Fledgling voice artist seeking counsel.

[Robert J. H.]

It might be helpful to see how white noise is modified after a pass through the whole recording system.

I've prepared a test file that you can play as input to the microphone (with some kind of playback system of course).

Spectral-test-patterns.flac

(933.72 KiB) Downloaded 49 times

It includes:
- Impulse train
- repeated sweeps, 100 to 10000 Hz
- Harmonic noise (all frequencies from the Audacity graphic EQ; 40 to 16000 Hz)
- White noise
It is all adapted for 48000 Hz (recording side) and 8192 samples (plot-spectrum)

I'll be dashed if none of them shows the peaks/troughs I'm suspecting.

Of course, the playback adds some non-linearity to the result. The smaller, the less low frequencies will be radiated. That's no problem for e.g. the impulse train.
Ian, I hope you'll find the time to make the test. Try to go as near as possible to the 0 dB limit while recording in order to have the highest signal-to-noise ratio.
Thanks very much.
RJ

[MichloIW]

It might be helpful to see how white noise is modified after a pass through the whole recording system.

I've prepared a test file that you can play as input to the microphone (with some kind of playback system of course).

Spectral-test-patterns.flac

It includes:
- Impulse train
- repeated sweeps, 100 to 10000 Hz
- Harmonic noise (all frequencies from the Audacity graphic EQ; 40 to 16000 Hz)
- White noise
It is all adapted for 48000 Hz (recording side) and 8192 samples (plot-spectrum)

I'll be dashed if none of them shows the peaks/troughs I'm suspecting.

Of course, the playback adds some non-linearity to the result. The smaller, the less low frequencies will be radiated. That's no problem for e.g. the impulse train.
Ian, I hope you'll find the time to make the test. Try to go as near as possible to the 0 dB limit while recording in order to have the highest signal-to-noise ratio.
Thanks very much.
RJ

Of course, I'll try that this evening.

Cheers.

[kozikowski]

I need to go back and remember how I made the popping noise. I have the project, but Audacity doesn't remember the effects settings or UNDO.

Attached is the Project screen for AMRadio5. There's a lot going on and I didn't clean up after myself, so all the trials are still there.

I do remember the conditions for the voice (track 5).

Record the vocal track clean and apply Effect > Equalization: AM Radio > OK.
Then Normalize to -1.

That's the basic AM presenter voice. I agree with flynwill that it needs to be a little dirtier. No AM station sounds like that, so try Effect > Leveller at Light or Medium. I got a similar pseudo-processed effect by leaning into the microphone while speaking (that's my MacBook Pro internal microphone at night in the garden). The up side is no room ambience or echoes. But then there's the cars and crickets.

Track 2 in that picture is the 5KHz sine wave whistle. Generate it and then Effect > Amplify to a new peak value of -35. Change that number to change the presence of the tone in the show — or don't do it at all if you think it's distracting.

Track 3 is the static. The simple narrative version is Generate > Noise: brown. Then Effect > Noise Gate which is then cranked to evil settings you would never use for quality audio. The occasional pops are the extreme peaks of the noise sound with everything else stripped off by Noise Gate. I think Effect > Amplify it so they were louder in the show. I need to get you the particulars about that. That, too can be tuned for theatrical effect.

Audacity will smash all the active tracks together when I export. The gray tracks will not be part of the show.

Koz

[MichloIW]

Thanks again, Koz.

It won't be until tomorrow but I'm looking forward to a bit of experimenting.

Cheers, mate.

[MichloIW]

Greetings,

three Spectral Tests are uploading right now to OneDrive.

Unfortunately I had forgotten that I've started hearing a little bit of noise from either downstairs or next door since they are now using their air conditioner(s) so I hope that doesn't invalidate the tests.

I was going to play it from my tablet at first but from what I read it wouldn't have done justice to the sounds being played. In the end I went with holding my laptop up to the pop filter and playing / recording on there whilst I held my breath.

I hope it helps.

https://onedrive.live.com/redir?resid=F ... =folder%2c

Cheers.

[kozikowski]

How are we measuring ACX Compliance? I see through searching that we seemed to have settled on WaveStats.

http://forum.audacityteam.org/viewtopic ... 454#p99454

I also see that WaveStats uses the "A" curve for RMS noise (normally pretty common). That's at variance with ACX specs which are either C or X.

'Sup?

Koz

[Robert J. H.]

How are we measuring ACX Compliance? I see through searching that we seemed to have settled on WaveStats.

http://forum.audacityteam.org/viewtopic ... 454#p99454

I also see that WaveStats uses the "A" curve for RMS noise (normally pretty common). That's at variance with ACX specs which are either C or X.

'Sup?

Koz

I always look at the normal RMS, The A-standard is fairly lower and would mean heavy compression must be applied to meet -18 to -23 dB.
Where have you read about the ACX measurement habitude?

[Paul L]

Sorry if this whole thread is tl;dr for me-- but what ACX guideline document are we referring to? I have never bothered with A weighted rms values nor seen mention of them in requirements. I aim for -20 rms overall, 3dB of headroom, backgound noise below -60 rms, and proper spacing (I still follow the old rules of .5s at the start, 2.5s after chapter title, 3.5s at the end; new requirements are less specific). I have even neglected the headroom in my earlier efforts. ACX never complained except for some mistakes in my spacings.

[Robert J. H.]

So, has any one noticed the can-like sound I've been talking about?
No? I must be paranoid then...

I've been able to analyse Ian's tests shortly.
What strikes me the most is the extreme presence of frequencies around 1 kHz and to a lesser degree the damped frequencies above.
I've radically corrected this in the following sample--quite revealing I dare to say...

Cattle - Chapter 1 (B-A Eq).mp3

(207.96 KiB) Downloaded 254 times

I've tried to do it with the EQ-effect. Unfortunately, the preview doesn't absolutely not play what it is supposed to play, i.e. the end result is not the same.

I've used the following line for the Nyquist Prompt instead:

Code: Select all

(eq-band 
   (eq-band s 1000 -15 1.8)
 2700 6 1)

Don't tell me that the voice isn't much clearer.

[Paul L]

There is no one American accent, neither is there just one British. All of them are mere regionalisms and most are very unlike what Shakespeare would have said. I love the different musics in all of them.

But if by British we mean "RP" English, I find the most distinguishing vowel I tend to forget if I am careless imitating it, is the "o" of "not" which is rounded. So, Ian, remember to un-round it to sound like a Yank. (Except certain New Englanders, the real "Yankees" ... details, details.)

http://aschmann.net/AmEng/#FatherBother