12

I am helping a friend (very talented aspiring singer songwriter in the folk music genre’) by making some free demo recordings of her music to promote her talent.

She has a nice voice but it’s on the higher end of the spectrum.

When instruments (particularly violin) are playing while she is singing, it’s hard to distinguish the vocals from the instruments on the recording and you can’t understand the lyrics she is singing.

Obviously I could lower the volume of all the instruments but I think the balance in volume between instruments and vocals is not really the problem. I suspect it’s more of a situation where certain instruments ( particularly the violin) occupy the same tonal frequency range as her vocal.

What can I do either in the recording process or post production or mastering process to better separate or differentiate the vocals from the instrumentation so that her voice comes through prominently in the mix? Should I adjust the EQ settings on the vocals or the instruments (or perhaps both – in opposite directions)? Should I use compression on the vocals or instruments? Should I try a different microphone on the vocals? Any suggestions are worth trying.

My question is specifically about separating vocals from instrumentation so that the lyrics can be understood. The input source for vocals with be a microphone and could either be a dynamic or condenser microphone. Some instruments might be recorded with a microphone while some could be direct input through an instrument cable.

It will likely be a trial and error process- and ultimately the best results may come from combining several different solutions - but any advice on some things I might try first would be greatly appreciated.

  • 2
    There are probably some tricks you can try, but maybe they need to work on the arrangement. How easy is it to distinguish the voice when they're playing, is it really a problem with the recording? – biggvsdiccvs Mar 16 '15 at 7:10
  • @biggvsdiccvs I think it may be mostly a problem with the arrangement of the instrumentation. I think violins probably should never compete with a female singer - both trying to sing at the same time. The violin is winning. What can I do to help the singer gain the upper hand? Kill the violin? That's what I'm thinking. Other ideas appreciated. – Rockin Cowboy Mar 16 '15 at 17:07
  • I was was also going to suggest cutting high frequencies on the violin, but see Todd Wilcox's excellent detailed answer. The only other thing I can think of right now is maybe not to record the violin at all with the vocals and add it to the mix later, possibly overdubbing it, as he suggested. – biggvsdiccvs Mar 16 '15 at 20:36
  • @biggvsdiccvs: cutting the violin's treble is actually likely counterproductive: it might just make the violin dull and the voice, in comparison, appear even thinner. Better cut the violin midrange, or just lower its level and perhaps even boost the treble a bit again. – leftaroundabout Mar 18 '15 at 2:03
15

Well this is a big question, so let me start with a big answer: http://www.amazon.com/Mixing-Secrets-Small-Studio-Senior/dp/0240815807/ref=sr_1_1?s=books&ie=UTF8&qid=1426508544&sr=1-1&keywords=mixing+secrets

Now for the smaller answer more approriate to this format.

  1. A good mix starts with the arrangement before you even record. If the violins are drowning out the more important vocal part, don't play (or mute) the violins at that part.
  2. If you have to have the violins there, make the vocals louder or the violins quieter or both. If you want the relative levels to be different in different sections, well that's what automation is for. Or you can copy the violin recording to a different track, set the different volume levels to the two copies and then mute and unmute at appropriate times.
  3. If level setting isn't do enough, then it's time to EQ. Cutting EQ is easy. Try cutting the high frequencies of the violin, or cut with a variable filter and then sweep it down in freqency until it starts to duck behind the vocals. Boosting EQ is much harder to do well and takes lots of practice. You could try boosting some part of the vocals but it can make them sound worse and not better.
  4. You need more? Let's get 3D. Make the violins go towards the back of the mix by giving them more reverb with a shorter pre-delay (0 to 30 ms or so). If you want any verb on the vocals to sweeten them, make sure they have a longer pre-delay than the violins (50 to 500 ms or even up to 2000 ms if it sounds good). If you're trying to send something far away, it really helps to cut the highs as well as add reverb. Natural air space cuts highs, so in real life highs roll off with distance.
  5. I skipped right over 2D didn't I? Well panning is an art. Sometimes it can sound really messed up. Modern style is that the most important things are always panned center (more or less). So leave the vocals alone. Throw the violins off to one side or the other, but don't pan anything hard left or hard right unless you are doing some special effect. If you have more than one violin track then pan separate tracks to the sides, which can also make the violins surround the listener a bit which is cool.

Yes, using different microphones on different instruments makes a difference, but what you're really doing is doing these mixing tricks before you even mix. Microphone and preamp choice, mic distance and placement, and all kinds of other recording choices are basically creative ways of doing levels, EQ, reverb, and panning during the recording process instead of during the mix. To that end, you want to use your best and brightest gear on the vocal track, mic the singer on-axis with a fairly short distance (4" - 12"), gain stage appropriately, etc. The goal is the get all the possible sound you can from the singer so you have it if you need it. For stuff that is supposed to be in the background, go the opposite way. Use an older mic with less frequency response, put it farther way, point it off axis, etc.

One more trick that's often used on backing vocals: Double track important parts that you still don't want to dominate the sound. That means perform and record the same part twice and play them both back in the mix. You'll want each track to be 3 - 6 dB lower in the mix when you double them up. You can pan them wide or keep them together to keep them smaller sounding. Reverb and EQ as above, etc. Basically the phase differences in the performances will reduce the clarity and dominance of those parts.

Have fun!

  • Good point about the violins. See my comment on user19443 answer. – Rockin Cowboy Mar 16 '15 at 17:04
  • From reading all of your comments and the other answers, I wonder if double-tracking the violin would be a good place to start. Also back the mic off the violins, get some of the room in the mic, point it off axis, and if you have a dynamic like an SM-57 or MD-421 around you might use that instead of a really responsive condensor to smooth out the violin a bit. Make a few takes and listen to it. Also record the vocal first if you can so you can compare where the violin takes sit compared to the existing vocal takes. – Todd Wilcox Mar 16 '15 at 18:33
  • I don't think you would have to worry about violins being "too loud"; I don't recall a situation ever where the violins needed to be softer. Real instruments are very different than synthetic ones. Now Brass I could see being a problem. – jjmusicnotes Mar 18 '15 at 3:04
  • Also, I wouldn't recommend cutting the high frequencies from the violins, they're pretty high instruments to begin with, so you'd be robbing their sound. All you really need to do is just emphasize certain frequencies to give the violins and the voice each their own shape. For example, emphasize the frequency of the tonic key for the voice and emphasize the frequency for the dominant for the violin. This is of course very rudimentary, but you get the idea. – jjmusicnotes Mar 18 '15 at 3:16
  • @jjmusicnotes: Really what we're dealing with is not a violin (or singer) itself, but a recording of a violin. Once it's been miked and recorded we can can give it as much or as little power bandwidth as we want, which will then cause more or less of a sensation of loudness when we convert the electronic signal back to an acoustic one. One way to look at the art of mixing is how to hand out the power bandwidth effectively to create the auditory experience we want for the listener. – Todd Wilcox Mar 18 '15 at 11:58
3

You can try through several mics and see what difference they make in the presence of the singer. Working with equalizers, however, results in "unnatural" sound. While microphones also influence the frequency response, they do it in a pretty consistent manner.

But it sounds like what you are basically trying to do here is putting lipstick on a pig. A violin has the same range as female voices (starting pretty much identical to an alto and covering soprano). It has a continuous tone, a rich overtone spectrum coming from its sawtooth-similar basic string action and a resonant body sized similar to vocal structures. It is one of the most popular instruments exactly because it is reminiscent in its tonal qualities of human voice and thus able to evoke a similarly emotional response.

As a result, you don't want to have a violin compete in the melodic space with a female singer. It's a bit different for a whole violin section: as a texture, you can merge it better with the background and it will also be somewhat more amenable to equalization without sounding weird.

Of course, there are exceptions: the "Agnus Dei" from Bach's B minor mass has an alto singer and a violin as solo instruments. However, you'll notice that their solistic phrases are mostly kept separate, with one having long notes or a pause where the other has more complex phrases. When they are indeed parallel, the vocal part tends to be at more dominant pitches and the violin tends to copy the vocal movements and actions and acts more as harmonization.

It also helps that those two instruments are the only ones competing at that pitch range: other than that, there is just a bass. So one has a better chance at hearing both of them separately, basically the essence of listening to baroque polyphony. But that's not the kind of continuous skill/attention expected from a listener to popular music.

  • 1
    I agree with what you said about the violin competing with the vocals. So one thing we can do is not play the violin while she is singing. I was not the musical genius who created the arrangement. I'm just trying to make a recording of what another artist thought should be played and sung. But I don't disagree that the arrangement (having the violin play at same time lyrics are sung) is the biggest part of the problem. If I can't solve it with other solutions in the mixing, I will suggest changing the arrangement - or I will just pull the violin way down whenever she is singing. – Rockin Cowboy Mar 16 '15 at 17:03
  • @RockinCowboy: If you know how to do sidechain compression you could use that to automatically pull down the violin while the singer is singing. With the right settings no one will even know it's happening. – Todd Wilcox Mar 18 '15 at 12:03
3

Are you recording "live" or building up the song track by track (perhaps in the mistaken idea that that's the "professional" way to do it?) Can the violin player hear the singer while playing her part (and is she LISTENING?) When good musicians are playing sensitively, very little "mixing" will be needed.

A real stereo recording could be great - put the whole group in a good-sounding room with a stereo mic pair in front of them. Maybe just one extra mic for the vocalist, panned center. More usual though these days to give everyone a close microphone and pan-pot some instruments away from centre. Not so much a common technique as an essential one!

You've been offered a selection of technical solutions to a problem mix. I strongly suggest you go back to source and solve any musical problems first!

Does the singer have good diction? Perhaps you can't hear the words because she isn't singing them clearly. Vocalists sometimes need persuading that "feel" isn't destroyed by clear diction.

One last thought. Sometimes a vocalist, lacking confidence in their own voice, asks to be DRENCHED in reverb and other effects. This won't help!

  • 1
    Very true! For a mostly acoustic recording, actually starting with a good acoustic sound is key to getting really good final result. – leftaroundabout Mar 18 '15 at 2:05
  • 1
    Yup. "Compose and arrange BEFORE you record" may sound obvious, but people don't :-) – Laurence Payne Mar 18 '15 at 11:52
2

Something is sitting where vocals should be. You have to find out what it is for each part of the song. Try muting every suspect and combinations of them until you find out what is your problem. Now, for each section of the song, you must know why there is a problem. You can see mixing as organizing sound into a 3D room where the 3 axis are Panning (sides), Frequency (height) and Volume (depth). There is also the "size" of each element, related to time effects. Here are some possibilities and solutions:

  • Solutions related to Volume:

The obvious one would be to turn the vocals up and the rest down, but there is more to volume than that. The dynamic range of the vocals might be too big. This means that the high volume parts are much louder than the low volume ones. The problem here is that if you turn your vocals louder so you can hear the quiet sections, the louder ones will be too loud. The solution to this is compression, as it will narrow the dynamic range. This is probably the most important advice to fit vocals in a mix. You can also try saturation effects like overdrive and distortion (they also make the dynamic range lower) instead of a compressor. Experiment with compression and saturation.

  • Solutions related to Panning:

This is probably not going to solve your problem, but it can help. If the vocals are a key part of the song - and they probably are - you should keep them centered, but you can try to get your problems to the sides so they get out of the way, as long as it sounds good.

  • Solutions related to Frequency:

This is really important. You mentioned the singer's voice has high frequencies. Try to roll these off from the violins, for instance. Also, try to identify more problem sources and what frequencies they have in common with the voice. If you spend some time on identifying all problem sources and EQing them properly, the results will be really good. Try to work mainly with subtractive EQ: Don't boost frequencies too much.

  • Solutions related to Time FX:

Maybe the voice should be bigger inside the track. Or smaller. This varies greatly with the song you are working in. If the voice track already has delays, reverbs or any spatial effects in it, get rid of them. Does it sound good? Try to reintroduce reverb or delay until it fits nicely.


Remember: 1. Identify source of conflict; 2. Try this four categories of solutions, one by one; 3. Move on to another source.

2

Along with filtering, EQ, and reverb (in reverse order of priority for the folk genre, probably) there is a pretty good way of creating the kind of separation you're looking for, if you aren't afraid to possibly lose some of the string 'body' if the mix ever gets played through a monaural audio system.

First, copy the competing string part to two tracks [panned hard left/right respectively], and on one of them, offset it by 2-4 ms (with a delay, fully wet - no dry signal on the delayed channel). Then on the delayed channel, invert the phase (many DAW's and mixers have a button for this directly in the console). For your vocal, leave it more or less centered on a single track. That way the strings will gain the illusion of spreading around the room, while the vocals will remain tight and focused in the middle.

You WILL have to meticulously tweak every detail of this method, from the delay used to how you use reverb and EQ to blend the results back together. The results, however, are absolutely worth the time invested and there are a lot of ways to improve on the general technique.

A rough outline of the techniques, and how they can be applied in complex ways, can be found here: http://www.independentrecording.net/irn/resources/phasepol/content/txt_wrap_stereo.htm

  • 2
    The split-delay-invert technique you describe is certainly worth knowing, but done this way it's actually equivalent to a single comb filter, and will give a weird hollow sound. Did you mean to recommend that with the split channels panned hard stereo? Then it can indeed be quite effective for getting the violin “out of the center” while still retaining the mono modes. I would however hesitate to do it this way in a folk recording, since it'll always sound quite artificial. For more natural sound, better put different takes of the violin to L/R. – leftaroundabout Mar 18 '15 at 1:53
  • Hrm yes, I'm sorry I did mean to add that. It's been a while since I've actually done anything on a mixing console and that kind of thing is near the limits of what I'm familiar with. Thanks for the additional tips! – Darren Ringer Mar 18 '15 at 2:00
2

Here's what I woud try :

1) Eq - if you can reduce the more piercing tones of the violin (upper midrange?) it'll allow the voice to prod through a bit. This is probably simplest to try if you have parametric eq on your mixer.

2) Panning violins to each side - as Todd Wilcox suggested this can result in the violins still being 'in the mix' but the vocals will sound more definite in the middle. The effect is lost if listening in mono though.

3) Compression. You could apply this across the whole of the piece or just the violins and guitars. Studio Compressors often have a separate input for the sound which is driving the compression (sometimes called 'trigger' signal). You don't hear that in the resulting mix- it's just a trigger signal. If you're apply the compressor to the whole mix, then you could set the trigger signal to be the whole song, but with the voice higher in the mix so that it's more persuasive over what the compressor does.

It follows that the mix going into the trigger isn't the same as the main mix so there might be some mixer jiggery-pokery to do there. I usually use a spare effects-send to do this.

So in a nutshell :

  • Apply compression over whole mix (at first at least)
  • Trigger signal for compressor is the whole mix plus a bit of extra vocals
  • You could also try applying the compression over just violins and vocals, or some subset of the mix.

The result will be that the voice will try to 'sit' just above the rest of the music in volume, as the compressor uses the a-bit-more-voice-than-the-rest to detemine the overall volume.

You have to be careful with this though. If you get too much voice in the compressor trigger signal, you can end up with the voice being too dominant and end up with a "voice like a fog horn" effect. An example can be heard in Whitney Houston's "And I will always love you" just after the keychange, where her voice pretty much drowns out the rest of the music. Also DJs use compressors in this way so that when they talk over a record, the record ducks out automatically & you can hear their -er- wisdom. That's a rather savage example though. You'd probably want something much more subtle than that !

2

In order to make a vocalist stand out in the mix, you need to do the following:

  1. Use a Compressor on the vocal track, with a hard compression ratio. This is realy an essential step in any vocal track. Just be sure to also put a Noise gate to eliminate the parts where the singer is quiet.
  2. EQ on the vocal track: using good reference speakers, it is your goal to first make the singer sound as if he was singing in front of you, which means realy correcting the recorded track to sound as close to a live performance. I can't realy tell you how to do this here because that depends alot on the type of microphone and the frequency response in which it records. You'll have to figure this out yourself. And then, once you've done that - put a hiss eliminator (this is usually a multiband compressor that is set to compress the high end 4Khz+ range to eliminate "SSS" sounds from jumping), and then - add brightness to the treble (3khz - 8khz) range to make the singer sound more "in front", and a small reverb.
  3. Not neccessary, but may benefit alot if you have this gear - put a sidechain compressor or sidechain EQ whenever the singer sings to "duck" away any interfering instruments like the violins or such.

This should do the job.

1

A basic first step would be to use filters/EQ to cut off the voice frequencies with the conflicting instruments (and maybe a bit of the low ones of the voice). Just try it with basic high/lowpass filters and you will see in a few minutes if this helps. If it does you can take more time to finetune it so as few frequencies are cut off as possible and still make it sound good.

It is a tedious process, but practice will help a lot and make you better and faster.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.