High Ratios. Again, real high ratios. 8:1 is a good starting place, but be prepared to go up to as much as 15:1 or higher. Now that you know what sibilance is, you’re never going to want to hear it again, and so a low threshold and high ratio is the way to go.
Trust your ears. If you’re still picking up too much sibilance, don’t be afraid to go beyond the settings we set out above. Your ears are the best indicator for both under and over de-essed voices. Does your guest all of a sudden sound super-nasally? Back off the ratio a little bit until his voice returns to normal.
Double up. If you’re having trouble with a particularly sibilant part, or the high end of a guest sounds very harsh after de-essing, try using two in tandem. I find first using a split band de-esser to tame sibilance, and then a gentler wide band to bring down the high end more naturally is a good way to tame harsh frequencies.
Compress, then De-ess. Keep your de-essers after your compressors in your signal chain. Because we’re using our compressors to bring up our volume, and de-essers to bring down problematic parts, it wouldn’t make sense to have our de-essers to bring down the volume only for the compressor to bring it back up. Use your head!
Airwindows DeBess is a great free de-esser, and Hornet Sybilla is my go-to paid de-esser
The most important setting to choose on a de-esser, however, is whether to use Split or Wide Band modes. Split-band mode literally splits the incoming audio into the frequencies with sibilance in them, and those without. In split band mode the de-esser will only compress the frequencies you’ve set it to look at, and leave all other frequencies untouched. This is handy in music production, where you might want to keep the airy 10-15 kHz sounds to make the singer sound more present.
Wide band, conversely, will compress the whole signal, whenever it detects sibilance. This can be very useful to podcasters with less expensive microphones, where sibilant words have a tendency to sound extra harsh across a wider spectrum than a de-esser would necessarily detect. That being said, you do have to dial in your settings more precisely, as this method will reduce your whole signal, meaning there is more chance of accidentally compressing non-sibilant words. To help you strike the right balance, here are some tips to keep in mind:
Keep your threshold low. I mean real low, around -40 to -20 dB. Because sibilance cuts through so much harsher than other sounds, we want to start compressing it as soon as we detect any on the track.
High Ratios. Again, real high ratios. 8:1 is a good starting place, but be prepared to go up to as much as 15:1 or higher. Now that you know what sibilance is, you’re never going to want to hear it again, and so a low threshold and high ratio is the way to go.
Trust your ears. If you’re still picking up too much sibilance, don’t be afraid to go beyond the settings we set out above. Your ears are the best indicator for both under and over de-essed voices. Does your guest all of a sudden sound super-nasally? Back off the ratio a little bit until his voice returns to normal.
Double up. If you’re having trouble with a particularly sibilant part, or the high end of a guest sounds very harsh after de-essing, try using two in tandem. I find first using a split band de-esser to tame sibilance, and then a gentler wide band to bring down the high end more naturally is a good way to tame harsh frequencies.
Compress, then De-ess. Keep your de-essers after your compressors in your signal chain. Because we’re using our compressors to bring up our volume, and de-essers to bring down problematic parts, it wouldn’t make sense to have our de-essers to bring down the volume only for the compressor to bring it back up. Use your head!
Airwindows DeBess is a great free de-esser, and Hornet Sybilla is my go-to paid de-esser