Jump to content

MIX and ISO Tracks levels


vale

Recommended Posts

12 hours ago, Glen Trew said:

2) Don't be afraid to use the mixer's and the recorder's dynamic range. In fact, be afraid to not use it. The audio quality does not degrade when going into the "yellow" and then degrade further when going into the "red" (technically, it actually improves). The perfect recording should have the highest peak reach 0dBfs.

 

I don't necessarily agree with this, it sounds like the old "don't waste any bits" argument of the 16 bit era. I have found that it really depends on the analog front end and ADC.  Mic gain and ISO recording levels are linked (if pre-fade ISOs are recorded) so if you have a cheap mixer/ recorder (Zoom F8, etc) gaining up to "use every bit" may result in a "pinched" sound from the pres giving out before the ADC clips while under-cranking the mic gain may result in an "anemic" sound. I usually shoot for peaks in the -18-12dBFS range and as a post mixer I expect to make up 10 dB of gain or more on ISOs, no big deal as long as there's good S/N ratio, which again can be a problem with cheap mixers/ recorders when printing low levels, regardless of 24bit resolution.  On quality mixers/ recorders that most here use this is less of an issue since a good quality pre and ADC will be linear across a wide range but many cheap interfaces and mixers/ recorders underperform in this area so it is something to be mindful of.

I wouldn't say that a recording approaching 0 dBFS would constitute a "perfect recording" unless I knew the recording chain. 

Link to comment
Share on other sites

10 hours ago, borjam said:

At 24 bits it's safe to leave a big headroom, you will still have plenty of usable resolution. Leaving 8 bits of headroom at 24 bit will still give you full resolution 16 bit signals. Of course you can't enjoy that luxury recording at 16 bits, you would have only 8 bits which would really suck.

 

Yes, but 24 bits only provides a theoretical dynamic range of 144 dB or thereabouts. 

The a-d converter of the 688 for example only has a dynamic range of 114dB and even Zaxcom‘s NeverClip only has a range of 137dB. 

So leaving too much headroom and thinking that all is fine because it’s 24-bits can quickly be a grave mistake.  

Link to comment
Share on other sites

 

7 hours ago, Werner Althaus said:

I don't necessarily agree with this, it sounds like the old "don't waste any bits" argument of the 16 bit era. I have found that it really depends on the analog front end and ADC. 

 

If the analog input module levels are aligned with the recorder iso track levels, and any links in the chain in between (i.e. proper gain structure), then the possibility of one distorting while the other is largely unused is not a concern, whether or not it's a Zoom or Sonosax device. Of course, riding gain in an attempt to keep levels at or near maximum would not only be way too risky, but also in poor taste with regards to natural dynamics, so no one is suggesting riding gain to keep all levels near 0dBfs. But what seems to be happening more often - recording levels pointlessly low - is because of some being needlessly uncomfortable with peaks going much past -20dBfs or -10dBfs, because of the incorrect assumption that it's starting to sound bad.

 

Then there's the opposite problem I've seen with some who mix and record with their peak meters often pegging at full scale, with the assumption that there's useable headroom beyond zero, like was sometimes the case in the analog days. But that is a problem now because, with digital, if the meters are calibrated correctly, there is nothing useable beyond full scale.

6 hours ago, Werner Althaus said:

I wouldn't say that a recording approaching 0 dBFS would constitute a "perfect recording" unless I knew the recording chain. 

The assumption should be that a professional user knows their recording chain and will have all the links properly align and configured.

Link to comment
Share on other sites

10 hours ago, Glen Trew said:

 

 

If the analog input module levels are aligned with the recorder iso track levels, and any links in the chain in between (i.e. proper gain structure), then the possibility of one distorting while the other is largely unused is not a concern, whether or not it's a Zoom or Sonosax device. 

 

Maybe I didn't explain clearly what I was trying to say but my point was that in practice I can gain up a nice Schoeps mic to give me "good healthy and clean" levels (peaks around -3 or 4 dBFS on my Zoom F8) only to find that the mic sounds "pinched" or 'slightly distorted and shrill"while backing off the mic pre's gain and effectively lowering the ISO levels will improve the response significantly but going too far the other direction into "super-safe" territory will result in a lifeless, dull recording. These linearity issues are  a fairly common problem with budget audio interfaces and mixers. I tend to gain stage according to the gear I'm using, the better the chain, the hotter (or lower, not a paradox since improved linearity affects both ends of the dynamic range) I feel I can print, within reason. The frontend on SD 442/ 664/ 552 or 633 , while all sounding different, have never given me any grief in this area but the Zoom definitely has as have the mic and line inputs of cheap cameras. I'll hit a Sony PDW700 different than a Canon C 100 and I remember the Sony HDW-F900 sounding way better when hit hard than any camcorder on the market today. So IMHO proper gain staging is one thing but knowing how your gear responds to varying levels is also important.

 

Link to comment
Share on other sites

3 hours ago, Werner Althaus said:

 

Maybe I didn't explain clearly what I was trying to say but my point was that in practice I can gain up a nice Schoeps mic to give me "good healthy and clean" levels (peaks around -3 or 4 dBFS on my Zoom F8) only to find that the mic sounds "pinched" or 'slightly distorted and shrill"while backing off the mic pre's gain and effectively lowering the ISO levels will improve the response significantly but going too far the other direction into "super-safe" territory will result in a lifeless, dull recording. These linearity issues are  a fairly common problem with budget audio interfaces and mixers. I tend to gain stage according to the gear I'm using, the better the chain, the hotter (or lower, not a paradox since improved linearity affects both ends of the dynamic range) I feel I can print, within reason. The frontend on SD 442/ 664/ 552 or 633 , while all sounding different, have never given me any grief in this area but the Zoom definitely has as have the mic and line inputs of cheap cameras. I'll hit a Sony PDW700 different than a Canon C 100 and I remember the Sony HDW-F900 sounding way better when hit hard than any camcorder on the market today. So IMHO proper gain staging is one thing but knowing how your gear responds to varying levels is also important.

 

Yes I think the same way. Feels the same if I use my Zoom F8. Don't have this problem with Sound Devices 788T

Link to comment
Share on other sites

The dynamic range of a speech recordings can be quite high because of factors such as varying subject-to-microphone distance, whispers to screams, etc. You can make the case to record the complete dynamic range on both the iso track and the mix. The argument for controlling dynamic range is just as valid.

 

Every time a microphone is moved relative to the sound source, the amplitude changes. Great boom ops use that principle to manually control dynamic range. Mixers continually ride faders to limit dynamic range. It is all going to depend on what the scene/project/staff expect. Some post mixers like fat tracks that have little dynamics. Some like to make that decision on their own. Again, project-dependent.

 

As far as recording, as stated above, 24-bit file containers give us latitude to leave plenty of headroom. The analog part of a microphone preamp circuit largely determines its noise performance.

Link to comment
Share on other sites

4 hours ago, jon_tatooles said:

The dynamic range of a speech recordings can be quite high because of factors such as varying subject-to-microphone distance, whispers to screams, etc. You can make the case to record the complete dynamic range on both the iso track and the mix. 

 

All true, if we know exactly what the source (actor) levels are going to be, which we never do, even with rehearsals. Therefore, we set the trim (which controls the iso track level) lower than we think we'll need, to allow for the occasional surprise peak, which means the iso tracks will usually be low, then making it up in the mix with the fader and our ears. 

Link to comment
Share on other sites

As a rough guide you can check this table of S/N ratios for different bit depths. 

 

The problem is when people use "signal to noise ratio" and have been trained in analog. Typically, analog s/n is specified to nominal, and headroom is on top of that.

 

So a Nagra IV with 66 dB published s/n at 7.5 actually could record 70 dB cleanly, and maybe with 76 dB with acceptable distortion.

 

The chart's "digital s/n" of 96 dB for 16 bits doesn't include any headroom. Depending on the device, one more dB will either give you horrible distortion, or be squashed. 

 

I've always used the convention of calling the digital measurement "dynamic range" rather than s/n, to avoid this confusion.

 

(Of course the chart also doesn't consider electronic noise in the preamp or ADC. I can't think of any 24 bit recorder in common use today that actually gives you 144 dB dynamic range from the mic inputs.)

Link to comment
Share on other sites

5 hours ago, Jay Rose said:

As a rough guide you can check this table of S/N ratios for different bit depths. 

 

The problem is when people use "signal to noise ratio" and have been trained in analog. Typically, analog s/n is specified to nominal, and headroom is on top of that.

 

So a Nagra IV with 66 dB published s/n at 7.5 actually could record 70 dB cleanly, and maybe with 76 dB with acceptable distortion.

 

The chart's "digital s/n" of 96 dB for 16 bits doesn't include any headroom. Depending on the device, one more dB will either give you horrible distortion, or be squashed. 

 

I've always used the convention of calling the digital measurement "dynamic range" rather than s/n, to avoid this confusion.

 

(Of course the chart also doesn't consider electronic noise in the preamp or ADC. I can't think of any 24 bit recorder in common use today that actually gives you 144 dB dynamic range from the mic inputs.)

That's really interesting, I never really thought about S/N vs dynamic range in analog vs digital systems and what it means exactly , so thank you for that insight. S/N is a confusing term anyway in this day and age when people refer to a microphone's isolation of wanted audio against unwanted audio as "Signal to noise ratio", arguing that directional mics have "better S/N", etc.

 

 I also don't think the theoretical dynamic range of a digital system as determined by some formula is of much use in determining the usable dynamic range of a given system. That's where the ears come in.

Take 16bit, giving us a range that should be adequate for anything you throw at it. I used to record a few orchestral performances and operas in the late 90's/ early 2000's on Tascam DA-88s (16bit). A few years earlier I was working some with blackface ADATs (16bit) and the usable dynamic range of these two systems struck me as vastly different because the ADATs sounded pretty ratty when pushed close to 0 dBFS while the Tascam handled it much better IMO. On the other end however they both exhibited a high degree of quantization noise, dither/ noise shaping wasn't what it is today. So while the 16bit machines theoretically had a dynamic range of 96dB plus change, in reality they performed (much) worse than a good analog recorder with SR. 

Even today those numbers tossed around regarding dynamic range don't warrant much attention IMO ("ADC A is better because it has 137 dB of dynamic range, ADC B is bad because it only has 127" etc.) . The Zoom F8 (sorry for picking on this one, it's actually a nice machine for the money) claims a dynamic range of 120 dB, in reality I treat it as little more than half that. 

Maybe I'm wrong but the only 3 digit number in decibels I care about is the maximum SPL a mic can handle, that number actually matters as we all know. 

Thanks again.

Link to comment
Share on other sites

I’m still using PPM metering on most of my consoles-just what I’ve grown accustomed to over the years. Calibrated for -8dbu PPM for -20Dbfs (O VU equivalent) on the recorders. Have never had a complaint from post regarding levels.

 

There a a few boards that are equipped with limiters on the input channels, but I try not to hit those too hard for the iso’s. And I try to avoid changing the input gain trim mid-take if possible, just to avoid the subsequent change in level when post receives the tracks, as it makes it a bit more difficult doing signal processing when the levels are bouncing around.

 

-Scott

 

 

Link to comment
Share on other sites

On 12/4/2018 at 10:51 PM, Constantin said:

 

Yes, but 24 bits only provides a theoretical dynamic range of 144 dB or thereabouts. 

The a-d converter of the 688 for example only has a dynamic range of 114dB and even Zaxcom‘s NeverClip only has a range of 137dB. 

So leaving too much headroom and thinking that all is fine because it’s 24-bits can quickly be a grave mistake.  

 

Of course, and my apologies for the poor explanation. In the 16 bit world every bit was sacred ;) In the 24 bit world with dithering you can sacrifice several for headroom without bad consequences.

 

On 12/6/2018 at 6:32 PM, Jay Rose said:

As a rough guide you can check this table of S/N ratios for different bit depths. 

 

The problem is when people use "signal to noise ratio" and have been trained in analog. Typically, analog s/n is specified to nominal, and headroom is on top of that.

 

So a Nagra IV with 66 dB published s/n at 7.5 actually could record 70 dB cleanly, and maybe with 76 dB with acceptable distortion.

 

The chart's "digital s/n" of 96 dB for 16 bits doesn't include any headroom. Depending on the device, one more dB will either give you horrible distortion, or be squashed. 

 

I've always used the convention of calling the digital measurement "dynamic range" rather than s/n, to avoid this confusion.

 

(Of course the chart also doesn't consider electronic noise in the preamp or ADC. I can't think of any 24 bit recorder in common use today that actually gives you 144 dB dynamic range from the mic inputs.)

 

You are right, as I said my post was written too quickly. Quantization noise is deceiving because it is not a “constant” noise signal, but it’s correlated to the recorded signal, the weaker the signal the worse. 

 

In my experience anyway (recording live jazz concerts mostly) I can leave a headroom of even 20 dB when recording drums. 20 dB related to the level when I ask the drummer to play loud during the sound check. Of course I know that during the actual concert it will be much louder. 

 

 

Link to comment
Share on other sites

On 12/2/2018 at 8:40 PM, vale said:

 

Editor and his assistant are happy with my levels (they're editing the show just using my MIX track).

The problem came out on the dialogue editing side whe they start working with the iso tracks.


Well that is sad, as the sound post department should be more experienced working tracks. You'd more expect that type of ignorant comment to come from the visual side of things who are just cutting the picture

Link to comment
Share on other sites

Iron - tut tut - don't blame other departments. I've worked with picture departments on sound who have taught me stuff and could blow many sound dept folk away with their ability (privately one in mind with an Oscar nomination: his ex assistants taught me, they all helped me, and everyone helped to get the job done).

 

Scott, I also love and trust the PPM - after all, it's how I learnt my trade. And whilst having assisted in music I never had the same trust (actually, embarrassingly, less trust but understanding through practice) in the VU on the consoles.

 

Even when I don't plug the PPMs in I still hang onto them: I know the 'trust' is from a broadcast basis but it's also from knowing what's what in practice.

Link to comment
Share on other sites

On 12/8/2018 at 3:13 PM, borjam said:

In the 16 bit world every bit was sacred

 

...which made it so painful when Sony decks were shipped with -20 dBFS nominal, matching +4 dBu on their analog jacks, so -12 dBFS (+16 dBu analog) became the peak level most US broadcasters would accept. Video CEs would argue with me that "yes, sending us DATs with -16 dBFS nominal or even -1 dBFS peak would be a cleaner track... but our guys would have to knock it down before they could lay it back on the DigiBeta master. So the net would be even worse, because of the additional math the playback deck would be adding.

 

We've got better ways to measure loudness now, and deeper bit rates. But the old standard is still hanging on in ways... just like the deliberately slowed down QWERTY keyboard is still with us.

 

Link to comment
Share on other sites

4 hours ago, Jay Rose said:

 

...which made it so painful when Sony decks were shipped with -20 dBFS nominal, matching +4 dBu on their analog jacks, so -12 dBFS (+16 dBu analog) became the peak level most US broadcasters would accept. Video CEs would argue with me that "yes, sending us DATs with -16 dBFS nominal or even -1 dBFS peak would be a cleaner track... but our guys would have to knock it down before they could lay it back on the DigiBeta master. So the net would be even worse, because of the additional math the playback deck would be adding.

 

We've got better ways to measure loudness now, and deeper bit rates. But the old standard is still hanging on in ways... just like the deliberately slowed down QWERTY keyboard is still with us.

 

 

I get it but I felt that the Sony decks -20 dBFS reference = +4dBu with a front end that could take it all the way up to 0dBFS was a beautiful thing in the field and a great match to fieldmixers that actually were capable of outputting clean +24dBu. In the meantime companies like SD have moved the goalposts and now have made 0dBu the defacto reference with a max of +20dBu output.Not crazy about that one.

. For post to sacrifice 8-10 dB headroom at the top of the scale due to broadcast delivery specs had probably more to do with legacy technology in the analog transmission realm where 10dB of headroom  was all you got. I still have to mix live with a -10dBFS peak ceiling for certain networks, not my favorite thing in the world but I get where they are coming from and why they are reluctant to adopt loudness normalized TruePeak levels of -2dBFS.

Link to comment
Share on other sites

5 hours ago, Jay Rose said:

 

We've got better ways to measure loudness now, and deeper bit rates. But the old standard is still hanging on in ways... just like the deliberately slowed down QWERTY keyboard is still with us.

 

 

A good companion to you excellent book is "Mastering Audio, the Art and the Science" by Bob Katz and not only for people working on post. Except for a murky explanation of clock jitter and Internet file transfer in the first edition, concepts such as dynamics processing, digital audio, gain staging, etc, are very well explained.

 

And I liked the K-system for example. 

 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...