a form of vocal pops?

(1/1)

AndyH:
This is two selections from a recent transfer. I’m curious to know if the particulars I’ll point out are the result of recording technique, an equipment artifact, a normal aspect of human speech that I just don’t manage to recognize outside of recordings, or whatever. I would also be interested to know if someone has a clean-up technique easier and faster than manual treatment of each individual instance. That is, a technique that quashes these things but doesn’t effect the rest of the recording noticeably.

This is from cassette, I’ve seen the same thing in extractions from CD and, at least to a small extent, from LPs. The problem, in my experience, occurs mainly in speech rather than singing, which could well explain the relative scarcity on LP transfers. I’ve extracted a good many audio books from CDs, but not many from LPs. This particular cassette source has them by the hundreds, if not thousands, and thus is especially obnoxious in this respect, although of considerable interest otherwise.

The problem is transient noise spikes, occurring in speaking but not, I believe, part of the normal result of speaking, at least when away from a microphone. More often than not, these transient spikes comes in little groups rather than just one at a time.

If your decoding of mp3s matches mine, the first instance is several spikes between 0:00.200 and 0:00.887. In the next snatch, between 0:03.575  and 0:03.868, although there are lesser occurrences elsewhere. I get 0:05.303 as the total length.

They are probably easier to hear as part of the entire sample rather than by selecting only the ranges I’ve pointed out. They are quite readily visible close up in Spectral View.

SteveG:
Sounds like sibilance to me. Since it comes out of human mouths, usually accompanied by a blast of air, then microphones are rather more susceptible to it than other humans (who generally tend to be further away) tend to be. It is a human failing - one that can often be trained out of people - but there are also microphone techniques that can minimise it, and it doesn't sound as though anything in particular was done about it on these recordings.

The problem with treating it is that unfortunately, the 's' sound you actually want is in exactly the same range as the hiss that accompanies it, so it's very difficult to remove effectively at all without losing some, or all, of the articulation. I've yet to hear a treatment that doesn't either sound worse, or make no real difference to the audio. There are all sorts of plugins available, and even attempts in Audition to filter it, but they all come up against the same issue in the end - it's really difficult to differentiate between the wanted and the unwanted sound. Some speakers are so lazy from this POV that it all rolls into one regardless. And in a perverse way, that's easier to deal with - just locate the relevant band, and reduce it with EQ.

AndyH:
This reader may have had little experience speaking in front of a microphone. At any rate, he probably had no training; I doubt if recording was his thing.

All of the many instances on this set of four cassettes seems easy to fix, but quite tedious. Perhaps that is what you mean: who would go to so much trouble. Fill Single Click seems to me to do the job without any downside. I have fussed with recorded singing where it wasn’t so straightforward, however. I was just hoping there was an easier (i.e. bulk processing) way (but didn’t bank much on the possibility).

Navigation

[0] Message Index