Audio Meter and Level : Calibration test tones ( download )

[Steve_Karl]

Here are some test tones useful for calibrating audio meters and also setting levels.

This morning, when getting my Zoom H2 working as an external mic feeding my HV40, ... I remembered making these yrs. ago, and will be using the 10 second tone at 1k 0dB to set the HV40 input level. I'll just load it onto my H2 and have it there when ever I need it.

The idea is to play this file on the H2 and set the input level on the HV40 very slightly below 0dB.
This will guarantee no clipping on the cameras recorded audio.
Of course the front end of the H2 might need watching for possible peaks.

Currently, after a short test this morning and for dialogue in a normal speaking voice, at about 2 feet from the camera, I have the H2 inputs at maximum and the output feeding the HV40 at 70. ( 70 whatever on the readout of the H2 when adjusting the playback volume )
Watching the LCD on the HV40 I could see that I could clip by clapping my hands, but the sound of the voice was great.
This is before remembering the test tones and before any leveling with the 1k tone.
I'll be getting to that an other day.

There are also other tone ranges that might be useful to some doing advanced audio work so I'm including them also.
( See attached image for details )

Today, I doubled the length of the 1k tone to 10 seconds so that one is not shown in the older picture, but is included in a .zip by itself, because it's probably the most useful one of the bunch.

http://www.sightsea.com/music/1K_10_Sec_0dB.zip

http://www.sightsea.com/music/ALL_Test_Tones.zip

Caution: These tones are at 0dB which is as HOT as it gets. Be very careful of your ears, speakers and other gear when using these. Turn your volume knobs down before playing them. These could easily destroy speakers and also cause permanent damage to our ears if we're not cautious.

[Janke]

Hmmmm - you should never go to 0 dB in digital recordings. In fact, TV stations here require that you limit everything you deliver to them at -9 dB peak...

The usual suggestion is to keep the general level at -12 dB, then equalize in post. If you record with a H2 or similar recorder in 24 bits, you have enough bits to boost as much as you ever want...

[Steve_Karl]

I'm not suggesting recording at near 0dB, i.e. unless I have the gear to control it, but there's nothing wrong with doing a 1 hour tape that occasionally get's as hot as -0.1 dB. The ONLY real issue is to never hit 0dB i.e. in ones recording, and on a piece of hardware like a Camera or a Zoom. All else is fair game, i.e. unless one is not using enough of the 16 or 24 bits to get a decent recording, which is quite possible on a zoom at 24bit and very easily possible on an HV40 at 16bit.

For a calibration / test tone 0dB is fine and perfect for my needs. These are not recordings.
They're all pure sine waves synthetically generated using Sound Forge which is a controlled environment.
They are not recordings. They're were synthetically generated and not recorded.

Open any of them in a .wav editor and you'll see there are no truncated peaks. They're all perfect.
If someone needs them to be -.1 dB or -9dB then they can be changed with no harm to the audio.
And of course I'm not suggesting that anyone send these 0dB test tones off to their local TV station or record at 0dB.

Every venue has their own requirements.
That's usually designed to keep the uninformed from making their job harder or impossible.

My requirements are nothing ever hotter than -0.1 dB but that's because I do my own mastering. If I can get closer to 0 with out ever hitting 0 then there's no problem with it.
And, If I have to send it off else where, I have no problems meeting their requirements, and even if I did screw up and hit 0 in a recording, it won't be a problem unless one can hear the clipping. I've often fixed problems like that with the pencil tool in Sound Forge for clients or friends.

24 bit recording does give you more bit range, but the extra 8 bits is at the bottom of the stack, i.e. relative to 16 bits, and both 16 and 24 bits can never be boosted above 0dB. 0dB for both, is exactly the same 0dB.

24 bit's can not be boosted more than 16 bits, since 0 is 0 for both, however 24 bits gives one more dynamic range and distance from problems in the very low volume ranges, particularly in very quit passages when fading out to infinity.

You're probably thinking of working in a DAW or NLE and seeing that it's ok to clip a 24bit track as long as one never clips the master out. However, that clipped 24bit track still never gets louder than 0 dB.

Nothing exists above 0 or below infinity in the realm of audio.
They are absolutes like pure black and pure white.

-------
Foot note:
I originally made these to calibrate the meters in my Soundcraft 400B console
to get a reference in relationship to the output of my 4 computers 8 sound cards. (***) With 8 cards and sometimes more than one stereo out per card I needed to be able to critically evaluate the signal levels and balance them accurately at the analog console.

My audio rig consists of 4 PCs and 8 Sound cards.
My master work station has 2 Echo Layla 24 cards and 1 Echo Layla 3G card.
Those 3 cards in the master machine receive a digital signal from the satellite machines over fiber optic cable.

The other 3 machines (running Giga Studio 3 and also other various VSTi ) have either Echo Layla or Echo Gina cards which output 24bit / 44 over the toss link to the mater machine when rendering / capturing the final audio for the master recording.
For monitoring ( during composition ) they output 24/44 to the analog ins of the Soundcraft console. (***)

I most always render / capture my audio at 24/44.
If I need to be 16/48 I just output the mix out of Sonar at 16/48.
Sonar can resample during mixdown with no problems.

[cdlynch]

24 bit recording does give you more bit range, but the extra 8 bits is at the bottom of the stack, i.e. relative to 16 bits, and both 16 and 24 bits can never be boosted above 0dB. 0dB for both, is exactly the same 0dB.

24 bit's can not be boosted more than 16 bits, since 0 is 0 for both, however 24 bits gives one more dynamic range and distance from problems in the very low volume ranges, particularly in very quit passages when fading out to infinity.

You are correct that 0dB is the same for 16 and 24 bits, but it is not correct that the extra 8 bits is at the bottom of the range. Increased bit depth distributes added resolution evenly through the entire range of possible sample locations. The extra precision can be measured at any amplitude, not just quiet passages.

When Janke suggested recording so that you peak around -12 dB, he's following the professional standard. It allows you to avoid accidental clips (pencil editing in Sound Forge can help, but is generally an insult to sound no matter how steady your hand! Adobe Audition has a much better solution, but clip avoidance is always the best option) and you don't have to worry much about loss of detail. Even 12 dB down, you are still working with FAR more resolution than a 16 bit file.

Consider that every time you add one bit, you DOUBLE your resolution. 16 bit is 65,536 (2^16) possible sample positions for amplitude. 17 bit is 131,072 (2^17), and so on. By the time you've doubled your way to 24 bit, you have 16,777,216 (2^24) sample positions. 16 bit has .39% the resolution of 24 bit, so 12dB isn't much to worry about.

You're probably thinking of working in a DAW or NLE and seeing that it's ok to clip a 24bit track as long as one never clips the master out. However, that clipped 24bit track still never gets louder than 0 dB.

What? It's never OK to clip a file regardless of its relation to the master out, unless the style calls for nasty odd order harmonic distortion. The file is still clipped even though it's below 0dB in the master out, and you won't see flat tops if you've mixed it with other sources, but the damage is there. If you gain stage properly, no track will clip, including at every plug-in on each track, and the sum of all your tracks won't clip the output either.

[Steve_Karl]

Understood, and I stand corrected abut the added resolution over the whole range in a 24 bit recording, but when it comes to just seeing and hearing audible levels ... ahhh never mind. It's of no consequence really.

And, I totally understand the recommendation of playing it safe at -12dB when recording but if I can get consistent non clipped recordings with my upper limit showing as a higher level, then there's no reason not to go for it, especially since the HV40 is stuck at 16 bits.

Quote Originally Posted by Steve_Karl View Post
You're probably thinking of working in a DAW or NLE and seeing that it's ok to clip a 24bit track as long as one never clips the master out.

I didn't say clip a file, I said a track, meaning a track level meter. Most all DAW software now days is working at 32bit (or 64 in some) internal resolution. You can see truncation at the top in the audio view for that track, however I've never heard anything other than a limiting effect when and if that happens.
I was just trying to find a logical interpretation for how janke might think that:
"If you record with a H2 or similar recorder in 24 bits, you have enough bits to boost as much as you ever want... "

Pencil editing is certainly less of an insult to *ones ears* than an audible click. I have no worries about insulting the inanimate object we call sound.

Obviously ( very obviously ) it's better not to clip at all, but I'll be finding the limitations of the gear as I test it myself rather than just playing it safe at -12dB, and mainly because of the HV40 is 16bit only.

[cdlynch]

OK, I get your track clipping thing, as long as the DAW in question processes with 32 bit floating point. Different DAWs will do it differently. Some might require that the actual wav file be 32 bit float for that to hold true. Also, not all plug-ins operate in 32 bit float, and might convert on-the-fly to 24 bits, which will clip any overs before being passed to the next effect, or bus. Best practices for gain staging still apply unless you can verify EVERY technical detail of your DAW, and every plug-in you use. No small feat.

Wait a minute, you're using the external recorder just to get sound into your HV?!? That's MADNESS!!! The HV cams have terrible sound with unnecessary processing, even if you don't use their mics!

Yes, there's some benefit to using the external recorder instead of the raw HV cam, but using it that way you're going through two stages of not-so-great AD/DA and ending up with 16 bits, just so you can avoid synchronizing in post! I realize not everything needs to be hi-fi, but you're not doing yourself any favors. If you got the external to improve your sound, you certainly have the option to improve it even more by changing your workflow a bit.

[Steve_Karl]

Wait a minute, you're using the external recorder just to get sound into your HV?!? That's MADNESS!!! The HV cams have terrible sound with unnecessary processing, even if you don't use their mics!

Yes. I suspected as much the first time I captured and it's the main the reason I wanted to stay in the upper 6dB range or higher for these reference tracks. Nothing is a final audio take with what I'm doing now, since it's all music video oriented. I'd never use the HV40 to record any audio that I know was going to be used in the final.

I know I can record on the H2 (even also, and both) and easily sync. but this idea is just to quickly improve the stuff I'm doing now by using the better mics in the h2 and not having to deal with the extra sync step. We're just collecting media right now and I spent 20 minutes setting it up as it is now, to save me hours in the next few months.

The H2 wind screen is a perk also.
We shot some stuff on a high hill over looking the Allegheny River a few weeks ago and the wind noise was sometimes wiping out the reference audio being played from a boom box.

Sonar is 32 bit floating point and I never use plugins on the track insert. There always on a bus.

[cdlynch]

Ah, I get it now, sorry! I retract my claim of madness! But if it's all reference audio, why worry about noise floor and clipping enough to use calibration tones? Just play your reference audio, set the level to say, -4 dB, and party like it's 1999!

Sonar is 32 bit floating point and I never use plugins on the track insert. There always on a bus.

Even if they're on a bus, if the plug-in on it doesn't support 32 bit float and your send has overs, you'll clip the plug-in input.

[Steve_Karl]

I've clipped plug in inputs often enough to know it never produces audible problems.
The calibration tones will be used to be actually able to set the levels to where I can get a good level and insure no clipping at the HV40.

I've spent much more time trying to explain it to you than to do it.