$\newcommand{\Z}{\mathbb{Z}}$We can assume $X$ path-connected, by restricting to each path-component.
Then yes, vanishing cohomology does imply vanishing homology. In Hatcher's Algebraic topology, proposition 3F.12, it is proven that $\tilde{H}^*(X) = 0$ implies that $\tilde{H}_n(X)$ is finitely generated for all $n$, because otherwise $H^*(X)$ would be uncountable.
Now the universal coefficient theorem tells you that for every $n > 0$, $$\begin{align}
\hom(H_n(X), \Z) & = 0
& \mathrm{Ext}(H_n(X), \Z) & = 0
\end{align}$$
But since $H_n(X)$ is a finitely generated abelian group, you can write it as
$$H_n(X) = \Z^r \oplus \Z/p_1^{\alpha_1} \oplus \dots \oplus \Z/p_s^{\alpha_s}$$
- The first equation gives you $\hom(H_n(X), \Z) \simeq \Z^r = 0 \Rightarrow r = 0$;
- The second equation gives you $\mathrm{Ext}(H_n(X), \Z) \simeq \Z/p_1^{\alpha_1} \oplus \dots \oplus \Z/p_s^{\alpha_s} = 0 \Rightarrow s = 0$.
Therefore $H_n(X) = 0$.