Also, you’re not booming any more. That problem vanished with the shiny walls and the echoes. As I posted, all of those rescue tools can go away.
Koz
Also, you’re not booming any more. That problem vanished with the shiny walls and the echoes. As I posted, all of those rescue tools can go away.
Yes, this is a great improvement indeed.
Do you know why you’re clicking?
Unfortunately, no. For the raw audio tests, the NTG3 was on a solid stand, about 10 inches high. So I’m guessing it is me. Is the clicking still there? I’m not sure I can hear it myself: https://forum.audacityteam.org/download/file.php?id=31669
I’m not sure I can hear it myself:
Now you’re scaring me. Audacity doesn’t work in the subjunctive mood.
“We might require click removal had the clicks been audible.”
You posted a mono (one blue wave) track. If you post mono, you can announce out to 20 seconds before the forum cuts you off. Post a WAV file of you reading one of your own works in the way you would want to present it to a customer/client. Leave one clean second at the top instead of two.
Koz
I’m wondering if those clicks were from my computer, now distanced and under a cardboard box?
Alternatively, perhaps they are introduced by the change pitch? I do prefer some reduced pitch, and apparently my students do too. I’m not proud. Whatever works best.
Here is a longer test, raw no processing.
Here is the same test, with the only processing being a change of pitch of -4%
Still clicks, or clicks all gone?
They usually correspond when your voice is loud, so not extraneous (e.g. computer) noise from the in the room.
They are on the raw version, so not caused by pitch shifting, (but will become more obvious by shifting down)
Sounds like clicking within the microphone.
Can be fixed with Paul-L’s declicker plugin …
I’m putting together my chains.
The main target for this work is Udemy, where students are used to louder audio. So I have had comments mine was too low, which has review rating and other negative implications, so would like to make mine as loud as (acceptably) possible.
Would it be ok to RMS normalize to -18 dB and then soft limit to -3.0 dB?
That may be a job for Chris’s Compressor. Chris wrote this processor so he could listen to opera in the noisy car without constantly turning the volume up and down.
https://theaudacitytopodcast.com/chriss-dynamic-compressor-plugin-for-audacity/
Notes:
– I change the first value, Compress ratio from the default 0.5 to 0.77. That makes the show louder and denser.
– It shows up under the effects menu as Compress dynamics.
– It doesn’t like running off the end of the file, so add “something” at both ends of the work, run Chris and then cut it off later. My show was part of other material, so I never noticed the end-file thing.
I use this on a talk show where the hosts have very different voices. It mops up the differences and makes the whole show louder, all without me guessing at compression “knee” points and volume slopes of other compressors.
Give it a shot.
Chris won’t be improving it any more. He reached “end of life” a while ago.
Koz
IIRC Chris’ is only a one-band compressor.
Gmulti has 3, (#3 can be used to de-ess)
Gmulti allegedly can be used on Windows, Mac & Linux,
but I can only testify the 32-bit version works in Audcaity on Windows.
I’m leery of compression mainly because I’ve read about it taking out some of the natural feel, and the relative emphasis of syllables, words, and phrases is so important in the kind of otherwise dry material I’m covering.
Would there be a downside to just RMS normalizing to -18 dB, and then soft limiting to -3.0 dB? In the tests I’ve done, to my ear, I can’t tell any damage from the soft limiting. But then again, we are talking about a barely adequate ear.
Here is an example to see what the result would be. Same audio as last sample, with the following:
- Declicked, settings as per Trebor post
- Change pitch -4%
- Equalization with low rolloff for speech
- RMS normalize -18 dB
- Soft limit -3 dB
Acceptable, or too loud?
Acceptable, or too loud?
Acceptable is up to you and the customers. You don’t have to meet anybody else’s standards, so it’s up to you.
You could also soft-limit louder, say -2dB rather than -3dB.
Soft-Limiter is not accidental. There are other harsher and stiffer limiters. In extreme or with the wrong one, you could start to sound like an air traffic controller. Soft Limit is like wrapping the limiting action with many layers of flannel. It’s working, but you can’t tell by listening what it’s doing.
One of the goals of the Mastering Suite was that you still sound exactly like you, except you pass audiobook standards.
Koz
That’s -16.9db RMS, (rather than -18), which is little bit too loud for me, (& my volume is only on 25%).
When the complainant(s) said “too low” maybe they meant too bassy, rather than too quiet.
It is possible to make the voice less intelligible by boosting the bass.
[For some reason, (mic rattle ?), “accountability” sounds like “accountRability”].
For some reason, (mic rattle ?)
Fair warning, if you apply enough effects, filters, modifications, and corrections, you may get tools fighting each other and odd artifacts start to crawl out from under the rocks.
You may start to run out of fidelity. 44100 sampling frequency used in Audio CDs was a compromise. It’s only unconditionally stable out to just over 16,000Hz sound, out of 20,000Hz, the highest possible audible sound pitch. It uses electronic tricks and hearing limits to cover the difference. Studios use 96000 sampling to record your Grammy award-winning album music, not 44100.
That’s still not dreadful, FM radio (remember that) sound only goes out to 15,000Hz, but you used pitch shifting.
**_- Declicked, settings as per Trebor post
- Change pitch -4%_**
Suddenly, all those high pitch errors safely hidden up where only dogs can hear them might be audible to people. I bet that’s where the “mic rattle” is coming from.
Koz
Well, I’ve learned a couple things.
That’s -16.9db RMS, (rather than -18), which is little bit too loud for me, (& my volume is only on 25%).
The entire track is -18.24, but the excerpt is as you found -16.9. Of course… it is an RMS. This learning argues for trying some compression after all. I will experiment a bit. Should this could come before all the other processing?
Btw, my guess as to the reason people like courses loud on Udemy is because many of the courses are loud so they get used to it, and because probably at least half listen on portable devices with often tiny speakers.
Change pitch -4%. Suddenly, all those high pitch errors safely hidden up where only dogs can hear them might be audible to people. I bet that’s where the “mic rattle” is coming from.
Aha. I thought carefully before buying this expensive NTG3 microphone, and have been confused about how it could have clicking. Now things make sense. So, assuming that the pitch change provides so much improvement over many hours of listening that it is worth it, which I believe I’ve validated from many students listening to previously made audio, should the declick then come at the end of the chain?
The entire track is -18.24, but the excerpt is as you found -16.9. Of course…
The tools work on overall averages. It’s always going to be possible to find “celebrity” passages within any one chapter or performance. The metaphor is someone telling you a fascinating story over cups of tea. It doesn’t work if the performance stops for several seconds while they put the kettle on.
That performance is going to be miles away from the overall average goal if the tools are applied.
That’s why I suggested Chris’s Compressor or other similar tools. They “know” when the content doesn’t lend itself to simple processing and they make adjustments.
reason people like courses loud on Udemy is because many of the courses are loud so they get used to it, and because probably at least half listen on portable devices with often tiny speakers.
So maybe the dense, aggressively-processed Air Traffic Controller model is the best bet.
“Continental twenty-six heavy. Clear for departure two-four left.”
Koz
half listen on portable devices with often tiny speakers.
I again question the wisdom of lowering the vocal pitch. The Waze app in your car is not a gutsy, broadcasting, male voice. It’s a woman’s voice.
“You have arrived at your destination on the left.”
Much easier to deal with in earbuds, too.
So obviously you need a sex change.
Koz
I again question the wisdom of lowering the vocal pitch.
I totally get the advantages of an unprocessed voice. In general, the less done and more natural most things are, the better.
However, nothing counts more than actual results, and if you listen to both the unprocessed and pitch lowered samples here:
https://forum.audacityteam.org/t/searching-for-an-effect/192/1
The hands down winner for me, as audio for a course of between two and eight hours, is the second. Just so much easier on the ears. Since it is my voice, I can say it: less grating, more relaxed, more pleasant.
So obviously you need a sex change.
But hey, other than my voice, I already love myself, so what would be the sense in that!
I’ll check out the results with compression, and post back (could be a few days).
Hi
Have you tried using a Lavalier Microphone.
Have you tried using a Lavalier Microphone.
Yes, I tried several. I could not find one that worked well. I also found they had no direction capabilities, so picked up any sound in the vicinity, even if quite far away.
Lavaliers work by being close to the performer, but most of them are non-directional and will pick up the whole room. Pretty much the exact opposite of the shotgun you’re using.
There was a show called (I think) “Secrets of New York” that used a directional lavalier. That was the only way they could get narrator voices recorded in the middle of Grand Central Station (actual show). Directional lavaliers have a down side. If they slip on your clothing, you’re dead.
Real news studios work well, but the early Rachel Maddow shows were not using a studio. Her lavalier microphone picked up every little sound in the room they were using. I accused them of recording on the NBC loading dock and one show featured someone assembling an Ikea credenza in the background. They started using a real-time noise gate and I use those shows as examples of how not to use a noise gate.
They don’t do that anymore.
Koz