Two mics into Tonedexter

[James May]

Originally Posted by Doug Young

Those devices can be interesting, I've had both the physical and the UAD plugin. There's also the Auto-Align plugin. I've experimented a bit with them and ToneDexter, and one situation was really interesting. I was training ToneDexter by recording both the pickup and mic, then playing it back to do the training. So I thought it might work well to phase-align the mic and the pickup, especially on one guitar where the wavemap was coming out a bit distant-sounding. But nope... What happened was that ToneDexter got stuck and wouldn't train at all. Asking James or Andy, they explained that ToneDexter actually uses the time delay/phase between the mics and the pickup in some way, and by eliminating it, I basically short-circuited the process. This was an early rev of tonedexter, so no idea if this still applies...

ToneDexter will only get stuck training if the mic signal comes exactly time aligned with the pickup signal, or beforehand. If it is at least a smidge behind in time (say >100us or so) then all will be well. This is because the algorithm is based on the concept that the pickup signal is "causing" what the the mic hears. It can't be causal if it happens beforehand.

Originally Posted by Doug Young

To me, and this may not be accurate - James can correct me - I don't think a focus on the same kind of "good" sound that we go for when recording necessarily translates to a better wavemap. The mic is providing data to the algorithm and the data that works best as input to the algorithm may not correspond to a good recording. Stereo micing techniques, for example, seem totally unrelated. James has also said that room acoustics don't play a big role, whereas they're critical for recording. There's always a tendency to think that more is better - if 1 mic works, maybe 2, and if that works, why not 3 or 4? My suspicion is that a clean simple signal as input to the algorithm might be preferable.

It's true that what's best for a solo guitar recording will not necessarily be best for ToneDexter training. And, ToneDexter is a monophonic device, so stereo miking techniques will be lost, and is some cases detrimental.

My experiment detailed above was designed to answer the question I get a lot: will 2 (or more) mics, mixed together, give a better WaveMap? The answer is possibly, but only better than a poorly positioned single mic. In general, one well placed mic will be better and certainly easier to do.

Regarding the continuously variable phase alignment tool mentioned. It should be noted that phase alignment is not the same thing as time alignment. Phase alignment can only perfectly time align two sources at one frequency. In practice, a range of frequencies are close enough to gain improvements in some specific band, say lows coming from a bass guitar. But if you listen carefully to two mics on an acoustic guitar that are at different distances from the instrument, the only way to correct this is with actual time alignment. Either from moving the recorded tracks, or with a digital delay line.

[Doug Young]

Originally Posted by James May

T
Regarding the continuously variable phase alignment tool mentioned. It should be noted that phase alignment is not the same thing as time alignment. Phase alignment can only perfectly time align two sources at one frequency.

Important point. In practice, these tools act more like a tone control - you basically get to choose what frequency is most aligned (i.e. least phase cancellation).