STP vs. Sound Soap?

Question

Has anyone compared the sound reduction capabilities of STP vs. Sound Soap from Bias? And in a situation where STP's Set Noise Print / Reduce Noise function doesn't help make the voice all that much clearer, would Sound Soap possibly be better equipped to keep the baby while throwing out the bath water?
I'm currently needing to transcribe a 17-minute speech heard over a ballroom PA system, and recorded as an experiemnt on a tiny digital pocket memo recorder using its built in mic (the kind of thing where you'd normally hold it close to your mouth and go "memo to self...").
Needless to say, there's a LOT of hiss, and it appears to be across the full frequency spectrum. You can understand the speaker when he puts some oomph in his delivery, but when he goes a bit soft it's often a struggle to make out the exact words he's saying. Have tried STP's noise reduction feature (in various degrees), but am finding that the more noise I remove, the more "warped" the voice becomes, and as for intelligibility (which is all I'm trying for) it's pretty much a wash (pardon the pun).
So should I shell out for Sound Soap and give it a try, or am I likely to hear that much of a difference?
John Bertram
Toronto
G5 dual 1.8   Mac OS X (10.4.3)  

Answer 1

We use both SoundSoap and STP. STP seems to work best with consistent low level 'airy" type background noise. It tends to chirp badly as you apply higher levels of effect. One suggestion we have used is to keep the level of noise reduction minimal but make several passes to target as narrow a frequency range as possible. On the type of noise you are describing we have had some limited success using STP.
We also use SoundSoap for more broadband type noise. It works remarkably well and is efficient. It also will chirp badly when you apply too much effect. It unfortunatly doesn't work too well when applied in successive passes. Like STP it has an "enhanced" slider to adjust for frequency loss and seems to work reasonably well.
For me it is nearly a toss up when it comes to the two apps performance. As the earlier writer said, if the noise is in the same frequency range as the recorded voice you will lose some of the vocal and even the best noise reduction software won't prevent that. You are likely going to have to settle for some noise in the background.
One last suggestion is to use the graphic equalizer in FCP to target the specific region of voice you want to hear. You can use this EQ to knock out the very high frequencies and most pervasive rumbles and boost the vocal frequencies that are closest to your speakers natural range. Careful not to over do it because it will cause the voice to digitally quantize and sound awful if you apply too much gain to the frequencies. Try this on a small clip and save it out as an aiff before running the noise reduction in STP on the clip. This will accentuate the noise but give you a stronger signal for the noise reduction to work on. I have used this technique to lift a female voice from a room with an air conditioner droning in the background with fairly decent results.