Making a file "low-fi"

Special Interest Groups Audio Processing
1

Happy springtime Wednesday, fellas. I hope everyone’s doing awesomely. I’ve been searching online for a week for a plug-in that I thought would be easy to find. This is how my stories always start out, isn’t it!

I make a lot of videos that involve voice-overs. They’re about very old video games and 8-bit programming tricks, and are for my reference only, so I’m not concerned with how good they sound to anyone else.

After the repeated listening and editing, my own voice is really starting to get on my nerves. I sound different recorded digitally than I ever did in analogue. So, first, I use Limiter 2 to force down the volume spikes in my natural “enthusiastic” speaking voice. Then I use Spitfish and/or Lisp to get rid of the sharp sibilants, which makes the sound so bassy that I have to atone by lowering the bottom EQ faders. Then the sharpness is, of course, back. So it’s a balancing act, and I get the feeling that it’s all unnecessary if I just want to lower the quality so much that high/low frequencies don’t matter.

I listen to a podcast that usually squashes six+ hours of spoken material into a relatively small MP3 file. Their voices are low-quality but understandable – no “ringing” artifacts, but certainly no dynamic range. It sounds as if they’re on a telephone with the treble turned all the way down. That would be perfect.

I’m thinking, “If I could just make my voice sound that low-fi and dynamics-free, I’d be able to speak without being paranoid about every accidental spittle and volume peak…”

I tried the “Telephone” VST plug-in, but all it does is crank up the mids and highs, and add a bunch of deliberate buzzing. Even the “Telephone” pre-set in the Audacity EQ sharpens everything. I’ve tried saving at a lower bit rate, but it speeds up my voice and I sound like a chipmunk. It’s amusing, but not what I’m going for.

I’ve tried searching “Skype quality,” “low podcast quality,” etc., etc., and I’ve found nothing.

Does anyone know of a good way to squash the hell out of my voice and make it sound like it’s coming out of a cheap speaker, devoid of extreme highs and lows? I don’t even mind a bit of distortion. I considered buying one of those suction-cup mics and speaking into my old phone receiver while recording, but it has to be plugged into a phone to render sound through the earpiece, and I can’t think of any number to call that would give me empty air!

Thank you.

2

Something like telephone-equalization is the way to go , You just have to experiment with the upper and lower frequency thresholds …

Spectrogram of ''Dubya, low-pass, progressively lower theshold''.png
Or alternatively lower the sample-rate , ( to say 22050Hz ), which will sharply cut off the frequencies at half the sample rate , (11kHz). Using “resample” in the “tracks” menu to do that doesn’t cause the chipmunk effect.

3

Cool; I’ll try resampling. As much as I’ve been using Audacity, I’m surprised that I didn’t notice that selection.
Thanks, as always, Trebor!

4

If you want “skype quality” you may need to resample at 16000Hz , maybe even lower.

5

Still searching for a plug-in – or something – that squashes the highs and lows. If anyone happens to know of such an effect…well, I suppose that’s why I threw this out there. One never knows.

My USB mic is apparently just “too good” to be affected much by resampling. I even tried cutting off the highest and lowest six frequencies in the EQ, and then resampling at 8K. It sounded even tinnier, and didn’t degrade much in quality at all. I’ve run things through the Spitfish De-Esser three times, tried an effect called Degrade (which hikes up the bass ridiculously), etc. Nothing sounds like that old-phone-quality, no-treble voice that I hear in long, compressed podcasts.

But I appreciate your taking the time to offer the advice. Maybe I just need to buy a very bad mic, or something outboard that can accept one of my old, warm mics with 1/8" plugs that I’ve had since the '80s. Thanks again!

6

If you can post a few seconds of what you sound like now on your “too good” mic , and what you used to sound like on the podcast , we can probably suggest an equalization curve to make the new sound more like the old.

7

My USB mic is apparently just “too good”

Make? Model Number?

If you have to go through all that just to make a recording that sounded like your older analog recordings, there is something seriously wrong with your system. Crashing sibilants in particular are always a warning sign of a mismatch or other damage.

You can get a startlingly bad recording by using the microphone wrong.

How are you listening? Can you hold your speakers in one hand? Headphones?

Koz

8

Try this:

  1. Amplify effect (default settings)

  2. Compressor effect:
    Threshold -30
    Noise floor: -30
    Ratio: 10:1
    Attack time: 0.1
    Release time: 1.0
    Make-up gain: No (not selected)
    Based on peaks: Yes (selected)

  3. Equalization effect: Settings something like this (tweak to taste - use the Preview button before deciding on the exact settings:

    eq.png
    eq.png703×456 33 KB

9

Good point; I guess it would have been helpful including the actual mic model! And that’s what the strange thing is – it’s not exactly a high-end mic. It’s a Bad Axx 904-UM-600, which is a USB condenser mic (tight cardioid, I’m sure). If I could afford an outboard compressor, a Shure or an old U47 or something, you can bet that I wouldn’t hesitate.

By “too good,” I meant that there are extreme highs and lows in there. For a sixty-odd-dollar mic, that thing has a hell of a range. I’m at work and don’t have any raw samples at hand, but I’ll post one next week, at the risk of bugging everyone. Maybe I’ll just knock out a couple of front teeth to change my sharp pronunciations of things. (Kidding. Mostly.) I sound like a gay Linus when I hear myself back. Nothing against gays or Peanuts, mind you.

Thanks for the EQ suggestion, Steve. I’ll try that at home tonight. I really appreciate you guys reading my weird query. Hell, if text-to-speech programs that sounded halfway comprehensible didn’t cost an arm and a leg, I’d probably just go with one of those.

10

There are free online text-to-speech converters , some realistic, some a bit “Stephen Hawking” …

:slight_smile:
Microsoft Windows does come with text-to-speech , but IIRC legally it’s speech output cannot be broadcast ,
( i.e. personal use only ), a Podcast is a broadcast. ( Probably true of Adobe Reader’s text-to-speech too).

11

That’s like saying “don’t use that blue ink for writing your correspondence, it’s for personal use only”.
I do use my TTS voices regularly, they were certainly expensive enough. The voices that come along with MS aren’t not worth a consideration anyways.
The voices of the MS speech platform are much better, but they may have the same restrictions.

@ABQChris
You could include a sample that sounds “pleasant” to your ears, we can maybe construct a filter that approaches that ideal.

12

Thanks for the link, Trebor! (Funny wave file!)