Adjusting Levels, Recording / Post-process

[himd_user]

I have listened to some of the audio clips in the Gallery. I have also uploaded a few of my own. Hope to upload more soon.

One thing I am curious about is adjusting levels. Several of the recordings in the Gallery were VERY SOFT. Had to crank my PC speakers up all the way.

Usually I record with a lot of head room. I do this out of habit from recording in SP mode on an MD deck, where I noticed clipping really sounds BAD.

So with HiMD, I record with a lot of room (maybe -6 dB max) then upload to PC, then edit in Cool Edit Pro, and hard limit with amplification to get the levels to hover between -12dB and -4dB. Even for voice interviews, this is what I do, and it sounds OK.

However I recently ripped a Tool(hard rock band) CD to WAV , and noticed the levels are near constant peaking between -3dB and 0dB. No headroom! All red! All the time! And still sounds crisp.

Should I be amplifying my recordings and saturating with hard limiting (i.e. "compression"?) so that the audio signal hovers i.e. between -4dB and -1dB ?

If anyone has experience with "production" level post-process, please comment?

. himd_user

[dex Otaku]

One thing I am curious about is adjusting levels.  Several of the recordings in the Gallery were VERY SOFT.  Had to crank my PC speakers up all the way.

Some of us are more 'purists' than others. I think I tend to ride the fence on this; it depends on the nature of the recording.

With most recordings that I want to maintain a "natural" sounding dynamic range with, I use RMS normalisation at -20dBfs. This maintains a lot of headroom over the average level of the recording, and matches the fashion with which most classical CDs are produced [-20 to -18dBfs, which is 0VU on AES/EBU equipment's meters if I'm not mistaken].

To point something out that most people don't ever realise - the volume control on any stereo system is an attenuator. It subtracts from the signal going into the amplifier. If you turn it 'down' all the way, you're subtracting 100% of the signal. If you turn it up all the way, you're subtracting nothing.

In reality, your amplifier is always amplifying at the same level, it's just what's going into it that's altered [otherwise the volume control would have to have massive heat sinks on it and would likely cause fires in most living rooms].

The thing to remember is that if you're using the volume control, you're subtracting from the signal, and you're reducing the dynamic range of the content along with it. The amp's self-noise is actually a constant - hiss that gets louder as you 'turn up' the volume means that your amp has cheap, noisy op-amps in it that are generating that noise. The amp itself isn't the source [of that in particular.]

It's difficult to say there's any kind of standards with music or stereo systems, but to me listening to a natural-sounding recording means you can have 0dB attenuation [the volume all the way 'up'] and the 'normal' passages will be at a comfortable level. Quiet passages will actually be quiet, and loud passages will be as loud as your amplification and speakers can handle. I have actually mastered recordings [meant to be played on the system they were mastered on] this way in the past.

Pop music isn't meant to be that way, nor are radio broadcasts, voice recordings in general, &c. The only recordings that come close to having a potentially natural dynamic range are film soundtracks. Sound for film is heavily standardised [and has been, within certain limits, since the 1950s or so, before THX was ever even a thought in anyone's head] to maintain the same volume level for dialogue regardless of what venue the film is being played in. Later other standards were set to ensure that a certain amount of amplifier/speaker SPL headroom was always available for sound effects and music. This is not to say that every theatre followed the standards loosely let alone strictly, but at the least the standard for dialogue [at around 72dB actual SPL] has been around and loosely maintained for a long time.

The same does not exist for music, as playback systems vary wildly in terms of capability, as do listening environments. Going by today's trends, one would think that mastering engineers expect that nearly 100% of music is going to listened to within 20m of a construction site or aeroport nearly 100% of the time.

Note that I don't advocate reaching for a natural dynamic range with most listening material. Whatever is appropriate for the material in question is what's appropriate for it, and that ends up being a matter of opinion as well as fitting things within the limitations of its carrier system [like compressing music for playback on FM radio].

I record with a lot of room (maybe -6 dB max)

Wow. 6dB of headroom is almost nothing, actually. I usually record with at least 15-20dB headroom, i.e. levels averaging [not peaking] one or two segments -below- the first dot on HiMD's meters [which is the -12dBfs mark, and for all intents and purposes, MD and HiMD's 0VU mark].

. . . and hard limit with amplification to get the levels to hover between -12dB and -4dB.  Even for voice interviews, this is what I do, and it sounds OK.

Sounds reasonable, though I personally prefer not to hard limit anything if I can avoid it. Two words: listening fatigue. On highly dynamic and very short passages, sure, but otherwise all it leads to is stress to your ears [and your equipment].

However I recently ripped a Tool(hard rock band) CD to WAV , and noticed the levels are near constant peaking between -3dB and 0dB.  No headroom!  All red!  All the time!  And still sounds crisp.

It's called bit-pushing and has been the trend in mastering since the late 1990s. Basically, they reduce the dynamic range of the recording to almost zero using dynamic compression and limiting, which has the benefit of raising the average volume of the recording so that when played on cheap equipment it sounds loud and 'clear'. The gist in competition is to make things sound as loud as possible, most specifically to make it sound louder than whatever other band's record.

The end result is that CDs are apparently much louder, have virtually no dynamic range, are full of readily measurable distortion, and cause severe listening fatigue after only brief periods of exposure.

Severe compression and bit-pushing also make sound more difficult to compress with lossy codecs without audible distortion or artifacting, incidentally.

Should I be amplifying my recordings and saturating with hard limiting (i.e. "compression"?) so that the audio signal hovers i.e. between -4dB and -1dB ?

If anyone has experience with "production" level post-process, please comment?

If you want to make pop records, sure. If you want to make listenable recordings that don't cause people's ears to bleed and amplifiers to overheat, then no.

I don't advise bit-pushing. It just sounds like crap to me. Most of the recordings I've purchased in the past 8 years have left me wondering what the studio engineers' and the musicians' actual "vision" of the recording was before the mastering engineers took it and mutilated it so severely that it became unrecognisable - but hey, it's radio-ready, and it sounds loud on the average underpowered transistor radio, car stereo, boombox, portable player, computer, home stereo... you get the picture.

I, personally, don't like having to cut the band around 4-8kHz by as much as 15dB to make modern CDs listenable without causing me physical pain.

[dex Otaku]

Before I forget about this completely, I'll try and actually answer the question in a way that is relevant as well..

For voice recordings, I tend to RMS normalise to pretty high levels. Sound Forge's defaults are a reasonable average in industry terms.

The following presets are there:

* Normalise RMS to -10dB [speech]

* Normalise RMS to -16dB [music]

* Normalise RMS to -6dB [very loud!]

There's another one there, too:

*Normalise RMS to -20dB [dex]

Heh.

[greenmachine]

I consider Audition's RMS histogram to be very interesting:

The first graph shows a very loud, dynamically highly compressed heavy metal track, the second relaxed pop music. Note the difference in the dB scale.

[attachmentid=403][attachmentid=404]

Edited June 25, 2005 by greenmachine