Page 1 of 1

Un-mini the mini-speaker

Posted: Sat Oct 18, 2008 4:44 am
by Okieman
I recorded a cell phone call with a friend by holding the mic end of a little audio recorder next to where the sound comes out of the flip up on the cell. it actually worked ok except her voice (surprise) sounds like it's coming out of a tiny little speaker. Is there some way I can oomph it up? Lower the pitch and raise the volume? Tinker with the compression? Other suggestions?

Re: Un-mini the mini-speaker

Posted: Sat Oct 18, 2008 11:54 am
by steve
There is only a certain amount of improvement that you can make - no amount of post production processing can put back what is not there, but what you can do, is to make the most of what IS there.

This is using Audacity 1.3.5

First, have a look at what you are up against:
Select a representative section of the recording, and select "Plot Spectrum" from the "Analyze" menu.
Increase the "size" parameter to a larger number, say 8192.
Change the "Axis" parameter to "Log frequency".

You will now see something that probably looks similar to this:
Screenshot.png
spectrum of telephone recording
Screenshot.png (13.33 KiB) Viewed 451 times
You will notice in this example that the bass begins to drop away sharply below about 200Hz, and the treble also drops sharply above 1700Hz.
There is also a noticeable "bulge" at around 300 Hz, and again at around 1400Hz.

The trick now, is to use the Equalizer to make some correction to these limitations. Do not be too ambitious.. In our example, there is so little sound below 100Hz, or above 6000Hz that it is pointless trying to salvage any sound in those frequency ranges, as all we will achieve is increasing the background noise. Also, for the human voice, the frequency range is predominantly in the range of 100Hz to 8kHz (8000 Hz).

We now set the Equalizer to boost the lower and upper frequencies, within the range that is practical, and reduce the frequencies that are over emphasized.

Some rough guidance on frequencies:
100Hz - too much will sound boomy and muddy
500Hz - too much will sound boxy, too little will sound thin.
1000Hz - Hearing is most sensitive around this frequency - this is around the centre of a typical telephone speaker response.
2000Hz (2kHz) - this can add clarity and make words sound more distinct, particularly bringing out "T" and "S" sounds. Too much can sound harsh.
4kHz and above - adds presence to the sound - too much and you will just hear lots of hiss.

Here is a picture of the Equalizer settings for correcting the above example:
Screenshot-1.png
Equaliser settings
Screenshot-1.png (21.07 KiB) Viewed 450 times
If use the "plot spectrum" function again, we see that the frequeny response is now much smoother, and we have managed to widen the frequency band of our recording. Most importantly, it sounds better.
Screenshot-2.png
After Equalisation
Screenshot-2.png (13.51 KiB) Viewed 451 times
An unfortunate side effect is that the amount of background noise has increased, but the noise reduction effect can help to minimise that.

Note. The "Preview" function in Audacity 1.3.5 is buggy, so if you have problems previewing the Eq settings, just apply your settings to a small section, then use "undo" if you need to change the settings. You will need to experiment a bit to get the optimum settings, but in Audacity 1.3.5, once you have the best settings, you can save the settings for future use.