Two issues when recording vocals - are these normal?

Hi, these two things have been puzzling me for some time.

 

First, I always need to turn on low-latency mode when recording vocals because there's too much lag. I have a fairly decent-spec computer (2017 MBP, 3.1 GHz Quad-Core Intel Core i7 with 16 GB 2133 MHz LPDDR3) going through a 2019 Focusrite Scarlett (have also tried an M-Audio interface). Buffer in Logic is set to 128, and have experimented with that, too. Is this common to have to turn on low-latency or should there be things I can tweak?

 

Second, if I create an audio track, add plugins like Autotune/Waves Tune and CLA Vocals on to it, and arm the track for recording, I can hear the Audio FX plugins when the song isn't playing, and I can hear them when I'm recording, but I can't always* hear them if I simply press PLAY.

 

To be clear, if I'm recording then Logic is recording the effects fine, but sometimes I just want to play the track and sing along without recording and it's nice to have autotune and other effects working while I do. This issue seems to be independent of the plugins I use, and I've tried switching low-latency mode and on off, etc. but I can't find any way to remedy it. Again, is this expected behaviour?

 

* The strange thing is, Logic isn't working consistently - sometimes I can just hit play and still hear the effects. Other times I can't.

If you've got CPU intensive plugins in your chain while you're tracking, you'll struggle to not have latency (Autotune needs to hear the audio before it can process it, hence the latency with that particular plugin). Take it out (remove it, don't just disable it), you'll be fine. If you've got low latency mode on, you won't always hear the effects - they're disabled. Easier to take Autotune out and just have a little reverb to help you with pitch if that's the issue. Tune the vocals after recording. Someone's gonna say 'or sing in tune', you watch ...

Here's an exercise for you:

 

Create the next project so that you only have EQ and compressor in the single channels, and one single, low CPU reverb like SilverVerb, driven from Sends. Steer clear of CPU hungry synths like Alchemy or Sculpture. Now you can set your IO buffer to 64 or even 32 samples and record your vocals without latency.

 

Oh, and - sing in tune...

I always thought that Logic records the signal *before* the channel strip - basically "pre" everything. The resulting WAV file is the raw signal. The only way I have been able to "print" an EQ into a recorded signal is to place the EQ (or compressor or whatever) on an input channel using the environment.

 

I assume (I haven't really tried) that placing a gain insert on an input would add gain in the 32/64 bit processing realm, so no clipping if I add a ton of gain.

Yes, on the first part.

 

On the second - if the processing is done in the input before getting printed to the file, remember the file you are recording is 24-bit, so it would be entirely possible to gain up the stream significantly and end up with a massively clipped 24-bit file - I haven't tried to see what happens in that circumstance though.

The quick experiment says that adding Gain on an Input channel doesn't work in the 32-bit realm - it just shoves everything up and clips the signal just fine thank you very much.

 

The answer for me is still the same - get a nice level on the input (trim) that never gets into the red - add all the gain I want on the tracks, subtract excess signal at the stereo output using a gain plugin before the Adaptive Limiter

 

24-bit recording. Check.

One suggestion I’d make:

 

One can spend forever trying to make bad tracks sound good and never get there. Or one can spend a few months honing their skills as a musician/singer and get better results faster (in the long run).

 

So in short, learn to sing in tune and ditch pitch correction.

 

At the very least avoid pitch correction while singing... singing one thing while hearing another (off pitch while hearing it corrected) will throw you off even if latency weren’t an issue.

This.

The quick experiment says that adding Gain on an Input channel doesn't work in the 32-bit realm - it just shoves everything up and clips the signal just fine thank you very much.

 

Well, the gain plugin might work/process at 32bit - but that 32bit precision is going to be lost as the file is printed at 24-bit fixed point, so anything about 0dBFS will be clipped, just as at the output of the mixer.

One suggestion I’d make:

 

So in short, learn to sing in tune and ditch pitch correction.

 

And another ...

I can't say about AutoTune, but I know WavesTune wasn't designed to be tracked through live. It was a tool designed to help fix issues after the fact, like FlexPitch. WavesTune RT is what you will want for live tracking or even Ovox if you want to do the vocoder thing.

 

I'm with the guys above though, it's always more desirable to learn to sing. I'm not trying to sound patronising, and I don't think the other guys are either, there is actually a very real practice application in learning how to sing.

 

For one, you feel very empowered and liberated, you feel more free and unbound by technology, as a result its also faster to lay down good takes, and the fast the session run the bigger the chance you have of not running out of steam and losing interest in it. In this day and age where everyone is relying on technology you will stand out, it gets more credibility, making you one of the people others will choose to work with.

 

There's basically lots to gain and nothing to lose. It does take time so don't expect miracles over night. But its always something you can get going while using pitch tuners in the mean time.

while i do agree with everyones sentiments about singing, some genres like modern hip hop pretty much everyone uses autotune in a weird way. Maybe hearing the end product of the autotuned vocal is integral to OP's genre in whatever he does? What if OP does know how to sing, but is watering down his vocals purposely? I digress.

 

like someone else said though, low latency mode will turn off plug ins.

while i do agree with everyones sentiments about singing, some genres like modern hip hop pretty much everyone uses autotune in a weird way. Maybe hearing the end product of the autotuned vocal is integral to OP's genre in whatever he does? What if OP does know how to sing, but is watering down his vocals purposely? I digress.

 

like someone else said though, low latency mode will turn off plug ins.

 

It's not to decry the relevance of autotune in modern music. In actual fact, it works much more effectively and quickly if you are on pitch than if you were off pitch. So it has some very real relevant practical applications. There are actually a few hip-hop artists who can actually sing themselves, but they still use autotune themselves for the effect. It doesnt have to be either or, it can be both. T

while i do agree with everyones sentiments about singing, some genres like modern hip hop pretty much everyone uses autotune in a weird way. Maybe hearing the end product of the autotuned vocal is integral to OP's genre in whatever he does? What if OP does know how to sing, but is watering down his vocals purposely? I digress.

 

like someone else said though, low latency mode will turn off plug ins.

 

It's not to decry the relevance of autotune in modern music. In actual fact, it works much more effectively and quickly if you are on pitch than if you were off pitch. So it has some very real relevant practical applications. There are actually a few hip-hop artists who can actually sing themselves, but they still use autotune themselves for the effect. It doesnt have to be either or, it can be both. T

except, that's not correct if you have a different idea of what "works much more effectively" means. I am aware that the more "off key" you sing, the more the autotune effect is prominent (it sounds like a robotic toddler is drowning in water). The more accurate you sing, the less that sound is prominent. Who is to say that OP isn't going for that robotic baby water cry noise intentionally?? Again, not my preference, but i think this is a creative choice that OP may feel comfortable with.

 

I feel like this thread is derailing though, so if this tune talk is getting annoying, i'll stop.

And, try this: Use a multi-step workflow.

 

Bounce or freeze the project into an audio file that will be played through the singer's phones. Turn everything else off. This generates an audio file, and the computer doesn't have to do that "in real time."

 

The immediate project now consists of playing this audio track (which requires no CPU resources to speak of) through the singer's headphones, and capturing the vocal. Now, you probably can afford to put auto-tune back into it, if s/he'd like to hear it "tuned," because you've now probably got CPU resources to spare. But you really don't want to do too much to the performance at this point: don't do right now what you might wish to (re-)do later. (You could, for instance, "send a tuned version to the phones" while capturing the raw.)

 

After recording the raw or minimally cleaned-up vocals, you can now separately "sprinkle the magic-mojo fairy dust" on it to produce a new, cleaned-up audio track for use in the final project, leaving the raw-vocals muted. This also does not have to be done "in real time."

 

It's not to decry the relevance of autotune in modern music. In actual fact, it works much more effectively and quickly if you are on pitch than if you were off pitch. So it has some very real relevant practical applications. There are actually a few hip-hop artists who can actually sing themselves, but they still use autotune themselves for the effect. It doesnt have to be either or, it can be both. T

except, that's not correct if you have a different idea of what "works much more effectively" means. I am aware that the more "off key" you sing, the more the autotune effect is prominent (it sounds like a robotic toddler is drowning in water). The more accurate you sing, the less that sound is prominent. Who is to say that OP isn't going for that robotic baby water cry noise intentionally?? Again, not my preference, but i think this is a creative choice that OP may feel comfortable with.

 

I feel like this thread is derailing though, so if this tune talk is getting annoying, i'll stop.

 

There are always exceptions to any rule/guideline. However, a person who can sing in tune and is efficient at it can also sing out of tune if they choose to and work the autotune correct. The converse isnt true, however, someone who can't sign takes a long time to track and along time to tune, the only benefit is they may be able;e to get a half decent autotune sound. Consider though, that thing has been trending for the last 10 or years, how much time do you think we have left before it will be "considered "unfashionable".

 

Consider also, I actual gave advice on how to achieve the effect they wanted, but only ADDED that learning to sign helps speed up productivity. So it's not like I was being counterproductive with my original statement. I was only seeking to provide additional advice. To be clear, I actually own Ovox and VocalSynth to get that creative vocal effect you were talking about. So its not like I condone it or anything, just seeking to help, someone better themselves and their craft.

 

Anyway, my point is, if learn to sign your options are much more open and you become more effect and easier for others to work with. You can still sing out of tune, you can still work the autotune. You're also seen as "more professional" making you a solid choice over anyone who can't sing. You gain a lot and you lose nothing. Its actual something worth considering for many people.

Consider though, that thing has been trending for the last 10 or years, how much time do you think we have left before it will be "considered "unfashionable".

 

I thought that would happen... a *long* time ago. *sigh*

 

Still waiting, there seems to be no sign it's going away (amongst the people who can't sing that is)...

Consider though, that thing has been trending for the last 10 or years, how much time do you think we have left before it will be "considered "unfashionable".

 

I thought that would happen... a *long* time ago. *sigh*

 

Still waiting, there seems to be no sign it's going away (amongst the people who can't sing that is)...

 

I actually think the duration has been lengthened because of the vocoder trend now. The lines between artificial pitch correction and vocoding can be blurred a bit. Vocoding at least goes hand in hand with the retro 80s vibe that seems to be rearing its head. I can see all the kicking around for quite a number of years yet, so vocoding and by extension, synthetic pitch correction may be with us for quite a bit longer now.

 

Actually I can see the next evolution of pitch and vocoder plugins follow this trend as developers squeeze both these tools into one hybrid plugin, for better and for worse. That in turn will help sustain these trends as well. You can get close to that already with VocalSynth or Ovox, but I can see that improving.

 

I'm not against pitch correction as an "effect", but when everything's smothered with it and there's very little way of musicianship or song craft, then it does get a bit much. Sure I like a bit of sauce (ketchup) on my hotdog, but I really don't like overloading it too the point you can't even taste the hotdog any more.

Thanks guys - never got any notification about replies and now see all these!

 

Although, as well intentioned as they may be, the comments about learning to sing *are* unhelpful because, for one thing, they take an example and make assumptions about the use-case. For one thing I've spent 30 years trying to sing and I still can't - sorry, I haven't got another 30 years to get a bit better - not everyone can. But I have my reasons for wanting to do this way - the ontopic answers are appreciated though

Vocoding can be done beautifully - Daft Punk, Royksopp, but also old hands/great singers like Stevie Wonder.

from about 2.55 to 3.25