I managed to get a substantial improvement by using a time alignment process. The result is not great, but it's a lot better. The technique is to compensate for the time difference of about 2.5ms between the same voice picked up by the two mics. Since the problem is symmetrical, the same time is used for processing both mic tracks.

What you need to do is find a delay value that gives you the most acceptable sound. It's easiest to experiment on sections where you are the speaker. Using your DAW, put your mic on one track and her mic on the other track. Choose a short section of you speaking and slide one of the tracks around in time until you get the best result. Note the time slip value. Now create a process that examines the amplitude of the two input tracks over successive periods of 50ms or so to determine who is speaking at that time. Slip the other track forward by the chosen amount for the duration of time that you determine for that speaker. It doesn't make it perfect, but it's a lot better, acoustically speaking. I was doing my best to process the audio and avoid the content, but that part of it was difficult.

Thank you very much for taking the time to look into this and reply. I understand your solution. It sounds like it would be a lot of work in postproduction. Do you think there is some type of compressor/gate that could automatically shut down the audio input of the person who is not talking and automatically go back and forth so only one mic input is active at any one point? I guess I'm asking if there is a hardware solution to this problem so I don't have to spend so much time afterwards making it sound good. If there is a hardware solution, is there a particular device you would recommend if I wanted the best possible solution? Thanks again for taking the time to help me.

Well, yes, that's how I started, but the trouble comes with your colleague's voice. It may not be possible to set the gate levels to work in the intended fashion with both your voices.

I know that I described a method that involved time-shifting, but a similar decision principle could be applied to gating if you can tolerate the sound of gated voice channels. Since your voice is the one with much greater amplitude, you could have a decision process that said that if you were speaking above a set threshold (in your mic channel), the second mic was gated off, otherwise your mic channel was off and the other mic on. No time-shifting involved, but it may go unsettlingly wrong on voice overlaps.

Thank you. What do you think about scrapping the entire dual microphone thing and just going with a single high-quality omnidirectional mic sitting above our heads?

Your two technical problems are unequal voice levels and phasing due to multiple microphones picking up the same source. If you used a single microphone, you would solve the second problem, but you may have even greater troubles with unequal voice amplitudes. There are standard solutions to this type of situation using carefully-positioned narrow-angle (shotgun) microphones, but you need to be working in a good room acoustic for them to give good results. Headset mics are also a way round this, but if you are doing video, you would have to get the visually-aesthetic type used for TV.

I thought you were trying to rescue an existing recording rather than re-record it?

thank you for all of your suggestions. I record our program live and then edited afterwards to resell. This is something that I do every week. Ongoing. Appreciate all of your help.

What lavs are you using? Omnidirectional mics will pick up a lot to the side. Cardioid mics will decrease pickup to the side. Super- or hyper-cardioid mics would further decrease bleed, but you do have to aim them carefully.

Slipping one track a little for the whole production should be a quick and easy job. You just line up your track with the bleed of your voice in her track. Fortunately she isn't bleeding into your mic, otherwise there wouldn't be much you could do.

I just purchased a new mic on Amazon. The Rode NT2-A. I think it will work well. It's flexible to play with too. thanks for all of your input!

Single mic is always best. A shotgun or a multi-pattern studio VO mic works great. Two Omni lavs that close together is always problems. i have had some success with using a gate on the lower mic but best to use one mic. the Rode NT-2A is a good mic for this micing situation

Given that separate lavs for each speaker is the standard practice for thousands of different TV productions there's no real reason it can't work. If the current lavs are omnis then switch to something directional. Also, some effort on the speakers' parts to project at an appropriate volume seems like an obvious thing to try. As a last resort you might look into automixers.

I agree with Bouldersound. All "speaking" should be at normal conversational levels and let the the audio mixer make it right. I have also reversed phase on one of the mics to help with frequency cancellations and a little better isolation. Sit a little farther apart too. the camera can adjust the two-shot better than the mics. I never use omni Lavs on location any longer. Always a cardiod or super-cardiod pattern. I have had success with figure-8 patterns on occasion as well. In most instances I will wire the talent with Lavs. And hang a schoeps directional shotgun on a C-Stand splitting the two talents for a balance and choice in post.

Good luck.