Q It Up: Do you mic and process female VOs differently than male VOs? Do you have them work the mic differently? Are your EQ settings different? Compressor/processor settings? What do you do to key into the warmth of the female voice without compromising her own natural flavors, tones and frequency windows? Do you use a different mic for female VO? What works for you?
Judy Smith, Wysox, PA: I apply different compression to every voice, male or female.
First I’ll look for any nasty peaks which are completely abnormal from the rest of the waveform, and try to control them either by hard limiting, or just tweaking the amplification in that particular vowel or consonant.
Then, I normalize the file to 70%, and apply compression at 3:1, maybe 4:1.
After mixdown, I apply a hard limiter again, to knock off any peaks that may appear out of the ordinary, but I don’t overdo it. I like to see some dynamics in my final waveform.
I generally don’t use any EQ on voiceovers, unless I’m trying to achieve a particular effect for a concert spot, or something to that effect, where I’m trying to carve out a spot in the mix.
However, if I’m recording a singer, I tend to use some parametric EQ, because certain frequencies tend to get a little harsh as a singer pushes their voice.
Mark Nelson, Sioux Falls, SD: I don’t necessarily process male and female voices differently, but I base processing tweaks on the character of the voice. Mellow, pleasant, warm sounding voices get a minus 18 threshold dynamic pre-processing, and then are amplified before mixing.
Louder voices with inconsistent volume or brasher, more grating voices get a minus 12dB dynamics pre-processing and then are amplified.
Joshua Mackey, MackeyVoiceTalent.com: I don’t process female voiceovers differently than male voiceovers. Processing tends to have much less to do with gender and much more to do with the person’s tone, microphone, recording atmosphere and pre-processing. I have created a preset for a specific person’s voice and found it to work for another person’s voice, even a different gender. I process ALL voiceovers slightly differently with regards to EQ, but compression is generally the same for all voiceovers.
The first thing I do to a voiceover is normalize it to 97% - sometimes this includes manually decreasing large spikes if the voiceover is highly dynamic.
I then add EQ to brighten the vocal, and gender rarely has anything to do with it. It’s typically more about the room noise, the microphone, the artist’s overall tone, and any pre-processing that took place (which I highly discourage for voiceover artists - send audio raw!).
I find that killing some of the low end, boosting some of the high end and creating a mid-drop USUALLY takes place with varying degree. Once I get a nice, clean voiceover, I add it to the mix.
From there I tweak the EQ for all individual elements (voiceover, music, sfx) so they all blend nicely together. I also make sure there are no transient spikes. I mix that down and perform the final hard limiting, fattening the audio and normalizing it -.2 dB.
If I’ve done everything else properly, the mix sounds good. If not, I go back to the individual elements and tweak the settings.
I rarely alter the EQ for a final mix.
John Pallarino, Greenville, SC: All voiceovers have their unique signature and a lot of it has to do with the mic chain that the talent is using.
If I get a good clean vocal I start with looking at the frequency analysis to see if I see anything that sticks out. Most of the time the first thing I see whether if its male or female is the low end which always gets a nice high filter pass at 60-100hz. Most of that is room noise and proximity of the talent talking in to the mic.
Females usually will a get a nice boost in the 160 range if they are thin or talking too far from the mic.
Males tend to get a cut around the 160 range if they sound boomy. Then I move to the mid-range section which again has a lot to do with the mic chain. Males tend to have a natural boost around 2.5 and females around 3.5 which may or may not get a cut depending on the style of the read.
The high end is where most of my boosting or cutting will occur along with some de-essing if necessary.
After I play with the EQ, I add some compression to tighten things up and smooth out any EQ I may have added.
Roy Cunningham roycunnicngham.com: By their nature, the characteristic of male and female voice is different enough to warrant separate processing of each. For me resonance is key. (A bit of research enables one to understand the anatomy of both and perhaps better understand the reasons why the two are indeed so different.)
I usually start to fine tune male vocals beginning from approximately 380hz down, then I go to 5k and above. I usually start to fine tune female vocals beginning from approximately 400 to around 2200Hz, then I place them into the mix.
“Making a hole” for vocals in any mix is old hat for almost any producer - discovering where the “hole” begins and ends precisely is a different pursuit.
The next process for both is to discover the dominant resonance by using a filter, in my case Cakewalk’s “Bitfilter” set on 100% “WET” to sweep for the frequency range that “pops” out of the mix.
For me, all productions - whether original music, broadcast audio for commercials or jingles - narratives and so forth rely on a balance of vocal to music - for me they’re all a song!
My introduction into broadcast production was rather crude, punch the VO… mix the music way down. As my pursuit to get better continues, I really appreciate hearing all parts of a mix equally clearly.
Sweeping the core frequency of a vocal in each production is crucial to the unique sound of the talent. There are no set rules for me in this. Once I find the frequency that “pops out” of the mix, I go back to further fine tune the voice by adding “air” from 5k up in a male voice using a high shelf with a boost from 0.6 to 0.8db, and adding “air” for a female voice from ~2900 to 8k using a bell curve with “Q” set to the individual case.
As of late, Eventide’s “Ultra Channel” has really helped clarify vocals. It contains so much control over the signal, and so many presets from which to start the fine tuning, it has made a big difference in how my productions sound. It’s also very good in the finalisation of an entire mix.
I also work with Native Instruments’ “Supercharger GT” and various emulations of the LA2, Fairchild and Urei 1176 and a variety of the tools in SONAR. I use as many different EQ, compression, and delay tools as I have projects. No one voice sounds the same as another.
I work hard to avoid a standard studio rule while maintaining an overall quality one might describe as “my sound”.
Andrew “Spaldo” Fry, Hardy Audio, Ballarat, Victoria: Processing of voice over tracks can be a very complex task, one which I spend my days giving a lot of attention.
We record and produce our own talent in house at our own studios, which as any recording engineer will appreciate, gives you a lot of control over the multitude of voices that you have working in front of the mic. And it goes without saying that each voice has its own distinct characteristics, often being highlighted by the type/brand of mic you are using.
This can be a bigger problem when using external talent i.e., talent operating from their own home or other facilities, where desk settings and room acoustics come into play. As a result, some of the VO tracks we receive require EQ treatment or processing to remove extraneous noise such as ground noise, interference or in one case recently, traffic and bird sounds!
iZotope RX2 is an extremely effective tool in achieving this end -- we can isolate problem areas in audio and correct where required, without loss of audio quality. I also prefer to de-breath where necessary and fix any problems in Sound Forge before transferring audio across to be mixed into Pro Tools, as I personally like to keep any mixing processes to a minimum.
Mixing in Pro Tools is a different situation. As I mentioned earlier, each voice has its individual characteristics which are really highlighted when using compression, aural enhancers and mastering tools.
In the case of female versus male VO, for example, I personally tend to find sibilance and midrange clipping a problem with females, especially with harder sell reads. I like to plug in a compressor on the vocal channel, such as a Dyn 3 with fairly soft settings (14db gain and -19db threshold) and plug in a parametric EQ before the compressor to isolate any problem areas. Some females have a tendency to clip vowels at the start of particular words, so a soft attack and quick release on the comp often works well to counter this.
Male voices tend to have an inherently broader range – from low and ‘bassy’ to shouty and harsh – and so any processing required tends to be more drastic. For example, less compression is required on a soft sell male whereas I’ll use harder knee, quick release settings with a much lower threshold for a harder sell. Again, I’ll use an EQ in the chain to correct any problems – i.e., too much bottom end or sibilance.
Final mastering is usually done in Sound Forge, where I can use Express Dynamics where necessary to fix overall loudness of a mixed track and prepare the track to meet radio and television requirements. Any glitches can be corrected or further EQ can be added where necessary.
But here I should point out that listening is key. As a former employer pointed out to me, ‘we are audio engineers, not desktop publishers’. That is, don’t rely on the tools to automatically take care of the work. Always set aside time to review the work you have done, and correct or remix where you need to.
Andrew Frame, BAFSoundWorks.com: In a broad answer, yes. We don’t do anything different for either gender. The heavy lifting is in the pre-production stage.
We ask all talent, regardless of experience or preference to please send us raw audio. No mic processing of any kind. Straight audio from the microphone to the recorder with no adjustments of any kind, save peak limiting to eliminate transients.
The first thing is to get a signal level up to -2dB peaks, my preferred working level. This is normally a couple passes with a normalizer, limiter, and occasional waveform reshaping.
I save at this point to a 32-bit. single-channel mono WAV with a different filename, so I always have the original, raw voiceover to go back to if something gets kludged.
A gate is applied depending on the noise floor. I find I get better results on a noisy voiceover by making multiple passes with a lower threshold, than one pass with a severe threshold. For example, two or three passes with a gate set for -45dB, than one pass set for -36dB.
For EQ, I mostly deal with the boost sitting around 100Hz on a lot of voiceovers. I favor subtractive EQ when possible, so I’ll put a fat parametric notch around that point, tuning the width and depth for the specific voice. Sometimes resonance is as low as 80Hz, or as high as 120Hz. If it’s a regular talent, I’ll save those filter settings.
The last pre-production step is to go through, remove all the breaths and artifacts, then adjust the word spacing if needed for a more consistent, smoother flow.
After a mix, we’ll use either a light or heavy companding depending on the desired outcome. Car spots, for example, are crunched heavier than a ballet store, though, not by much. Dry voice tracks destined for television stations, get either no companding, or a very light pass depending on the station production manager’s wishes.
Ronnie Kohrt [ronkohrt[at]gmail.com], WIXM, WWMP, WIFY, WCAT, WTWK: Yes, I do process them differently and in fact, I process ALL my voices individually with different settings based on the sound. Everything is situational for me. Everyone’s voice is different in so many ways! With great producing, we take these great voices, and make them even better! This goes for male or female, child or adult.
While a great producer will always have his secrets :) I always go for a warm, sweet sound with females. That said, it is a situation game based on clients. I hold certain ‘defaults’ that we look to preview after the initial voicing and modify from there based on how it sounds to the ear. It is definitely a new realm to dive into as a producer and an exciting one at that! It tests your versatility to know different EQ settings, how software and mics effect a female voice and how to coach a female voice talent when voicing for clients, spots or imaging. We’re lucky to have multiple female voices in-house as well as ready at the arm if needed for imaging on my stations.
Steve Mitchell [skm123[at]roadrunner.com]: Do you mic and process female VOs differently than male VOs? Yes. Do you have them work the mic differently? No. I have everybody work at least 8 inches away, even with a blast screen. Are your EQ settings different? Compressor/processor settings? Yes, dependent on the type of read. What do you do to key into the warmth of the female voice without compromising her own natural flavors, tones and frequency windows? Nothing - if she has a bedroom voice, so much the better. Do you use a different mic for female VO? What works for you? Yes. I like to use an RCA 77-d for the female talent and either an RE27, or an MXL V69 tube for the guys’, depending on their timber.
Dave Cox [dave[at]2dogsdigital.com], 2Dogs, Moline, IL: Yes, we record male and female very differently. The “go to” is a Neumann-47fet into an Avalon 737sp, for females. We cut a little at 1.2k and depending on what the final product will be, different release times for the compressor. The fatter sounding Neumann and tube pre/comp help to tame any harshness. De-Essing and noise… all post. Sibilance is the enemy and Waves De-Esser set narrow at 8k (give or take a little) is the cure.
Rick Harrington [production[at]cjbq.com]: Typically, I don’t make major changes to the voiceover talent, male or female. I will use some slight equalization to give it a boost, just very slightly. I also use a little bit of compression to tweak the levels overall, bringing the low levels up, and the higher levels down. But for the most part, I prefer to keep the natural sound of her voice as best as possible. Two reasons: (1) they don’t sound processed, and (2) they sound like they’re talking to their buddy. It’s a natural response the listener can easily identify with, and if they sound over-processed/EQ’d, etc., it sounds unnatural and the listener can easily tune out.
Von Coffman [vcoffman[at]intermountainradio.com], Bonneville Radio, Salt Lake City, UT: Wow, what a great question. I really had to stop and think about it… do I mic them differently? Do they work it differently? What about EQ and compression?
I have three pro female voices that do a bulk of the female reads here at Bonneville. They have completely different sounding voices, so each has to be processed differently… but not really all that much. There have been many discussions about different mics for female vs. male. Here we use a Neumann TLM 103. We at one time had three different mics in the studio, a Rode Broadcaster, a Shure SM5B and the Neumann. The Shure was a little too tubby, the Rode to tinny, but the Neumann seemed to be a great middle ground (just right).
The processor I use is an Aphex 230. It has the Big Bottom and the Aural Exciter built into it. I use both, but not too much, for female or male. If the Neumann is set up on a somewhat neutral footing with the Aphex, I’m getting is a more polished natural flavor. Then, I can play all I want in post to make her sound bigger, smaller… whatever.
I don’t use a lot of compression inside the mix, but I do pay close attention to the EQ. I dump 60hz, but pay close attention around 100hz. My cardinal rule is if you shine the bottom you have to shine the top. The more lower end you add, the more air (upper end). I add usually around 13.5 kHz. I am also a big believer in the notch filter on music and FX (again EQ). On my final mix I do use a Graphic EQ, a Peak Limiter and a Normalizer on the master bus to finish it out.
Howard Kazuska [kzooradioworks[at]shaw.ca]: In the “old days” I would use a AKG C214 mic which really seemed to capture the female voice nice and clean/clear without too much low-end so I wouldn’t have to EQ the voice too much. I would do several takes (with the “tape” rolling) to see how the female was working the mic in the studio. If she was too close or too far then we would adjust her position. Standing always works best for females and males.
As well, I would adjust the windscreen so that it was angled instead of being even with the mic., unless there were lots of “Ps” in the spot. I would record the voice without any EQ or effects and then add about 10-15% compression/limiter in post-production with some EQ depending how the voice sat against music and sfx.
The acoustics of the mic studio is worth its weight in gold and saves having to spend time on EQ. Plus if there was a revision weeks later, it’s a lot easier to match the revised line and drop it in without it sounding like an edit (unless there was more copy added, which always mucks up the pace/read).
A couple of other good mics for females with deeper voices are the Shure KSM27 and my favourite for just about any voice - male or female - is the Rode NTK.
As a productions house doing radio spots for ad agencies, I found when keeping my spots as pure as possible that they cut thru the clutter as each radio station already has heavy compression/limiters on their final transmission chain.
Chris Diestler [chrisd[at]santafe.com] Hutton Broadcasting, Santa Fe, NM: When I’m doing the recording of female voices versus male voices, it really depends on their vocal frequency range. If they’re a soprano, I’m likely to avoid doing a hard limit and just stick with a soft limit compression. ometimes I’ll run a hiss reduction too if they’re really high up there. For EQ I might bump up the bass slightly. If the female has a voice in the lower register, I will typically process it much like a male voice. When selecting backgrounds and fx I may try to offset high-frequency voices with low-frequency sounds so they stick out more easily. The microphones we use are most often the Shure SM7, especially for higher voices, but if someone’s got a really deep voice I will occasionally use the Electro-Voice RE20. If only I could afford Sennheisers! We run clean signals to the editing suite and do the processing from there.
Dave Savage [DaveSavage[at]ClearChannel.com], Creative Services Group, Clear Channel Media + Entertainment, Atlanta, Georgia: I don’t treat VOs differently based on the gender of the talent other than if a male has a really deep voice I will roll off some of the low end, it makes it a little crisper but you can still tell he has a deep voice. I usually don’t do anything different to a female voice.
I do treat VOs differently based on the part they have in the copy. If they are two people having a conversation I will roll off all the low end, like everything below 175 to 200 Hz. It gives it a less full sound and really sets it apart from the announcer VO. If it’s an announcer VO, I’ll add a little high end and many times use the “Broadcast” preset on the Multiband Compressor in Adobe CS6. If the bed is fighting with the vocal, I’ll run a frequency analysis on the VO to see the most prominent frequency in that person’s read then lower those frequencies by a few db on the bed.
PART 2 NEXT MONTH!