In general you will only be able to get an approximation. The best approximation to equal temperament tuning will be $$ n = [12 \log_2(B/A)] \\ H = (B/A)^{\tfrac 1 n} \\ W = H^2 \\ $$ where $\;[\dots]\;$ is rounding to the nearest integer.

Note how this always makes $\;A H^n = B\;$, by design: in essence your question was really simply, "How can I find an $\;H\;$ so that multiplying $\;A\;$ by it a couple of times, I get exactly $\;B\;$?" Also, note how if rounding were removed, the result would always be $\;H = 2^{\tfrac 1 {12}}\;$, the equal temperament semitone.

This gives you a semitone $\;H\;$ that is $\;1200 \log_2 H - 100\;$ cents off of an equal temperament semitone.

In your specific example of $\;A = 1.07\;$ and $\;B = 1.4437\;$ this gives $\;n=5\;$ semitones (so the interval that 'is' $\;B\;$ is the fourth, not the tritone) with $\;H=1.061741\dots\;$ and $\;W=1.127294\dots\;$, where H is almost 4 cents too large, and the octave is almost 45 cents too large.

I'm assuming from the way you phrase your question, that your whole tone is always exactly twice a semitone.

In general you will only be able to get an approximation, in the sense that 5 whole tones and 2 semitones (or 12 semitones) never exactly add up to an octave ($\;2\;$). The best approximation to equal temperament tuning will be $$ n = [12 \log_2(B/A)] \\ H = (B/A)^{\tfrac 1 n} \\ W = H^2 \\ $$ where $\;[\dots]\;$ is rounding to the nearest integer.

This always makes $\;A H^n = B\;$, by design: in essence your question was really simply, "How can I find an $\;H\;$ so that multiplying $\;A\;$ by it a couple of times, I get exactly $\;B\;$?"

(Also, if rounding were removed, the result would always be exactly $\;H = 2^{\tfrac 1 {12}}\;$, the equal temperament semitone.)

The above gives you a semitone $\;H\;$ that is $\;1200 \log_2 H - 100\;$ cents off of an equal temperament semitone.

In your specific example of $\;A = 1.07\;$ and $\;B = 1.4437\;$ this gives $\;n=5\;$ semitones (so the interval that 'is' $\;B\;$ is the fourth, not the tritone) with $\;H=1.061741\dots\;$ and $\;W=1.127294\dots\;$, where $\;H\;$ is almost 4 cents too large, and the octave is almost 45 cents too large.

In general you will only be able to get an approximation. The best approximation to equal temperament tuning will be $$ n = [12 \log_2(B/A)] \\ H = (B/A)^{\tfrac 1 n} \\ W = H^2 \\ $$ where $\;[\dots]\;$ is rounding to the nearest integer.

(Without giving any real background, note how if rounding were removed, the result would always be $\;H = 2^{\tfrac 1 {12}}\;$, the equal temperament semitone.)

This gives you a semitone $\;H\;$ that is $\;1200 \log_2 H - 100\;$ cents off of an equal temperament semitone.

In your specific example of $\;A = 1.07\;$ and $\;B = 1.4437\;$ this gives $\;n=5\;$ semitones (so the interval is the fourth, not the tritone) with $\;H=1.061741\dots\;$ and $\;W=1.127294\dots\;$, where H is almost 4 cents off, and the octave is almost 45 cents off.

In general you will only be able to get an approximation. The best approximation to equal temperament tuning will be $$ n = [12 \log_2(B/A)] \\ H = (B/A)^{\tfrac 1 n} \\ W = H^2 \\ $$ where $\;[\dots]\;$ is rounding to the nearest integer.

Note how this always makes $\;A H^n = B\;$, by design: in essence your question was really simply, "How can I find an $\;H\;$ so that multiplying $\;A\;$ by it a couple of times, I get exactly $\;B\;$?" Also, note how if rounding were removed, the result would always be $\;H = 2^{\tfrac 1 {12}}\;$, the equal temperament semitone.

This gives you a semitone $\;H\;$ that is $\;1200 \log_2 H - 100\;$ cents off of an equal temperament semitone.

In your specific example of $\;A = 1.07\;$ and $\;B = 1.4437\;$ this gives $\;n=5\;$ semitones (so the interval that 'is' $\;B\;$ is the fourth, not the tritone) with $\;H=1.061741\dots\;$ and $\;W=1.127294\dots\;$, where H is almost 4 cents too large, and the octave is almost 45 cents too large.