Ed's comment about 32 bit is the tell for today's digital mixing environment.
A large number of DAW's today use at least 32 bit float internally, there is NO clipping happening internally. And then there are those that use 64 bit float (PT) internally.
This was a very real concern in the past, but as computers have become faster and programs evolved it has faded as a problem if minimal attention is paid to running your system.
I love tape machines, I have 1 ATR 102, 1 ATR 124 and 4 Studer decks among others. But I hear more truthful audio coming off my digital systems, the audio I hear coming off my tape machines is interesting and very nice, maybe more seductive at times, sometimes less seductive, but I do not hear a loss coming off the digital return compared to the mic.
And of course the final question is how does it sound compared to the sound in the room? My answer is, a pale imitation is what the mic captures. The task at hand is then to make a recording that delivers the emotion of the music/event to the final listener. A task that so far no technology can do without considerable human intervention.
The last time I was in a studio was yesterday, and the day before, and the day before.
Cheers
Alan
A large number of DAW's today use at least 32 bit float internally, there is NO clipping happening internally. And then there are those that use 64 bit float (PT) internally.
This was a very real concern in the past, but as computers have become faster and programs evolved it has faded as a problem if minimal attention is paid to running your system.
I love tape machines, I have 1 ATR 102, 1 ATR 124 and 4 Studer decks among others. But I hear more truthful audio coming off my digital systems, the audio I hear coming off my tape machines is interesting and very nice, maybe more seductive at times, sometimes less seductive, but I do not hear a loss coming off the digital return compared to the mic.
And of course the final question is how does it sound compared to the sound in the room? My answer is, a pale imitation is what the mic captures. The task at hand is then to make a recording that delivers the emotion of the music/event to the final listener. A task that so far no technology can do without considerable human intervention.
The last time I was in a studio was yesterday, and the day before, and the day before.
Cheers
Alan
Last edited:
what I find today, and what PT tends to as a default, is
-16 dbfs = +4dbu = 0VU
personally I recommend -18 or -20 dbfs = 0VU, but many studio owners are still hesitant to use the more headroom settings and are stuck in the push it up mindset from the 16 bit days.
cheers
Alan
-16 dbfs = +4dbu = 0VU
personally I recommend -18 or -20 dbfs = 0VU, but many studio owners are still hesitant to use the more headroom settings and are stuck in the push it up mindset from the 16 bit days.
cheers
Alan
Nice as habit. Thank-you. Have a Merry Christmas.
I'm sorry my English is not good enough. But, even when i re read, it looks obvious for me.
BTW, you don't answered my question about your professional expérience ...
I'm no more in the business since ~10 years, now, but the last time i worked, it was for a french TV named Canal+.
The "PAD"* had to follow this norm for Audio:
A reference tone:
A1: 1000 Hz intermittent at -18 dBFS
A2: 1000 Hz continu à -18 dBFS
Then the maximal audio level had to be at -9 dBFS. (10 db lost for nothing)
* "PAD" is the file or the tape the lab send to the TV channel for future diffusion.
Last edited:
Hi Chris, Actually 24-bit and 32-bit are the same thing in terms of the resolution they offer. 24-bit integer has 24 significant bits. 32-bit float also has 24 significant bits. 24-bit DACs typically can work with either file format.
Thanks.
An other thing too. Since i was young, there is a confusion with the VU meters.
Some set them at 0 for +4dBm (1.23V continuous) out of the desk which i believe is a mistake.
Some to show 0dB for -4dBm to let the traditional 8dB of room to compensate their inertia (optimized for speech peak factor) .
Only the peak meters are supposed to show 0dB for +4dBm (reference level) at the out.
The choice of working level for 0VU is made for the installation. The old level for ABC, NBC, etc was +8dbm for 0VU. The most common industry level was +4dbm, but +12 and even +26 were used on occasion. There is no "correct" level, just the choice made for the installation.
in your last sentence, you state "Only the peak meters are supposed to show 0dB for +4dBm (reference level) at the out."
Bill's objection will be again that is should say 0VU for +4 dbm, not 0db for +4dbm.
and I would add, that today it should be dbu, not dbm as we typically do not use power matched systems. dbm is referenced to 1 milliwatt at 600 ohms load impedance, although other termination impedances may be used if specified. (50, 75, etc) Dbu is the voltage equivalent of dbm at 600 ohms.
There has also been a difference in tendency between Europe and the US. Europe has tended in the past to look at peak levels (ppm, etc) rather than the average level (VU). All of this is changing with the transition to digital, there are several competing systems for measuring loudness today, loudness units, etc. These are tending to be common across the pond. These are more directly concerned with broadcast, but tend to seep into music production. My new edition of Wavelab, now includes a loudness units meter I believe standardized for the EBU.
Cheers
Alan
Last edited:
I don't understand "for the installation". Can-you precise ?
(Agree with dBm Vs dBu, bad habit from me ...)
Don't agree with "Bill's objection will be again that is should say 0VU for +4 dbm, not 0db for +4dbm."
As i don't believe a peak meter give a dB->VU (for VU meter ;-)
Last edited:
building, studio, facility you are working in.
the tendency is to expect average when levels are concerned. So when you said 0db is -18dbfs, it was assumed by many, me included that you meant 0VU, Not something including a peak measure. In fact, almost all -dbfs calibration measures are done in rms, not peak level. RMS is not VU, but significantly closer than peak. Peak is peak and totally different measures and numbers are implied.
Last edited:
Ah, you mean during building the equipment ?
Well, some studios i used to work in had, halas, 0VU for 1.23V out on a reference tone. The first thing to verify when you work in, at it will change everything.
No, not during design of the equipment.
the choice of working level, it is generally archaic now, but it is still part of the standards. For instance, some Neve equipment has the designated working level of 0dbu, which means that if there is a VU meter on the desk it is 0.775V out at 0VU. You could recalibrate to 0VU = +4dbv, but you would have 4db less headroom.
It is a CHOICE made by the system designer when building the studio.
the choice of working level, it is generally archaic now, but it is still part of the standards. For instance, some Neve equipment has the designated working level of 0dbu, which means that if there is a VU meter on the desk it is 0.775V out at 0VU. You could recalibrate to 0VU = +4dbv, but you would have 4db less headroom.
It is a CHOICE made by the system designer when building the studio.
Off by a factor of two that you get for free with the floating point. ;-)
We can discuss recording methodologies until we're blue in the face but how many of us can actually do anything about it? Howie, Ed, and Alan excepted.
Good point, thank you.
Anyone with a computer can record, it has become widely democratized. And lots of people do. My daughter makes part of her living writing music for TV commercials and whatever else comes up. $500 for 30 seconds. No recording studio needed. Sometimes I help her out with some editing, or mastering. Help out a few friends too. Since I don't need the money I do it for fun.
Last edited:
Your English was fine, as Howie pointed out, your maths was wrong, and your units.
I am not sure why you are playing the 'what is your professional experience' card. Do you have to have >x hours in a recording studio in order to be able to do a dB calculation or read a specification?
Tournesol, I understand what you said, even if Bill quibbles with you about it. That is the way that I would express it, if I were relating it also to analog recording.
With analog recording in my day, OVu (as seen on a Vu meter) gave a +4dbm out, and we had to design for at least 25 dB above it in the record circuit to get just about everything. IF you clipped the electronics, even when the tape was mostly saturated (for all practical reasons) it would inject higher order harmonics into the tape, and could often be detected. We were not stuck on peak reading meters 30 years ago, but augmented Vu meters with peak reading LEDs: in my designs, green, yellow, red, set at 6db intervals, to detect what the Vu meter might ignore. Digital has taken another direction, and I am surprised that a mere 18dB is enough to eliminate clipping.
With analog recording in my day, OVu (as seen on a Vu meter) gave a +4dbm out, and we had to design for at least 25 dB above it in the record circuit to get just about everything. IF you clipped the electronics, even when the tape was mostly saturated (for all practical reasons) it would inject higher order harmonics into the tape, and could often be detected. We were not stuck on peak reading meters 30 years ago, but augmented Vu meters with peak reading LEDs: in my designs, green, yellow, red, set at 6db intervals, to detect what the Vu meter might ignore. Digital has taken another direction, and I am surprised that a mere 18dB is enough to eliminate clipping.
Back onto Rickie Lee and the track 'Ghetto of My mind' which has the highest DR* number:
DR22 0.00 dB -24.72 dB 6:16 03-Ghetto Of My Mind
Now I don't have any sharp tools in the box for analysing this, but here is what the blunt ones show. Attached are a couple of shots from Audacity, just with varying zoom. As you can see the percussive hits go waay above the rest of the program level. What I can't tell with a mk1 eyeball and the version of audacity on the PC I'm using won't calculate is if there is anything abnormal about the crest factor. Percussive sounds generally of course have a high crest factor, but I am intrigued as to how high.
The other interesting bit is that there does seem to be a significant degree of asymmetry in the percussion. So for those who can cope with the genre it's a possible for testing if you are sensitive to phase.
I would be interested in any comments from those versed in the art as this seems unusually uncompressed for a popular CD and an indication of what was thrown out in the last 10 years or so (this CD was 1989 and DDD).
* Assuming for now that DR is just two random letters assigning a measure of 'goodness' rather than having a meaning translatable to something else 🙂.
DR22 0.00 dB -24.72 dB 6:16 03-Ghetto Of My Mind
Now I don't have any sharp tools in the box for analysing this, but here is what the blunt ones show. Attached are a couple of shots from Audacity, just with varying zoom. As you can see the percussive hits go waay above the rest of the program level. What I can't tell with a mk1 eyeball and the version of audacity on the PC I'm using won't calculate is if there is anything abnormal about the crest factor. Percussive sounds generally of course have a high crest factor, but I am intrigued as to how high.
The other interesting bit is that there does seem to be a significant degree of asymmetry in the percussion. So for those who can cope with the genre it's a possible for testing if you are sensitive to phase.
I would be interested in any comments from those versed in the art as this seems unusually uncompressed for a popular CD and an indication of what was thrown out in the last 10 years or so (this CD was 1989 and DDD).
* Assuming for now that DR is just two random letters assigning a measure of 'goodness' rather than having a meaning translatable to something else 🙂.
Attachments
Thanks, J.C. Seems so obvious.
In fact, on my opinion, the question is: how long has a signal to be clipped before we can hear the distortion. I mean, can we hear a clipping in only half a wave at 3000 Hz ?
(About limiters or compressors, they have an "attack" time.)
Pro sound engineers know very well the peak/Vu relationship levels for most of the musical instruments. A second nature.
Of course, problems can occur with percussions. They are so diverses, so numerous, so different from each others that it is an entire world.
But,then, as we have digital meters that avoid any digital clipping to occur, we should be stupid to ignore them...
And, most , we are supposed to use our ears, in the same time ;-)
What I can say is this -18dB from 0dbfs (with 1.23V at the input of the recorder) was not so bad a compromise.
With a traditional peak meter (10mS) you will be safe most of the time whatever you record as long as you don't go deeply in the red. That is my personal experience. Of course, not with a Vu, even set at 0 for -8dB from the reference.
Well some digital recorders in the beginning of digital could had very bad behaviors when saturated. Do you remember ? We used to be more careful with them.
Now, if you have to record a kick drum, as an example, the method can change a lot from analog to digital. With an analog tape, some like to go deeply in the Vu-meter red and use the euphonics of the tape to *sculpt* the sound. No way with digital. No way.
There are a couple of freeware VST loudness meter plugins that should probably work with Audacity (using the VST Enabler):
SLM 128 - Free Loudness Meter VST 3 Plugin Released By Steinberg! - Bedroom Producers Blog
Youlean Loudness Meter - Free VST and AU plugin for PC and macOS
SLM 128 - Free Loudness Meter VST 3 Plugin Released By Steinberg! - Bedroom Producers Blog
Youlean Loudness Meter - Free VST and AU plugin for PC and macOS
May-be I can help ?
A (real) VU meter has an inertia. This inertia was fixed for it takes 300ms to the needle to reach 0dB Vu on a continuous signal of the reference level.
You understand that, if this same reference signal is shorter than those 300ms, the needle will indicate a lower level. (crest factor)
Because the crest factor of the speech was around 8dB ( 0dB on the VU brings ~ +8dB RMS on an oscilloscope) it was decided to tune the VU in order it shows 0dB when -8dB of the reference level.
A (real) peak meter has an inertia fixed to 10ms.
Led "crest" meters can have no inertia at all.
Some other instruments with memory, average calculation or more inertia can be used to give an idea of the "loudness" of a program.
Because, with a meter with no inertia, you can monitor the technical level on the tape ( or digital) but you have no idea of the subjective "loudness" of your program. Vu is quite good at this.
Last edited:
- Status
- Not open for further replies.