silence between words

I’ve seen this before but I don’t recall asking about it. I just extracted an audio book from audio CDs. Most of the pauses in speech, varying from under 0.5 seconds to several seconds, are digital silence. The transition is very abrupt between signal/no signal but sounds natural. There is no “pumping” or other artifact that I can hear. In some cases, seemingly a minority, pauses have some, essentially inaudible, background, measuring between -65dB and -72dB.

Is there an easy way to get this result? Is it likely to be a deliberate goal or the inadvertent result of doing something else?

Looks like a well set up noise gate to me.

... or well placed volume envelopes?

However you look at it, it's a result of incompetence - it's really annoying when people do that. The background noise (which should be present) should remain pretty much constant, at whatever level. When you do a project like that, the chances are that you'll end up with coughs, sneezes, whatever and what you should do is record a few minutes of 'silence' to cut up and use in those gaps - it's far less intrusive than BG noise in one gap, and what amounts to a bottomless pit in the next one.

Is there an easy way to get this result? Is it likely to be a deliberate goal or the inadvertent result of doing something else?

First question kinda impossible to respond to effectively isn't it?  4:27 AM here.  I just spent 2-2 1/2 hr. massaging what can succinctly, if inelegantly,  be called cross fades between a B3 and a rhythm guitar part processed through a manual (well, foot) wah effect.  Old school wah, boutique 18 watt class A amp (single 12) for which I spent much of yesterday trying to find the right mix of mics . . . ended up with an SM57 and an overpriced ribbon and am still not happy

anyway there were about six instances of call response between B3 and rhythm guitar that needed to be manually adjusted  envelops of the two instruments . . . Well tracks were recorded at different times  so in the transitions the relative levels required relatively subtle adjustment . . . (and there was one instance where the B3 came in on what I considered to be the wrong note . . .) took a couple of hours but for the budget that was pretty easy

did a lot of work with a guy, starting back in analog tape days, who used to drive me crazy manually ducking all the Pauses between vocal phrases . . . not talking about between choruses here but 'breath' between words and vocalist did not have to have abysmal microphone technique before this impulse kicked in . . . fortunately he was not paid by the hour, unfortunately other stuff he did was so immensely valuable that I either had to find work arounds or, occasionally, learn to live with it . . .  but apparently, with appropriate budget, it was pretty 'easy'

Digitally, by comparison it is ridiculously easy (but typically not necessary, or something that should be addressed prior to that point in an edit) . . . using a side chain of a compressor or gate probably requires nearly as much time to set up correctly (with greater variation in results) (certainly the first, second, third, fourth, fifth and sixth times one does that) as to simply manually nuke the stuff

course after nuking the stuff one is typically forced to lather on the digital reverb because human hearing evolved in such a way that most people find lack of spatial information to be irritating

anyway, pretty damn easy and  and you have not provided anywhere near enough information to even speculate as to whether what you noticed might have been intentional; quite possible that, that determination could not be made solely from the file in any case (as with projects with mentioned sound engineer: intentional on his part (though his results were consistent), the ducking was unintentional on mine, though I was intentionally not fighting him on it to access other aspects of his skill set  (and I make those type of trade offs every day all day, and some are easier the others).

I work in the other direction.  Of course, I try to minimize noise in the recording process, but there's going to be something in the background regardless.  If I have to edit in a gap, I replace the resulting silence with background noise copied from another pause.  Fluctuation in the level of the background noise is a distraction IMHO and not something you should strive for.

And often makes digital "restorations" of analogue audio or films unlistenable to for me

So that's three and a half votes for keeping the noise - albeit as unobtrusively as possible. That's all the justification Andy needs for getting rid of it completely, I guess... 

This was just another curiosity question, I don’t know how one could achieve the result, I wondered.  I can’t imagine someone having done this manually for the thousands of pauses in more than eleven hours of reading. I have no experience with extra hardware, such as compressors, so anything not software is likely to go over my head as useless to me.

This result can’t be from noise reduction; anything very extreme with that technique and there would be obvious ramp-up in noise approaching the next speech burst instead of the extremely sharp (on screen) division between silence and speech - at full volume. However this was done, it isn’t noticeable as any problem or distraction while listening. The parts that aren’t silent between words seems to contain normal vocal sounds. It is only the actual pauses that are silent; there is no room noise, no device noise, nothing.

I have no experience with extra hardware, such as compressors, so anything not software is likely to go over my head as useless to me.

Who said extra hardware? The dynamics processor can do this all on its own - no hardware required at all. Or don't you have that in your version of CE2000?

Oh, I thought oretez’s reference was to hardware use. So, this result can be easily achieved through the Dynamic Processing transform, without needing to address each pause individually?

Yes.  Set it up as a gate to compress when the level goes below a certain threshold.  Attack and decay will determine how quickly the gate responds to changes.

without knowing even the 'vintage' of recording AndyH was digitizing it's impossible to speculate accurately concerning 'how'  (where in manfc process this was dealt with, while tracking, in post production, etc.) let alone 'why' (if the artists, producers, label exec. were even aware that 'something' was done)

but my knee jerk gut reaction was that it was result of hardware gate in front of tape (or other recording medium), while the variation suggests that it might have been achieved via compressor side chain (a reasonably OK dedicated hardware gate can/could typically be set up (certainly in any half assed OK real studio) to produce reasonable consistent results, while structuring the process via a compressor algorithm I would general expect slightly more variability.  It was also pretty typical to automatically gate all low level non musical elements (i.e. speech)

but the question, if I remember correctly (which I might not), was 'how easy', whether it was intentional . . .

and whether hardware to tape or digital (or even analog) dynamics processing (CE2K has the ability to do this) in post production it's pretty easy, particularly if 'close enough' is good enough.  And while I'd find it a bit unusual for this to be subject of a committee meeting to analyze the process . . . inserting the gate would typically, consciously, be thought of as controlling, taming unruly audio not, typically, with the conscious idea, 'Oh, say why don't we eliminate all noise between words'.  Whether it was clear or not most of the people with whom I've worked (outside of an occasional unique obsessive/compulsive) would not perceive this as an appropriate goal ('noise' was a very different goblin in analog universe . . . in digital universe once we figured out how to effectively eradicate all noise, turned out that was not exactly the holy grail we had hoped it was going to be)

so: easy and at least an unintended consequence of intentional choice (as a knew jerk first & second guess)

. . . in digital universe once we figured out how to effectively eradicate all noise, turned out that was not exactly the holy grail we had hoped it was going to be)

The simple truth of the matter is that complete silence is an anathema to the listener.

I spent a good few years as a dubbing mixer and it was almost a standard practice to add some 'room noise' (usually take as a wild track and often electronically looped by me) into tracks where little noise was actually present.  The brain seems to be able to easily ignore such 'noise', whereas it objects, quite violently, to a complete absence of noise during a 'silence'.  The pumping of the background noise level is what most people object to.  The basic object of the exercise was to replace the dead silence between words/phrases with a noise that equated to the background noise of the speaker - this being far more acceptable to the listener.

. . . in digital universe once we figured out how to effectively eradicate all noise, turned out that was not exactly the holy grail we had hoped it was going to be)

The simple truth of the matter is that complete silence is an anathema to the listener.

I spent a good few years as a dubbing mixer and it was almost a standard practice to add some 'room noise' (usually take as a wild track and often electronically looped by me) into tracks where little noise was actually present.  The brain seems to be able to easily ignore such 'noise', whereas it objects, quite violently, to a complete absence of noise during a 'silence'.  The pumping of the background noise level is what most people object to.  The basic object of the exercise was to replace the dead silence between words/phrases with a noise that equated to the background noise of the speaker - this being far more acceptable to the listener.

Couldn't have put it better myself