[JasonHiller]

If I am having someone overdub on a prerecorded track, is it being printed in Pro Tools late? Is every converter different? Do DAWS handle this in different ways? I hear much about round trip latency but what about one way? Is the music being played out of a DAW coming out at exactly the time it is being played, like off a tape machine at the speed of electricity?

[Ryan Sutton]

<p style=”text-align: left;”>Jason, while I’m not a technical expert on the round trip timings in and out of Pro Tools, I can tell you that this really only effects me when there’s a high playback buffer setting engaged.  In my experience, if the playback buffer is set to 128 or below, you should be just fine.</p>

[Bob Katz]

Dear Jason:

Welcome to the Digido Fora!

There is a lot for you still to absorb and you’ve asked a multi-faceted question! Let’s start with the concept of a buffer: There is always a buffer required for digital audio playback, since the samples have to be assembled for playback, coming off of storage. As the samples are not stored in order of playback. Plus if you do an edit, then your DAW has to access one set of samples from one location and then marry them with another set of samples from another location.

That all takes time. To handle that, the samples are assembled into a buffer memory and then played out. The shorter the buffer, the less the latency or delay between the playback of any sample and the output to the world. The shorter the buffer, the more the chance of getting glitches as the computer tries to catch up with all the data. The longer the buffer, the more stable the playback, leaving room for complex operations like EQ, compression, etc.

DAWs have a built in mechanism where what you see on the screen (the waveform) can be made to appear synchronous with the sound. It’s a trick with the buffer where the computer delays the waveform view until the audio buffer has completed.

If you are overdubbing, buffers and latency can get in the way. So it’s important then to use as short a buffer as your system can tolerate without getting glitches or being unstable. Some DAWs have tricks of programming which allow you to overdub without perceptible delay, maybe a minimum delay of only a few milliseconds, which hopefully is not perceptible.

Yes, every converter is different, but most modern converters have a delay of only 2 or 3 milliseconds max. The higher the sample rate, the shorter the delay.

I suggest you experiment with using the longest buffer that Pro Tools provides and try to overdub. You’ll soon experience issues with the latency. Admin Ryan Sutton is a pro tools expert and could answer your questions in relation to overdubbing in Pro Tools specifically. I think Ryan may see this thread and “amplify” here!

[JasonHiller]

Wow thank you for the answers! This changes everything… 🙂

[Bob Olhsson]

Latency is a far worse problem with headphones than most people realize. A cheap analog mixer combining the DAW playback with a live feed split from a microphone or instrument can improve performance and speed up the recording process dramatically over even the lowest latency DAW. It’s less convenient but the results are well worth it in my experience.

Ironically, recording to a loudspeaker rather than with headphones is another way to solve the problem. I suspect the headphone issue has more to do with the acoustical sound mixing with the short delay from the digital audio than a longer delay from a speaker. Perhaps increasing the latency a bit is another solution.

[bob stjohn]

“The higher the sample rate, the shorter the delay.”  well, didn’t THIS blow my mind.  never thought of that.  as the majority of my work is mixing, i deal with latency on a daily basis.  and there is this “dirty little secret” to protools, mixing and latency…in regarding to automation.  it just isn’t as precise as we may believe. when you are mixing ITB, and you have a few plugs on the master fader, you start looking at delay compensation in excess of 8000 samples.  it becomes obvious when looking at the meters, and even MORE obvious when you are doing fader automation that requires precision (perhaps a cut on the beat, or automation for a specific word or guitar part, for example).  i had some delusion, that over the years, avid had addressed this issue…but obviously it’s still somewhat of an issue; my workaround is to just shut off delay compensation while doing precision moves (and suck it up and listen to all of the drum sampling plugs playing back delayed and out of sync) and turn it back on when i’m done.  One thing is for sure…as long as you have a master fader on your mix output, you’ll have latency when doing actual recording; so when tracking, i’ll typically use stems and as low as a buffer as possible, as well as NO master faders or routing live tracks through aux returns…

[Bob Katz]

Yup.

Let’s look at 8000 samples latency at different sample rates:

at 44.1 kHz sampling, it’s 22.7 microseconds per sample. So 8000 samples would be 181 milliseconds. Oy vey.

So at 88.2 kHz, that halves to 90 ish milliseconds. It’s still a lot of time, but somewhat better.

You can see why as Bob Olhsson says in this thread, it pays to overdub using analog to hear yourself, and yes, some DAWs provide special analog-style routing for overdubbing.

[bob stjohn]

30 years of this and no one really explained that math to me that simply…thank you bob!  when recording…i just don’t use anything that can add to the latency already inherent.  as i said…stems, no master faders, no native plug-ins (using a DSP based plug-in you can get away with a little without too much added latency).  but hey…that’s why i have outboard eq and comps to record with 🙂

[Bob Olhsson]

“Analog style” is NOT analog! Just back-to-back converters introduce a significant delay.

Fader latency is another issue. I could no longer ride vocals “off the floor looking the singer in the eye” using consoles having moving fader automation. DAWs are often even worse.

[Bob Katz]

Hi, Bob O. Back to back converters might add 2 ms. latency with today’s current converters at 96 kHz sampling. It’s about the smallest you can get. Not sure how that would affect a critical user like a singer wearing headphones. Personally I would give them analog foldback and that’s what I meant where some interfaces can give a pure analog foldback path to the performer during overdubbing.

[Bob Olhsson]

It would be interesting to try longer delays comparable to listening to speakers. I’ve never been in a position to experiment with this.

A number of interfaces define “analog” as not being fed through the computer as opposed to real analog.

[JasonHiller]

So much great info! I just tracked a record and the singer sang and played guitar in front of the monitors (ns10s) without headphones (his request) and it just came out great! The only latency (because we were recording all analog to the 8 track, no computer at all) was the fact that the drums were not completely baffled (because I love bleed) and were about 30 feet away from the singer so he was hearing the signal through the speakers first, the live sound was delayed. If I remembered how to do math I could give you some more info… 780mph… 5280ft in a mile… 30ft… someone help me…

For digital, the interfaces I use are Metric Halo ULN-8s and they have a balanced out that is pre A/D so I will try using those for headphone sends. But I’d like to know more about this headphone latency problem that Bob Olhsson is talking about! Should I just use individual monitors for the players?

Thank you!

[Bob Katz]

Hi, Jason. The full analog way you are working certainly creates fewer ergonomic and setup problems than with the digital lash ups. The acoustic time delay issue with the drumset is going to occur no matter how you feed foldback to the singer. I’m sorry to say this, but in this case the leakage that is so desirable is working against you in this situation. Instead of isolating the singer I would work on having them sing nearer to the drums to hear them acoustically (say by removing one ear of the headphones) but use a hypercardioid or figure 8 mike as well as gobos to minimize drum leakage into the mike. And not feed the drums to the cans at all.

[Bob Olhsson]

I would just use one monitor. The actual “old school” method was using no monitor with the musicians playing softly enough to hear each other and the singer.

What’s amusing is people isolating drums and then adding delay effects which actually simulate bleed. At Columbia in Nashville, SOP was the singer directly in front of the drums using a cardioid microphone. That way the bleed was clean and enhanced the drum sound much like today’s “room” mikes.

[JasonHiller]

I am getting ready to do another record, and the artist is willing to do it live to tape as is my preference, so I think I will try the monitor situation! However, it is not always easy to tell the singer or anybody in the band that you gotta get it right start to finish!  But it sure adds to the urgency and vibe of a tune. Punching in on the Ampex takes so long the song is almost over by the time I hit record. 🙂

Who is SOP?

 

…and I now would like to express extreme gratitude and thankfulness to be in a conversation with both of you Bob O and K and the other experts in this forum. It is truly humbing, heartwarming, exciting, and fills me with so much hope I can’t thank you enough. I’m honored.

[Bob Katz]

Live to 2 track has been my forte for many years! The tension is  live enough to create a spark :-). You can do several takes and edit them together you know. It can give you the best of both worlds. Doesn’t always work, especially if the musicians don’t work to a click track, but if you’re good at editing you can accomplish something.

“Isolating” the vocalist is often considered and as we’ve discussed, you’re damned if you do and you’re damned if you don’t. I recently mastered a great album where the vocalist had performed live in the studio with a big band using, gawd, an SM7. It was supposed to be a “work track” but you know how that goes, she fell in love with her performance. The producer did a marvelous job of masking the sibilance and other issues and the band sounded a bit washy and monoish at times due to leakage into the vocal mike in a live studio. But the end results were hair raising!

There are other ways to get great results miking a vocal in a live studio with musicians:

— semi gobos, giving her the ability to hear everyone or nearly everyone but blocking off at least some of the issues from the mike

— Use a great hypercardioid or figure 8 mike on the vocal

— VERY careful EQ on the vocal mike during the mixing. You certainly can filter below, say, 60 Hz, and if you have excellent wide-range monitoring, slip the filter upward in frequency VERY carefull, always erring on the side of vocal fidelity while living with the leakage

— Izotope RX on the vocal track. You can eliminate drum beats between the syllables. It works, it works well. Again, not being overzealous and do not take away the baby with the bathwater!

[Bob Katz]

Also, differential miking. Only for the brave, but it’s a valid technique.

Also, overdubbing a vocal to fix a line or two in post is not out of the question. Of course you get that hole in the ambience and I wonder if feeding the band into a speaker while she’s overdubbing, or simulating the leakage with mix tricks could possibly help to allow a vocal repair line. Personally, I’ve never done it, just thinking outside the box here.

[Bob Olhsson]

SOP means standard operating procedure. The first time I tried it, I was floored by how great it sounded. It would require recording full band inserts to edit out errors.

I would not use tape for a contemporary live recording because you will probably want to do quite a bit of editing.

[JasonHiller]

Ok great stuff. I just finished setting up all the mics and such, tracking tomorrow, no headphones all live to the 8 track. Drums, bass, acoustic guitar, electric guitar, and lead voc. Can’t wait to see the look on their faces when I tell them there are no headphones, oy! I know The Wallflowers first record was done like that, and I believe Bob Dylan’s Time Out Of Mind was like that too. And there were a few vocal overdubs that they did in front of the speakers. I should find out about more records done like this so I can tell the band.

thanks again and I’ll keep you posted.