Need some suggestions quickly

Attached is a short mp3 clip from a much larger spoken word wave file I just received as a restoration project.  My client wants to reduce the background buzz (from a cordless phone) and bring out the voices so there's more clarity to the conversation.  Simple noise reduction helps a little, but is there a better solution?

I'm currently using AA 3.01 and would appreciate any suggestions anyone might have. 

Good luck with this.  Personally I think there is really nothing that can be done about it.  It's broadband noise and can't be reduced without affecting the voice, either with EQ or Noise Reduction. 
Check it out in Spectral View and you will see what I mean about the broadband noise. 

The broadband noise issue is what threw me, Eric -- I wasn't sure how to deal with it -- if at all. 

Before you do the NR, you can reduce a lot of that noise by using the notch filter set at about 230, 460 and 920Hz, with a lot of attenuation (rather more than the default values), and this will reduce the buzz somewhat, because that has a definite pitch to it. If you apply the NR after this, you can hear a lot more of the conversation. You will have to experiment with this, but it's definitely better than NR alone.

*edit* I may be able to come up with a better set of numbers, but you may have to process this in stages, I think.

Thanks, Steve, for the suggestion.  If you have anything else, I'll take it.  

From the numbers you've given, it looks like you might be suggesting some 1st and 2nd overtone/harmonics settings. 

The thing to do, it turns out, is to set a reasonable value in the frequency analysis window (I used 8192) and notice that essentially, a lot of those spikes don't move much. If you look carefully at the frequencies, you'll get a list with 199, 299, 399 etc up to 899, and another set also spaced at 100Hz starting from 1.2kHz. If you set the notch filter to very or super narrow, with all these numbers in it, and the attenuation set so that it follows for all of them, you can wind the attenuation value up and down until you get the best-sounding result. You'll have to do this more than once because there aren't enough slots in the notch filter to do them all in one go, but with care you can reduce these quite a bit.

At that point, take a NR sample, but make sure that you've set the FFT to the highest value you can before taking it. Now, you should be able to shape the amount you use using the blue line so that on parts where there's no voice (below 200Hz and above about 3kHz) there's rather more NR, and a sensible amount where the speech is. If you do this carefully, you can hear quite a bit of it rather better than you could before. This is by no means perfect, but that really is quite a crap sample to work with!

tcatzere, do you have an old copy of Audition lying around?  AA1.5 and all all previous versions permitted an insane Q-value setting in the Parametric EQ filter.  Steve's right on with the notch filter idea, of course, but if you can use the old Parametric version, you may be able to go even narrower than the notch filter permits, and perhaps disturb what little is left of the conversation even less.